User login
Web Page Content and Quality Assessed for Shoulder Replacement
The Internet is becoming a primary source for obtaining medical information. This growing trend may have serious implications for the medical field. As patients increasingly regard the Internet as an essential tool for obtaining health-related information, questions have been raised regarding the quality of medical information available on the Internet.1 Studies have shown that health-related sites often present inaccurate, inconsistent, and outdated information that may have a negative impact on health care decisions made by patients.2
According to the US Census Bureau, 71.7% of American households report having access to the Internet.3 Of those who have access to Internet, approximately 72% have sought health information online over the last year.4 Among people older than age 65 years living in the United States, there has been a growing trend toward using the Internet, from 14% in 2000 to almost 60% in 2013, according to the Pew Research Internet Project.5 Most medical websites are viewed for information on diseases and treatment options.6 Since most patients want to be informed about treatment options, as well as risks and benefits for each treatment, access to credible information is essential for proper decision-making.7
To assess the quality of information on the Internet, we used DISCERN, a standardized questionnaire to aid consumers in judging Internet content.8 The DISCERN instrument, available at www.discern.org.uk, was designed by an expert group in the United Kingdom. First, an expert panel developed and tested the instrument, and then health care providers and self-help group members tested it further.8,9 The questionnaire had been found to have good interrater reliability, regardless of use by health professionals or consumers.8-10
More than 53,000 shoulder arthroplasties are performed in the United States annually, and the number is growing, with the main goal of pain relief from glenohumeral degenerative joint disease.11,12 The Internet has become a quasi–second opinion for patients trying to participate in their care. Given the prevalence of shoulder-related surgeries, it is critical to analyze and become familiar with the quality of information that patients read online in order to direct them to nonbiased, all-inclusive websites. In this study, we provide a summary assessment and comparison of the quality of online information pertaining to shoulder replacement, using medical (total shoulder replacement) and nontechnical (shoulder replacement) search terms.
Methods
Websites were identified using 3 search engines (Google, Yahoo, and Bing) and 2 search terms, shoulder replacement (SR) and total shoulder arthroplasty (TSA), on January 17, 2014. These 3 search engines were used because 77% of health care–related information online searches begin through a search engine (Google, Bing, Yahoo); only 13% begin at a health care–specialized website.4 These search terms were used after consulting with orthopedic residents and attending physicians in a focus group regarding the terminology used with patients. The first 30 websites in each search engine were identified consecutively and evaluated for category and quality of information using the DISCERN instrument.
A total of 180 websites (90 per search term) were reviewed. Each website was evaluated independently by 3 medical students. In the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram, we recorded how websites were identified, screened, and included (Figure 1).13 Websites that were duplicated within each search term and those that were inaccessible were used to determine the total number of noncommercial versus commercial websites, but were excluded from the final analysis. The first part of the analysis involved determining the type of website (eg, commercial vs noncommercial) based upon the html endings. All .com endings were classified as commercial websites; noncommercial included .gov, .org, .edu, and .net endings. Next, each website was categorized based on the target audience. Websites were grouped into health professional–oriented information, patient-oriented, advertisement, or “other.” These classifications were based on those described in previous works.14,15 The “other” category included images, YouTube videos, another search engine, and open forums, which were also excluded from the final analysis because they were not easily evaluable with the DISCERN instrument. Websites were considered health professional–oriented if they included journal articles, scholarly articles, and/or rehabilitation protocols. Patient-directed websites clearly stated the information was directed to patients or provided a general overview. Advertisement included sites that displayed ads or products for sale. Websites were evaluated for quality using the DISCERN instrument (Figure 2).
DISCERN has 3 subdivision scores: the reliable score (composed of the first 8 questions), the treatment options (the next 7 questions), and 1 final question that addresses the overall quality of the website and is rated independently of the first 15 questions. DISCERN uses 2 scales, a binary scale anchored on both extremes with the number 1 equaling complete absence of the criteria being measured, and the number 5 at the upper extreme, representing completeness of the quality being assessed. In between 1 and 5 is a partial ordinal scale measuring from 2 to 4, which indicates the information is present to some extent but not complete. The ordinal scale allows ranking of the criteria being assessed. Summarizing values from each of the 2 scales poses some concern: the scale is not a true binary scale because of the ordinal scale of the middle numbers (2-4), and as such, is not amenable to being an interval scale to calculate arithmetic means. To summarize the values from the 2 scales, we calculated the harmonic mean, the arithmetic mean, the geometric mean, and the median. The means were empirically compared with the median, and we used the harmonic mean to summarize scale values because it was the best approximation of the medians.
Results
A total of 90 websites were assessed with the search term total shoulder arthroplasty and another 90 with shoulder replacement. When 37 duplicate websites for TSA and 52 for SR were eliminated, 53 (59%) and 38 (42%) unique websites were evaluated for each search term, respectively (Figure 1). (These unique websites are included in the Appendix.) Between the 2 search terms, 20 websites were duplicated. Figure 3 shows the distribution of websites by category. Total shoulder arthroplasty provided the highest percentage of health professional–oriented information; SR had the greatest percentage of patient-oriented information. Both TSA and SR had nearly the same number of advertisements and websites labeled “other.” The percentage of noncommercial websites from each search engine is represented in Figure 4. For SR, Google had 40% (12/30) noncommercial websites compared with Yahoo at 53% (16/30) and Bing at 46% (14/30). Total shoulder arthroplasty had 43% (13/30) noncommercial websites on Google, 27% (8/30) on Yahoo, and 40% (12/30) on Bing. In total, SR had more noncommercial websites, 47% (42/90), compared with 37% (33/90) for TSA.
The mean of all 3 raters for reliablity (DISCERN questions 1-8) and treatment options (DISCERN questions 9-15) is represented in the Table. For both search terms, we found that websites identified as health professional–oriented had the highest reliable mean scores, followed by patient-oriented, and advertisement at the lowest (SR: P = .054; TSA: P = .134). For SR, treatment mean scores demonstrated similar results with health professional–oriented websites receiving the highest, followed by patient-oriented and advertisement (P = .005). However, the treatment mean scores for TSA differed with patient-oriented websites receiving higher scores than health professional–oriented websites, but this was not statistically significant (P= .407). Regarding search terms, there were no significant differences between mean reliable and treatment scores across all categories.
The average overall DISCERN score for TSA websites was 2.5 (range, 1-5), compared with 2.3 (range, 1-5) for SR websites. The overall reliable score (DISCERN questions 1-8) for TSA websites was 2.6 and 2.5 for SR websites (P < .001). For TSA websites, 38% (20/53) were classified as good, having an overall DISCERN score ≥3, versus 26% (10/38) of SR websites. The overall DISCERN score for health professional–oriented websites was 2.7, patient-oriented websites received a score of 2.6, and advertisements had the lowest score at 2.4.
Discussion
Both patients and health professionals obtain information on health care subjects through the Internet, which has become the primary resource for patients.15,16 However, there are no strict regulations of the content being written. This creates a challenge for the typical user to find credible and evidence-based information, which is important because misleading information could cause undue anxiety, among other effects.17,18 The aims of this study were to determine the quality of Internet information for shoulder replacement surgeries using the medical terminology total shoulder arthroplasty (TSA) and the nontechnical term shoulder replacement (SR), and to compare the results.
After analyzing the types of websites returned for both total shoulder arthroplasty and shoulder replacement (Figure 4), it was interesting to find that using nonmedical terminology as the search term provided more noncommercial websites compared with total shoulder arthroplasty. Furthermore, Yahoo provided the highest yield of noncommercial websites at 16, with Bing at 14, when using SR as the search term. We believe the increase in noncommercial websites returned for SR was greater than for TSA because SR yielded more patient-oriented websites, which usually had html endings of .edu and .org, as shown in Figure 3 (48% of SR websites offered patient-oriented information).
Although there were more noncommercial websites for SR, the majority of the DISCERN values between the 2 search terms did not differ significantly. This is a direct result of the number of sites (20) that were duplicated across both search terms. However as seen in the Table, TSA had similar reliable mean scores for advertisements and patient-oriented websites but a slightly higher reliable score for health professional–oriented websites. We correlated this with the increased number of health professional–oriented websites returned when using TSA as the search term (Figure 3). The health professional–oriented websites explained their aims and cited their sources more consistently than did patient-oriented sites and advertisements, resulting in higher reliable scores. Although patient-oriented websites frequently lacked citations, they provided information about multiple treatment options, which were more relevant to consumers. This resulted in nearly equivalent reliable scores. Treatment means for advertisements in both SR and TSA were similar. However, treatment means for professional-oriented websites in TSA were lower than those for SR because health professional–oriented websites often were only moderately relevant to consumers, with their focus usually on 1 treatment option or on rehabilitation protocols. Although the DISCERN scores were similar between the search terms, total shoulder arthroplasty provided more websites (20) classified as good—overall DISCERN score, ≥3—than SR did (10). Advertisement websites had similar overall DISCERN scores, which we anticipated because most of the advertisements were duplicated across the search terms.
Using the 2 search terms, academic websites and commercial websites, such as WebMD, consistently received higher reliable and overall DISCERN scores. Advertisement websites, which need to deliver a clear message, frequently scored high on explicitly stating their aims and relevance to consumers, but focused on their products without discussing the benefits of other treatment options. This is significant because Internet search engines, such as Google, offer sponsor links for which organizations pay to appear at the top of the search results. This creates the potential for consumers to receive biased information because most individuals only visit the top 10 websites generated by a search engine.19
We concluded that the quality of online information relating to SR and TSA was highly variable and frequently of moderate-to-poor quality, with most overall DISCERN scores <3. The quality of information found online for this study using the DISCERN instrument is consistent with those studies using DISCERN to evaluate other medical conditions (eg, bunions, chronic pain, general anesthesia, and anterior cruciate ligament reconstruction).2,9,15,19 These studies also concluded that online information varies tremendously in quality and completeness.
This study has several limitations. Websites were searched at a single time point and, because Internet resources are frequently updated, the results of this study could vary. Furthermore, although Google, Yahoo, and Bing are 3 of the most popular search engines, these are not the only resources patients use when searching the Internet for health-related information. Other search engines, such as Pubmed.gov and MSN.com, could provide additional websites for Internet users. Lastly, although DISCERN is validated to address the quality of information available online, it does not evaluate the accuracy of the information.8 Our use of DISCERN involves 2 scales, a binary yes/no (ratings, 1 and 5) and an ordinal scale (ratings, 2-4). As such, a single mean summary statistic cannot be calculated.
Conclusion
The information available on the Internet pertaining to TSA and SR is highly variable and provides mostly moderate-to-poor quality information based on the DISCERN instrument. Many websites failed to describe the benefits and the risks of different treatment options, including nonoperative management. Health care professionals should be aware that patients often refer to the Internet as a primary resource for obtaining medical information. It is important to direct patients to websites that provide accurate information, because patients who educate themselves about their conditions and actively participate in decision-making may have improved health outcomes.20-22 Overall, academic websites and commercial websites, such as WebMD and OrthoInfo, generally had higher DISCERN scores when using either search term. Of major concern is the potential for misleading advertisements or incorrect information that can negatively affect health outcomes. This study found that using nonmedical terminology (SR) provided more noncommercial and patient-oriented websites, especially through Yahoo. This study highlights the need for more comprehensive online information pertaining to shoulder replacement that can better serve as a resource for Internet users.
1. Eysenbach G, Powell J, Kuss O, Sa ER. Empirical studies assessing the quality of health information for consumers on the world wide web: a systematic review. JAMA. 2002;287(20):2691-2700.
2. Bruce-Brand RA, Baker JF, Byrne DP, Hogan NA, McCarthy T. Assessment of the quality and content of information on anterior cruciate ligament reconstruction on the internet. Arthroscopy. 2013;29(6):1095-1100.
3. Computer and internet use in the United States: population characteristics. US Census Bureau website. http://www.census.gov/hhes/computer/. Accessed December 11, 2015.
4. Fox S, Duggan M. Health online 2013. Pew Research Center website. http://pewinternet.org/Reports/2013/Health-online.aspx. Published January 15, 2013. Accessed November 24, 2015.
5. Smith A. Older adults and technology use. Pew Research Center website. http://www.pewinternet.org/2014/04/03/older-adults-and-technology-use. Published April 3, 2014. Accessed November 24, 2015.
6. Shuyler KS, Knight KM. What are patients seeking when they turn to the internet? Qualitative content analysis of questions asked by visitors to an orthopaedics web site. J Med Internet Res. 2003;5(4):e24.
7. Meredith P, Emberton M, Wood C, Smith J. Comparison of patients’ needs for information on prostate surgery with printed materials provided by surgeons. Qual Health Care. 1995;4(1):18-23.
8. Charnock D, Shepperd S, Needham G, Gann R. DISCERN: An instrument for judging the quality of written consumer health information on treatment choices. J Epidemiol Community Health. 1999;53(2):105-111.
9. Kaicker J, Debono VB, Dang W, Buckley N, Thabane L. Assessment of the quality and variability of health information on chronic pain websites using the DISCERN instrument. BMC Med. 2010;8(1):59.
10. Griffiths KM, Christensen H. Website quality indicators for consumers. J Med Internet Res. 2005;7(5):e55.
11. Wiater JM. Shoulder joint replacement. American Academy of Orthopedic Surgeons website. http://orthoinfo.aaos.org/topic.cfm?topic=A00094. Updated December 2011. Accessed November 24, 2015.
12. Kim SH, Wise BL, Zhang Y, Szabo RM. Increasing incidence of shoulder arthroplasty in the united states. J Bone Joint Surg Am. 2011;93(24):2249-2254.
13. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. Ann Intern Med. 2009;151(4):W65-W94.
14. Nason GJ, Baker JF, Byrne DP, Noel J, Moore D, Kiely PJ. Scoliosis-specific information on the internet: has the “information highway” led to better information provision? Spine. 2012;37(21):E1364-E1369.
15. Starman JS, Gettys FK, Capo JA, Fleischli JE, Norton HJ, Karunakar MA. Quality and content of internet-based information for ten common orthopaedic sports medicine diagnoses. J Bone Joint Surg Am. 2010;92(7):1612-1618.
16. Bernstein J, Ahn J, Veillette C. The future of orthopaedic information management. J Bone Joint Surg Am. 2012;94(13):e95.
17. Berland GK, Elliott MN, Morales LS, et al. Health information on the Internet: accessibility, quality, and readability in English and Spanish. JAMA. 2001;285(20):2612-2621.
18. Fallowfield LJ, Hall A, Maguire GP, Baum M. Psychological outcomes of different treatment policies in women with early breast cancer outside a clinical trial. BMJ. 1990;301(6752):575-580.
19. Chong YM, Fraval A, Chandrananth J, Plunkett V, Tran P. Assessment of the quality of web-based information on bunions. Foot Ankle Int. 2013;34(8):1134-1139.
20. Brody DS, Miller SM, Lerman CE, Smith DG, Caputo GC. Patient perception of involvement in medical care. J Gen Intern Med. 1989;4(6):506-511.
21. Greenfield S, Kaplan S, Ware JE Jr. Expanding patient involvement in care. Effects on patient outcomes. Ann Intern Med. 1985;102(4):520-528.
22. Kaplan SH, Greenfield S, Ware JE Jr. Assessing the effects of physician-patient interactions on the outcomes of chronic disease. Med Care. 1989;27(3 suppl):S110-S127.
The Internet is becoming a primary source for obtaining medical information. This growing trend may have serious implications for the medical field. As patients increasingly regard the Internet as an essential tool for obtaining health-related information, questions have been raised regarding the quality of medical information available on the Internet.1 Studies have shown that health-related sites often present inaccurate, inconsistent, and outdated information that may have a negative impact on health care decisions made by patients.2
According to the US Census Bureau, 71.7% of American households report having access to the Internet.3 Of those who have access to Internet, approximately 72% have sought health information online over the last year.4 Among people older than age 65 years living in the United States, there has been a growing trend toward using the Internet, from 14% in 2000 to almost 60% in 2013, according to the Pew Research Internet Project.5 Most medical websites are viewed for information on diseases and treatment options.6 Since most patients want to be informed about treatment options, as well as risks and benefits for each treatment, access to credible information is essential for proper decision-making.7
To assess the quality of information on the Internet, we used DISCERN, a standardized questionnaire to aid consumers in judging Internet content.8 The DISCERN instrument, available at www.discern.org.uk, was designed by an expert group in the United Kingdom. First, an expert panel developed and tested the instrument, and then health care providers and self-help group members tested it further.8,9 The questionnaire had been found to have good interrater reliability, regardless of use by health professionals or consumers.8-10
More than 53,000 shoulder arthroplasties are performed in the United States annually, and the number is growing, with the main goal of pain relief from glenohumeral degenerative joint disease.11,12 The Internet has become a quasi–second opinion for patients trying to participate in their care. Given the prevalence of shoulder-related surgeries, it is critical to analyze and become familiar with the quality of information that patients read online in order to direct them to nonbiased, all-inclusive websites. In this study, we provide a summary assessment and comparison of the quality of online information pertaining to shoulder replacement, using medical (total shoulder replacement) and nontechnical (shoulder replacement) search terms.
Methods
Websites were identified using 3 search engines (Google, Yahoo, and Bing) and 2 search terms, shoulder replacement (SR) and total shoulder arthroplasty (TSA), on January 17, 2014. These 3 search engines were used because 77% of health care–related information online searches begin through a search engine (Google, Bing, Yahoo); only 13% begin at a health care–specialized website.4 These search terms were used after consulting with orthopedic residents and attending physicians in a focus group regarding the terminology used with patients. The first 30 websites in each search engine were identified consecutively and evaluated for category and quality of information using the DISCERN instrument.
A total of 180 websites (90 per search term) were reviewed. Each website was evaluated independently by 3 medical students. In the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram, we recorded how websites were identified, screened, and included (Figure 1).13 Websites that were duplicated within each search term and those that were inaccessible were used to determine the total number of noncommercial versus commercial websites, but were excluded from the final analysis. The first part of the analysis involved determining the type of website (eg, commercial vs noncommercial) based upon the html endings. All .com endings were classified as commercial websites; noncommercial included .gov, .org, .edu, and .net endings. Next, each website was categorized based on the target audience. Websites were grouped into health professional–oriented information, patient-oriented, advertisement, or “other.” These classifications were based on those described in previous works.14,15 The “other” category included images, YouTube videos, another search engine, and open forums, which were also excluded from the final analysis because they were not easily evaluable with the DISCERN instrument. Websites were considered health professional–oriented if they included journal articles, scholarly articles, and/or rehabilitation protocols. Patient-directed websites clearly stated the information was directed to patients or provided a general overview. Advertisement included sites that displayed ads or products for sale. Websites were evaluated for quality using the DISCERN instrument (Figure 2).
DISCERN has 3 subdivision scores: the reliable score (composed of the first 8 questions), the treatment options (the next 7 questions), and 1 final question that addresses the overall quality of the website and is rated independently of the first 15 questions. DISCERN uses 2 scales, a binary scale anchored on both extremes with the number 1 equaling complete absence of the criteria being measured, and the number 5 at the upper extreme, representing completeness of the quality being assessed. In between 1 and 5 is a partial ordinal scale measuring from 2 to 4, which indicates the information is present to some extent but not complete. The ordinal scale allows ranking of the criteria being assessed. Summarizing values from each of the 2 scales poses some concern: the scale is not a true binary scale because of the ordinal scale of the middle numbers (2-4), and as such, is not amenable to being an interval scale to calculate arithmetic means. To summarize the values from the 2 scales, we calculated the harmonic mean, the arithmetic mean, the geometric mean, and the median. The means were empirically compared with the median, and we used the harmonic mean to summarize scale values because it was the best approximation of the medians.
Results
A total of 90 websites were assessed with the search term total shoulder arthroplasty and another 90 with shoulder replacement. When 37 duplicate websites for TSA and 52 for SR were eliminated, 53 (59%) and 38 (42%) unique websites were evaluated for each search term, respectively (Figure 1). (These unique websites are included in the Appendix.) Between the 2 search terms, 20 websites were duplicated. Figure 3 shows the distribution of websites by category. Total shoulder arthroplasty provided the highest percentage of health professional–oriented information; SR had the greatest percentage of patient-oriented information. Both TSA and SR had nearly the same number of advertisements and websites labeled “other.” The percentage of noncommercial websites from each search engine is represented in Figure 4. For SR, Google had 40% (12/30) noncommercial websites compared with Yahoo at 53% (16/30) and Bing at 46% (14/30). Total shoulder arthroplasty had 43% (13/30) noncommercial websites on Google, 27% (8/30) on Yahoo, and 40% (12/30) on Bing. In total, SR had more noncommercial websites, 47% (42/90), compared with 37% (33/90) for TSA.
The mean of all 3 raters for reliablity (DISCERN questions 1-8) and treatment options (DISCERN questions 9-15) is represented in the Table. For both search terms, we found that websites identified as health professional–oriented had the highest reliable mean scores, followed by patient-oriented, and advertisement at the lowest (SR: P = .054; TSA: P = .134). For SR, treatment mean scores demonstrated similar results with health professional–oriented websites receiving the highest, followed by patient-oriented and advertisement (P = .005). However, the treatment mean scores for TSA differed with patient-oriented websites receiving higher scores than health professional–oriented websites, but this was not statistically significant (P= .407). Regarding search terms, there were no significant differences between mean reliable and treatment scores across all categories.
The average overall DISCERN score for TSA websites was 2.5 (range, 1-5), compared with 2.3 (range, 1-5) for SR websites. The overall reliable score (DISCERN questions 1-8) for TSA websites was 2.6 and 2.5 for SR websites (P < .001). For TSA websites, 38% (20/53) were classified as good, having an overall DISCERN score ≥3, versus 26% (10/38) of SR websites. The overall DISCERN score for health professional–oriented websites was 2.7, patient-oriented websites received a score of 2.6, and advertisements had the lowest score at 2.4.
Discussion
Both patients and health professionals obtain information on health care subjects through the Internet, which has become the primary resource for patients.15,16 However, there are no strict regulations of the content being written. This creates a challenge for the typical user to find credible and evidence-based information, which is important because misleading information could cause undue anxiety, among other effects.17,18 The aims of this study were to determine the quality of Internet information for shoulder replacement surgeries using the medical terminology total shoulder arthroplasty (TSA) and the nontechnical term shoulder replacement (SR), and to compare the results.
After analyzing the types of websites returned for both total shoulder arthroplasty and shoulder replacement (Figure 4), it was interesting to find that using nonmedical terminology as the search term provided more noncommercial websites compared with total shoulder arthroplasty. Furthermore, Yahoo provided the highest yield of noncommercial websites at 16, with Bing at 14, when using SR as the search term. We believe the increase in noncommercial websites returned for SR was greater than for TSA because SR yielded more patient-oriented websites, which usually had html endings of .edu and .org, as shown in Figure 3 (48% of SR websites offered patient-oriented information).
Although there were more noncommercial websites for SR, the majority of the DISCERN values between the 2 search terms did not differ significantly. This is a direct result of the number of sites (20) that were duplicated across both search terms. However as seen in the Table, TSA had similar reliable mean scores for advertisements and patient-oriented websites but a slightly higher reliable score for health professional–oriented websites. We correlated this with the increased number of health professional–oriented websites returned when using TSA as the search term (Figure 3). The health professional–oriented websites explained their aims and cited their sources more consistently than did patient-oriented sites and advertisements, resulting in higher reliable scores. Although patient-oriented websites frequently lacked citations, they provided information about multiple treatment options, which were more relevant to consumers. This resulted in nearly equivalent reliable scores. Treatment means for advertisements in both SR and TSA were similar. However, treatment means for professional-oriented websites in TSA were lower than those for SR because health professional–oriented websites often were only moderately relevant to consumers, with their focus usually on 1 treatment option or on rehabilitation protocols. Although the DISCERN scores were similar between the search terms, total shoulder arthroplasty provided more websites (20) classified as good—overall DISCERN score, ≥3—than SR did (10). Advertisement websites had similar overall DISCERN scores, which we anticipated because most of the advertisements were duplicated across the search terms.
Using the 2 search terms, academic websites and commercial websites, such as WebMD, consistently received higher reliable and overall DISCERN scores. Advertisement websites, which need to deliver a clear message, frequently scored high on explicitly stating their aims and relevance to consumers, but focused on their products without discussing the benefits of other treatment options. This is significant because Internet search engines, such as Google, offer sponsor links for which organizations pay to appear at the top of the search results. This creates the potential for consumers to receive biased information because most individuals only visit the top 10 websites generated by a search engine.19
We concluded that the quality of online information relating to SR and TSA was highly variable and frequently of moderate-to-poor quality, with most overall DISCERN scores <3. The quality of information found online for this study using the DISCERN instrument is consistent with those studies using DISCERN to evaluate other medical conditions (eg, bunions, chronic pain, general anesthesia, and anterior cruciate ligament reconstruction).2,9,15,19 These studies also concluded that online information varies tremendously in quality and completeness.
This study has several limitations. Websites were searched at a single time point and, because Internet resources are frequently updated, the results of this study could vary. Furthermore, although Google, Yahoo, and Bing are 3 of the most popular search engines, these are not the only resources patients use when searching the Internet for health-related information. Other search engines, such as Pubmed.gov and MSN.com, could provide additional websites for Internet users. Lastly, although DISCERN is validated to address the quality of information available online, it does not evaluate the accuracy of the information.8 Our use of DISCERN involves 2 scales, a binary yes/no (ratings, 1 and 5) and an ordinal scale (ratings, 2-4). As such, a single mean summary statistic cannot be calculated.
Conclusion
The information available on the Internet pertaining to TSA and SR is highly variable and provides mostly moderate-to-poor quality information based on the DISCERN instrument. Many websites failed to describe the benefits and the risks of different treatment options, including nonoperative management. Health care professionals should be aware that patients often refer to the Internet as a primary resource for obtaining medical information. It is important to direct patients to websites that provide accurate information, because patients who educate themselves about their conditions and actively participate in decision-making may have improved health outcomes.20-22 Overall, academic websites and commercial websites, such as WebMD and OrthoInfo, generally had higher DISCERN scores when using either search term. Of major concern is the potential for misleading advertisements or incorrect information that can negatively affect health outcomes. This study found that using nonmedical terminology (SR) provided more noncommercial and patient-oriented websites, especially through Yahoo. This study highlights the need for more comprehensive online information pertaining to shoulder replacement that can better serve as a resource for Internet users.
The Internet is becoming a primary source for obtaining medical information. This growing trend may have serious implications for the medical field. As patients increasingly regard the Internet as an essential tool for obtaining health-related information, questions have been raised regarding the quality of medical information available on the Internet.1 Studies have shown that health-related sites often present inaccurate, inconsistent, and outdated information that may have a negative impact on health care decisions made by patients.2
According to the US Census Bureau, 71.7% of American households report having access to the Internet.3 Of those who have access to Internet, approximately 72% have sought health information online over the last year.4 Among people older than age 65 years living in the United States, there has been a growing trend toward using the Internet, from 14% in 2000 to almost 60% in 2013, according to the Pew Research Internet Project.5 Most medical websites are viewed for information on diseases and treatment options.6 Since most patients want to be informed about treatment options, as well as risks and benefits for each treatment, access to credible information is essential for proper decision-making.7
To assess the quality of information on the Internet, we used DISCERN, a standardized questionnaire to aid consumers in judging Internet content.8 The DISCERN instrument, available at www.discern.org.uk, was designed by an expert group in the United Kingdom. First, an expert panel developed and tested the instrument, and then health care providers and self-help group members tested it further.8,9 The questionnaire had been found to have good interrater reliability, regardless of use by health professionals or consumers.8-10
More than 53,000 shoulder arthroplasties are performed in the United States annually, and the number is growing, with the main goal of pain relief from glenohumeral degenerative joint disease.11,12 The Internet has become a quasi–second opinion for patients trying to participate in their care. Given the prevalence of shoulder-related surgeries, it is critical to analyze and become familiar with the quality of information that patients read online in order to direct them to nonbiased, all-inclusive websites. In this study, we provide a summary assessment and comparison of the quality of online information pertaining to shoulder replacement, using medical (total shoulder replacement) and nontechnical (shoulder replacement) search terms.
Methods
Websites were identified using 3 search engines (Google, Yahoo, and Bing) and 2 search terms, shoulder replacement (SR) and total shoulder arthroplasty (TSA), on January 17, 2014. These 3 search engines were used because 77% of health care–related information online searches begin through a search engine (Google, Bing, Yahoo); only 13% begin at a health care–specialized website.4 These search terms were used after consulting with orthopedic residents and attending physicians in a focus group regarding the terminology used with patients. The first 30 websites in each search engine were identified consecutively and evaluated for category and quality of information using the DISCERN instrument.
A total of 180 websites (90 per search term) were reviewed. Each website was evaluated independently by 3 medical students. In the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram, we recorded how websites were identified, screened, and included (Figure 1).13 Websites that were duplicated within each search term and those that were inaccessible were used to determine the total number of noncommercial versus commercial websites, but were excluded from the final analysis. The first part of the analysis involved determining the type of website (eg, commercial vs noncommercial) based upon the html endings. All .com endings were classified as commercial websites; noncommercial included .gov, .org, .edu, and .net endings. Next, each website was categorized based on the target audience. Websites were grouped into health professional–oriented information, patient-oriented, advertisement, or “other.” These classifications were based on those described in previous works.14,15 The “other” category included images, YouTube videos, another search engine, and open forums, which were also excluded from the final analysis because they were not easily evaluable with the DISCERN instrument. Websites were considered health professional–oriented if they included journal articles, scholarly articles, and/or rehabilitation protocols. Patient-directed websites clearly stated the information was directed to patients or provided a general overview. Advertisement included sites that displayed ads or products for sale. Websites were evaluated for quality using the DISCERN instrument (Figure 2).
DISCERN has 3 subdivision scores: the reliable score (composed of the first 8 questions), the treatment options (the next 7 questions), and 1 final question that addresses the overall quality of the website and is rated independently of the first 15 questions. DISCERN uses 2 scales, a binary scale anchored on both extremes with the number 1 equaling complete absence of the criteria being measured, and the number 5 at the upper extreme, representing completeness of the quality being assessed. In between 1 and 5 is a partial ordinal scale measuring from 2 to 4, which indicates the information is present to some extent but not complete. The ordinal scale allows ranking of the criteria being assessed. Summarizing values from each of the 2 scales poses some concern: the scale is not a true binary scale because of the ordinal scale of the middle numbers (2-4), and as such, is not amenable to being an interval scale to calculate arithmetic means. To summarize the values from the 2 scales, we calculated the harmonic mean, the arithmetic mean, the geometric mean, and the median. The means were empirically compared with the median, and we used the harmonic mean to summarize scale values because it was the best approximation of the medians.
Results
A total of 90 websites were assessed with the search term total shoulder arthroplasty and another 90 with shoulder replacement. When 37 duplicate websites for TSA and 52 for SR were eliminated, 53 (59%) and 38 (42%) unique websites were evaluated for each search term, respectively (Figure 1). (These unique websites are included in the Appendix.) Between the 2 search terms, 20 websites were duplicated. Figure 3 shows the distribution of websites by category. Total shoulder arthroplasty provided the highest percentage of health professional–oriented information; SR had the greatest percentage of patient-oriented information. Both TSA and SR had nearly the same number of advertisements and websites labeled “other.” The percentage of noncommercial websites from each search engine is represented in Figure 4. For SR, Google had 40% (12/30) noncommercial websites compared with Yahoo at 53% (16/30) and Bing at 46% (14/30). Total shoulder arthroplasty had 43% (13/30) noncommercial websites on Google, 27% (8/30) on Yahoo, and 40% (12/30) on Bing. In total, SR had more noncommercial websites, 47% (42/90), compared with 37% (33/90) for TSA.
The mean of all 3 raters for reliablity (DISCERN questions 1-8) and treatment options (DISCERN questions 9-15) is represented in the Table. For both search terms, we found that websites identified as health professional–oriented had the highest reliable mean scores, followed by patient-oriented, and advertisement at the lowest (SR: P = .054; TSA: P = .134). For SR, treatment mean scores demonstrated similar results with health professional–oriented websites receiving the highest, followed by patient-oriented and advertisement (P = .005). However, the treatment mean scores for TSA differed with patient-oriented websites receiving higher scores than health professional–oriented websites, but this was not statistically significant (P= .407). Regarding search terms, there were no significant differences between mean reliable and treatment scores across all categories.
The average overall DISCERN score for TSA websites was 2.5 (range, 1-5), compared with 2.3 (range, 1-5) for SR websites. The overall reliable score (DISCERN questions 1-8) for TSA websites was 2.6 and 2.5 for SR websites (P < .001). For TSA websites, 38% (20/53) were classified as good, having an overall DISCERN score ≥3, versus 26% (10/38) of SR websites. The overall DISCERN score for health professional–oriented websites was 2.7, patient-oriented websites received a score of 2.6, and advertisements had the lowest score at 2.4.
Discussion
Both patients and health professionals obtain information on health care subjects through the Internet, which has become the primary resource for patients.15,16 However, there are no strict regulations of the content being written. This creates a challenge for the typical user to find credible and evidence-based information, which is important because misleading information could cause undue anxiety, among other effects.17,18 The aims of this study were to determine the quality of Internet information for shoulder replacement surgeries using the medical terminology total shoulder arthroplasty (TSA) and the nontechnical term shoulder replacement (SR), and to compare the results.
After analyzing the types of websites returned for both total shoulder arthroplasty and shoulder replacement (Figure 4), it was interesting to find that using nonmedical terminology as the search term provided more noncommercial websites compared with total shoulder arthroplasty. Furthermore, Yahoo provided the highest yield of noncommercial websites at 16, with Bing at 14, when using SR as the search term. We believe the increase in noncommercial websites returned for SR was greater than for TSA because SR yielded more patient-oriented websites, which usually had html endings of .edu and .org, as shown in Figure 3 (48% of SR websites offered patient-oriented information).
Although there were more noncommercial websites for SR, the majority of the DISCERN values between the 2 search terms did not differ significantly. This is a direct result of the number of sites (20) that were duplicated across both search terms. However as seen in the Table, TSA had similar reliable mean scores for advertisements and patient-oriented websites but a slightly higher reliable score for health professional–oriented websites. We correlated this with the increased number of health professional–oriented websites returned when using TSA as the search term (Figure 3). The health professional–oriented websites explained their aims and cited their sources more consistently than did patient-oriented sites and advertisements, resulting in higher reliable scores. Although patient-oriented websites frequently lacked citations, they provided information about multiple treatment options, which were more relevant to consumers. This resulted in nearly equivalent reliable scores. Treatment means for advertisements in both SR and TSA were similar. However, treatment means for professional-oriented websites in TSA were lower than those for SR because health professional–oriented websites often were only moderately relevant to consumers, with their focus usually on 1 treatment option or on rehabilitation protocols. Although the DISCERN scores were similar between the search terms, total shoulder arthroplasty provided more websites (20) classified as good—overall DISCERN score, ≥3—than SR did (10). Advertisement websites had similar overall DISCERN scores, which we anticipated because most of the advertisements were duplicated across the search terms.
Using the 2 search terms, academic websites and commercial websites, such as WebMD, consistently received higher reliable and overall DISCERN scores. Advertisement websites, which need to deliver a clear message, frequently scored high on explicitly stating their aims and relevance to consumers, but focused on their products without discussing the benefits of other treatment options. This is significant because Internet search engines, such as Google, offer sponsor links for which organizations pay to appear at the top of the search results. This creates the potential for consumers to receive biased information because most individuals only visit the top 10 websites generated by a search engine.19
We concluded that the quality of online information relating to SR and TSA was highly variable and frequently of moderate-to-poor quality, with most overall DISCERN scores <3. The quality of information found online for this study using the DISCERN instrument is consistent with those studies using DISCERN to evaluate other medical conditions (eg, bunions, chronic pain, general anesthesia, and anterior cruciate ligament reconstruction).2,9,15,19 These studies also concluded that online information varies tremendously in quality and completeness.
This study has several limitations. Websites were searched at a single time point and, because Internet resources are frequently updated, the results of this study could vary. Furthermore, although Google, Yahoo, and Bing are 3 of the most popular search engines, these are not the only resources patients use when searching the Internet for health-related information. Other search engines, such as Pubmed.gov and MSN.com, could provide additional websites for Internet users. Lastly, although DISCERN is validated to address the quality of information available online, it does not evaluate the accuracy of the information.8 Our use of DISCERN involves 2 scales, a binary yes/no (ratings, 1 and 5) and an ordinal scale (ratings, 2-4). As such, a single mean summary statistic cannot be calculated.
Conclusion
The information available on the Internet pertaining to TSA and SR is highly variable and provides mostly moderate-to-poor quality information based on the DISCERN instrument. Many websites failed to describe the benefits and the risks of different treatment options, including nonoperative management. Health care professionals should be aware that patients often refer to the Internet as a primary resource for obtaining medical information. It is important to direct patients to websites that provide accurate information, because patients who educate themselves about their conditions and actively participate in decision-making may have improved health outcomes.20-22 Overall, academic websites and commercial websites, such as WebMD and OrthoInfo, generally had higher DISCERN scores when using either search term. Of major concern is the potential for misleading advertisements or incorrect information that can negatively affect health outcomes. This study found that using nonmedical terminology (SR) provided more noncommercial and patient-oriented websites, especially through Yahoo. This study highlights the need for more comprehensive online information pertaining to shoulder replacement that can better serve as a resource for Internet users.
1. Eysenbach G, Powell J, Kuss O, Sa ER. Empirical studies assessing the quality of health information for consumers on the world wide web: a systematic review. JAMA. 2002;287(20):2691-2700.
2. Bruce-Brand RA, Baker JF, Byrne DP, Hogan NA, McCarthy T. Assessment of the quality and content of information on anterior cruciate ligament reconstruction on the internet. Arthroscopy. 2013;29(6):1095-1100.
3. Computer and internet use in the United States: population characteristics. US Census Bureau website. http://www.census.gov/hhes/computer/. Accessed December 11, 2015.
4. Fox S, Duggan M. Health online 2013. Pew Research Center website. http://pewinternet.org/Reports/2013/Health-online.aspx. Published January 15, 2013. Accessed November 24, 2015.
5. Smith A. Older adults and technology use. Pew Research Center website. http://www.pewinternet.org/2014/04/03/older-adults-and-technology-use. Published April 3, 2014. Accessed November 24, 2015.
6. Shuyler KS, Knight KM. What are patients seeking when they turn to the internet? Qualitative content analysis of questions asked by visitors to an orthopaedics web site. J Med Internet Res. 2003;5(4):e24.
7. Meredith P, Emberton M, Wood C, Smith J. Comparison of patients’ needs for information on prostate surgery with printed materials provided by surgeons. Qual Health Care. 1995;4(1):18-23.
8. Charnock D, Shepperd S, Needham G, Gann R. DISCERN: An instrument for judging the quality of written consumer health information on treatment choices. J Epidemiol Community Health. 1999;53(2):105-111.
9. Kaicker J, Debono VB, Dang W, Buckley N, Thabane L. Assessment of the quality and variability of health information on chronic pain websites using the DISCERN instrument. BMC Med. 2010;8(1):59.
10. Griffiths KM, Christensen H. Website quality indicators for consumers. J Med Internet Res. 2005;7(5):e55.
11. Wiater JM. Shoulder joint replacement. American Academy of Orthopedic Surgeons website. http://orthoinfo.aaos.org/topic.cfm?topic=A00094. Updated December 2011. Accessed November 24, 2015.
12. Kim SH, Wise BL, Zhang Y, Szabo RM. Increasing incidence of shoulder arthroplasty in the united states. J Bone Joint Surg Am. 2011;93(24):2249-2254.
13. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. Ann Intern Med. 2009;151(4):W65-W94.
14. Nason GJ, Baker JF, Byrne DP, Noel J, Moore D, Kiely PJ. Scoliosis-specific information on the internet: has the “information highway” led to better information provision? Spine. 2012;37(21):E1364-E1369.
15. Starman JS, Gettys FK, Capo JA, Fleischli JE, Norton HJ, Karunakar MA. Quality and content of internet-based information for ten common orthopaedic sports medicine diagnoses. J Bone Joint Surg Am. 2010;92(7):1612-1618.
16. Bernstein J, Ahn J, Veillette C. The future of orthopaedic information management. J Bone Joint Surg Am. 2012;94(13):e95.
17. Berland GK, Elliott MN, Morales LS, et al. Health information on the Internet: accessibility, quality, and readability in English and Spanish. JAMA. 2001;285(20):2612-2621.
18. Fallowfield LJ, Hall A, Maguire GP, Baum M. Psychological outcomes of different treatment policies in women with early breast cancer outside a clinical trial. BMJ. 1990;301(6752):575-580.
19. Chong YM, Fraval A, Chandrananth J, Plunkett V, Tran P. Assessment of the quality of web-based information on bunions. Foot Ankle Int. 2013;34(8):1134-1139.
20. Brody DS, Miller SM, Lerman CE, Smith DG, Caputo GC. Patient perception of involvement in medical care. J Gen Intern Med. 1989;4(6):506-511.
21. Greenfield S, Kaplan S, Ware JE Jr. Expanding patient involvement in care. Effects on patient outcomes. Ann Intern Med. 1985;102(4):520-528.
22. Kaplan SH, Greenfield S, Ware JE Jr. Assessing the effects of physician-patient interactions on the outcomes of chronic disease. Med Care. 1989;27(3 suppl):S110-S127.
1. Eysenbach G, Powell J, Kuss O, Sa ER. Empirical studies assessing the quality of health information for consumers on the world wide web: a systematic review. JAMA. 2002;287(20):2691-2700.
2. Bruce-Brand RA, Baker JF, Byrne DP, Hogan NA, McCarthy T. Assessment of the quality and content of information on anterior cruciate ligament reconstruction on the internet. Arthroscopy. 2013;29(6):1095-1100.
3. Computer and internet use in the United States: population characteristics. US Census Bureau website. http://www.census.gov/hhes/computer/. Accessed December 11, 2015.
4. Fox S, Duggan M. Health online 2013. Pew Research Center website. http://pewinternet.org/Reports/2013/Health-online.aspx. Published January 15, 2013. Accessed November 24, 2015.
5. Smith A. Older adults and technology use. Pew Research Center website. http://www.pewinternet.org/2014/04/03/older-adults-and-technology-use. Published April 3, 2014. Accessed November 24, 2015.
6. Shuyler KS, Knight KM. What are patients seeking when they turn to the internet? Qualitative content analysis of questions asked by visitors to an orthopaedics web site. J Med Internet Res. 2003;5(4):e24.
7. Meredith P, Emberton M, Wood C, Smith J. Comparison of patients’ needs for information on prostate surgery with printed materials provided by surgeons. Qual Health Care. 1995;4(1):18-23.
8. Charnock D, Shepperd S, Needham G, Gann R. DISCERN: An instrument for judging the quality of written consumer health information on treatment choices. J Epidemiol Community Health. 1999;53(2):105-111.
9. Kaicker J, Debono VB, Dang W, Buckley N, Thabane L. Assessment of the quality and variability of health information on chronic pain websites using the DISCERN instrument. BMC Med. 2010;8(1):59.
10. Griffiths KM, Christensen H. Website quality indicators for consumers. J Med Internet Res. 2005;7(5):e55.
11. Wiater JM. Shoulder joint replacement. American Academy of Orthopedic Surgeons website. http://orthoinfo.aaos.org/topic.cfm?topic=A00094. Updated December 2011. Accessed November 24, 2015.
12. Kim SH, Wise BL, Zhang Y, Szabo RM. Increasing incidence of shoulder arthroplasty in the united states. J Bone Joint Surg Am. 2011;93(24):2249-2254.
13. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. Ann Intern Med. 2009;151(4):W65-W94.
14. Nason GJ, Baker JF, Byrne DP, Noel J, Moore D, Kiely PJ. Scoliosis-specific information on the internet: has the “information highway” led to better information provision? Spine. 2012;37(21):E1364-E1369.
15. Starman JS, Gettys FK, Capo JA, Fleischli JE, Norton HJ, Karunakar MA. Quality and content of internet-based information for ten common orthopaedic sports medicine diagnoses. J Bone Joint Surg Am. 2010;92(7):1612-1618.
16. Bernstein J, Ahn J, Veillette C. The future of orthopaedic information management. J Bone Joint Surg Am. 2012;94(13):e95.
17. Berland GK, Elliott MN, Morales LS, et al. Health information on the Internet: accessibility, quality, and readability in English and Spanish. JAMA. 2001;285(20):2612-2621.
18. Fallowfield LJ, Hall A, Maguire GP, Baum M. Psychological outcomes of different treatment policies in women with early breast cancer outside a clinical trial. BMJ. 1990;301(6752):575-580.
19. Chong YM, Fraval A, Chandrananth J, Plunkett V, Tran P. Assessment of the quality of web-based information on bunions. Foot Ankle Int. 2013;34(8):1134-1139.
20. Brody DS, Miller SM, Lerman CE, Smith DG, Caputo GC. Patient perception of involvement in medical care. J Gen Intern Med. 1989;4(6):506-511.
21. Greenfield S, Kaplan S, Ware JE Jr. Expanding patient involvement in care. Effects on patient outcomes. Ann Intern Med. 1985;102(4):520-528.
22. Kaplan SH, Greenfield S, Ware JE Jr. Assessing the effects of physician-patient interactions on the outcomes of chronic disease. Med Care. 1989;27(3 suppl):S110-S127.
Incidence, Risk Factors, and Outcome Trends of Acute Kidney Injury in Elective Total Hip and Knee Arthroplasty
Degenerative arthritis is a widespread chronic condition with an incidence of almost 43 million and annual health care costs of $60 billion in the United States alone.1 Although many cases can be managed symptomatically with medical therapy and intra-articular injections,2 many patients experience disease progression resulting in decreased ambulatory ability and work productivity. For these patients, elective hip and knee arthroplasties can drastically improve quality of life and functionality.3,4 Over the past decade, there has been a marked increase in the number of primary and revision total hip and knee arthroplasties performed in the United States. By 2030, the demand for primary total hip arthroplasties will grow an estimated 174%, to 572,000 procedures. Likewise, the demand for primary total knee arthroplasties is projected to grow by 673%, to 3.48 million procedures.5 However, though better surgical techniques and technology have led to improved functional outcomes, there is still substantial risk for complications in the perioperative period, especially in the geriatric population, in which substantial comorbidities are common.6-9
Acute kidney injury (AKI) is a common public health problem in hospitalized patients and in patients undergoing procedures. More than one-third of all AKI cases occur in surgical settings.10,11 Over the past decade, both community-acquired and in-hospital AKIs rapidly increased in incidence in all major clinical settings.12-14 Patients with AKI have high rates of adverse outcomes during hospitalization and discharge.11,15 Sequelae of AKIs include worsening chronic kidney disease (CKD) and progression to end-stage renal disease, necessitating either long-term dialysis or transplantation.12 This in turn leads to exacerbated disability, diminished quality of life, and disproportionate burden on health care resources.
Much of our knowledge about postoperative AKI has been derived from cardiovascular, thoracic, and abdominal surgery settings. However, there is a paucity of data on epidemiology and trends for either AKI or associated outcomes in patients undergoing major orthopedic surgery. The few studies to date either were single-center or had inadequate sample sizes for appropriately powered analysis of the risk factors and outcomes related to AKI.16
In the study reported here, we analyzed a large cohort of patients from a nationwide multicenter database to determine the incidence of and risk factors for AKI. We also examined the mortality and adverse discharges associated with AKI after major joint surgery. Lastly, we assessed temporal trends in both incidence and outcomes of AKI, including the death risk attributable to AKI.
Methods
Database
We extracted our study cohort from the Nationwide Inpatient Sample (NIS) and the National Inpatient Sample of Healthcare Cost and Utilization Project (HCUP) compiled by the Agency for Healthcare Research and Quality.17 NIS, the largest inpatient care database in the United States, stores data from almost 8 million stays in about 1000 hospitals across the country each year. Its participating hospital pool consists of about 20% of US community hospitals, resulting in a sampling frame comprising about 90% of all hospital discharges in the United States. This allows for calculation of precise, weighted nationwide estimates. Data elements within NIS are drawn from hospital discharge abstracts that indicate all procedures performed. NIS also stores information on patient characteristics, length of stay (LOS), discharge disposition, postoperative morbidity, and observed in-hospital mortality. However, it stores no information on long-term follow-up or complications after discharge.
Data Analysis
For the period 2002–2012, we queried the NIS database for hip and knee arthroplasties with primary diagnosis codes for osteoarthritis and secondary codes for AKI. We excluded patients under age 18 years and patients with diagnosis codes for hip and knee fracture/necrosis, inflammatory/infectious arthritis, or bone neoplasms (Table 1). We then extracted baseline characteristics of the study population. Patient-level characteristics included age, sex, race, quartile classification of median household income according to postal (ZIP) code, and primary payer (Medicare/Medicaid, private insurance, self-pay, no charge). Hospital-level characteristics included hospital location (urban, rural), hospital bed size (small, medium, large), region (Northeast, Midwest/North Central, South, West), and teaching status. We defined illness severity and likelihood of death using Deyo’s modification of the Charlson Comorbidity Index (CCI), which draws on principal and secondary ICD-9-CM (International Classification of Diseases, Ninth Revision-Clinical Modification) diagnosis codes, procedure codes, and patient demographics to estimate a patient’s mortality risk. This method reliably predicts mortality and readmission in the orthopedic population.18,19 We assessed the effect of AKI on 4 outcomes, including in-hospital mortality, discharge disposition, LOS, and cost of stay. Discharge disposition was grouped by either (a) home or short-term facility or (b) adverse discharge. Home or short-term facility covered routine, short-term hospital, against medical advice, home intravenous provider, another rehabilitation facility, another institution for outpatient services, institution for outpatient services, discharged alive, and destination unknown; adverse discharge covered skilled nursing facility, intermediate care, hospice home, hospice medical facility, long-term care hospital, and certified nursing facility. This dichotomization of discharge disposition is often used in studies of NIS data.20
Statistical Analyses
We compared the baseline characteristics of hospitalized patients with and without AKI. To test for significance, we used the χ2 test for categorical variables, the Student t test for normally distributed continuous variables, the Wilcoxon rank sum test for non-normally distributed continuous variables, and the Cochran-Armitage test for trends in AKI incidence. We used survey logistic regression models to calculate adjusted odds ratios (ORs) with 95% confidence intervals (95% CIs) in order to estimate the predictors of AKI and the impact of AKI on hospital outcomes. We constructed final models after adjusting for confounders, testing for potential interactions, and ensuring no multicolinearity between covariates. Last, we computed the risk proportion of death attributable to AKI, indicating the proportion of deaths that could potentially be avoided if AKI and its complications were abrogated.21
We performed all statistical analyses with SAS Version 9.3 (SAS Institute) using designated weight values to produce weighted national estimates. The threshold for statistical significance was set at P < .01 (with ORs and 95% CIs that excluded 1).
Results
AKI Incidence, Risk Factors, and Trends
We identified 7,235,251 patients who underwent elective hip or knee arthroplasty for osteoarthritis between 2002 and 2012—an estimate consistent with data from the Centers for Disease Control and Prevention.22 Of that total, 94,367 (1.3%) had AKI. The proportion of discharges diagnosed with AKI increased rapidly over the decade, from 0.5% in 2002 to 1.8% to 1.9% in the period 2010–2012. This upward trend was highly significant (Ptrend < .001) (Figure 1). Patients with AKI (vs patients without AKI) were more likely to be older (mean age, 70 vs 66 years; P < .001), male (50.8% vs 38.4%; P < .001), and black (10.07% vs 5.15%; P<. 001). They were also found to have a significantly higher comorbidity score (mean CCI, 2.8 vs 1.5; P < .001) and higher proportions of comorbidities, including hypertension, CKD, atrial fibrillation, diabetes mellitus (DM), congestive heart failure, chronic liver disease, and hepatitis C virus infection. In addition, AKI was associated with perioperative myocardial infarction (MI), sepsis, cardiac catheterization, and blood transfusion. Regarding socioeconomic characteristics, patients with AKI were more likely to have Medicare/Medicaid insurance (72.26% vs 58.06%; P < .001) and to belong to the extremes of income categories (Table 2).
Using multivariable logistic regression, we found that increased age (1.11 increase in adjusted OR for every year older; 95% CI, 1.09-1.14; P < .001), male sex (adjusted OR, 1.65; 95% CI, 1.60-1.71; P < .001), and black race (adjusted OR, 1.57; 95% CI, 1.45-1.69; P < .001) were significantly associated with postoperative AKI. Regarding comorbidities, baseline CKD (adjusted OR, 8.64; 95% CI, 8.14-9.18; P < .001) and congestive heart failure (adjusted OR, 2.74; 95% CI, 2.57-2.92; P< .0001) were most significantly associated with AKI. Perioperative events, including sepsis (adjusted OR, 35.64; 95% CI, 30.28-41.96; P < .0001), MI (adjusted OR, 6.14; 95% CI, 5.17-7.28; P < .0001), and blood transfusion (adjusted OR, 2.28; 95% CI, 2.15-2.42; P < .0001), were also strongly associated with postoperative AKI. Last, compared with urban hospitals and small hospital bed size, rural hospitals (adjusted OR, 0.70; 95% CI, 0.60-0.81; P< .001) and large bed size (adjusted OR, 0.82; 95% CI, 0.70-0.93; P = .003) were associated with lower probability of developing AKI (Table 3).
Figure 2 elucidates the frequency of AKI based on a combination of key preoperative comorbid conditions and postoperative complications—demonstrating that the proportion of AKI cases associated with other postoperative complications is significantly higher in the CKD and concomitant DM/CKD patient populations. Patients hospitalized with CKD exhibited higher rates of AKI in cases involving blood transfusion (20.9% vs 1.8%; P < .001), acute MI (48.9% vs 13.8%; P < .001), and sepsis (74.7% vs 36.3%;P< .001) relative to patients without CKD. Similarly, patients with concomitant DM/CKD exhibited higher rates of AKI in cases involving blood transfusion (23% vs 1.9%; P< .001), acute MI (51.1% vs 12.1%; P< .001), and sepsis (75% vs 38.2%; P < .001) relative to patients without either condition. However, patients hospitalized with DM alone exhibited only marginally higher rates of AKI in cases involving blood transfusion (4.7% vs 2%; P < .01) and acute MI (19.2% vs 16.7%; P< .01) and a lower rate in cases involving sepsis (38.2% vs 41.7%; P < .01) relative to patients without DM. These data suggest that CKD is the most significant clinically relevant risk factor for AKI and that CKD may synergize with DM to raise the risk for AKI.
Outcomes
We then analyzed the impact of AKI on hospital outcomes, including in-hospital mortality, discharge disposition, LOS, and cost of care. Mortality was significantly higher in patients with AKI than in patients without it (2.08% vs 0.06%; P < .001). Even after adjusting for confounders (eg, demographics, comorbidity burden, perioperative sepsis, hospital-level characteristics), AKI was still associated with strikingly higher odds of in-hospital death (adjusted OR, 11.32; 95% CI, 9.34-13.74; P < .001). However, analysis of temporal trends indicated that the odds for adjusted mortality associated with AKI decreased from 18.09 to 9.45 (Ptrend = .01) over the period 2002–2012 (Figure 3). This decrease in odds of death was countered by an increase in incidence of AKI, resulting in a stable attributable risk proportion (97.9% in 2002 to 97.3% in 2012; Ptrend = .90) (Table 4). Regarding discharge disposition, patients with AKI were much less likely to be discharged home (41.35% vs 62.59%; P < .001) and more likely to be discharged to long-term care (56.37% vs 37.03%; P< .001). After adjustment for confounders, AKI was associated with significantly increased odds of adverse discharge (adjusted OR, 2.24; 95% CI, 2.12-2.36; P< .001). Analysis of temporal trends revealed no appreciable decrease in the adjusted odds of adverse discharge between 2002 (adjusted OR, 1.87; 95% CI, 1.37-2.55; P < .001) and 2012 (adjusted OR, 1.93; 95% CI, 1.76-2.11; P < .001) (Figure 4, Table 5). Last, both mean LOS (5 days vs 3 days; P < .001) and mean cost of hospitalization (US $22,269 vs $15,757; P < .001) were significantly higher in patients with AKI.
Discussion
In this study, we found that the incidence of AKI among hospitalized patients increased 4-fold between 2002 and 2012. Moreover, we identified numerous patient-specific, hospital-specific, perioperative risk factors for AKI. Most important, we found that AKI was associated with a strikingly higher risk of in-hospital death, and surviving patients were more likely to experience adverse discharge. Although the adjusted mortality rate associated with AKI decreased over that decade, the attributable risk proportion remained stable.
Few studies have addressed this significant public health concern. In one recent study in Australia, Kimmel and colleagues16 identified risk factors for AKI but lacked data on AKI outcomes. In a study of complications and mortality occurring after orthopedic surgery, Belmont and colleagues22 categorized complications as either local or systemic but did not examine renal complications. Only 2 other major studies have been conducted on renal outcomes associated with major joint surgery, and both were limited to patients with acute hip fractures. The first included acute fracture surgery patients and omitted elective joint surgery patients, and it evaluated admission renal function but not postoperative AKI.22 The second study had a sample size of only 170 patients.23 Thus, the literature leaves us with a crucial knowledge gap in renal outcomes and their postoperative impact in elective arthroplasties.
The present study filled this information gap by examining the incidence, risk factors, outcomes, and temporal trends of AKI after elective hip and knee arthroplasties. The increasing incidence of AKI in this surgical setting is similar to that of AKI in other surgical settings (cardiac and noncardiac).21 Although our analysis was limited by lack of perioperative management data, patients undergoing elective joint arthroplasty can experience kidney dysfunction for several reasons, including volume depletion, postoperative sepsis, and influence of medications, such as nonsteroidal anti-inflammatory drugs (NSAIDs), especially in older patients with more comorbidities and a higher burden of CKD. Each of these factors can cause renal dysfunction in patients having orthopedic procedures.24 Moreover, NSAID use among elective joint arthroplasty patients is likely higher because of an emphasis on multimodal analgesia, as recent randomized controlled trials have demonstrated the efficacy of NSAID use in controlling pain without increasing bleeding.25-27 Our results also demonstrated that the absolute incidence of AKI after orthopedic surgery is relatively low. One possible explanation for this phenomenon is that the definitions used were based on ICD-9-CM codes that underestimate the true incidence of AKI.
Consistent with other studies, we found that certain key preoperative comorbid conditions and postoperative events were associated with higher AKI risk. We stratified the rate of AKI associated with each postoperative event (sepsis, acute MI, cardiac catheterization, need for transfusion) by DM/CKD comorbidity. CKD was associated with significantly higher AKI risk across all postoperative complications. This information may provide clinicians with bedside information that can be used to determine which patients may be at higher or lower risk for AKI.
Our analysis of patient outcomes revealed that, though AKI was relatively uncommon, it increased the risk for death during hospitalization more than 10-fold between 2002 and 2012. Although the adjusted OR of in-hospital mortality decreased over the decade studied, the concurrent increase in AKI incidence caused the attributable risk of death associated with AKI to essentially remain the same. This observation is consistent with recent reports from cardiac surgery settings.21 These data together suggest that ameliorating occurrences of AKI would decrease mortality and increase quality of care for patients undergoing elective joint surgeries.
We also examined the effect of AKI on resource use by studying LOS, costs, and risk for adverse discharge. Much as in other surgical settings, AKI increased both LOS and overall hospitalization costs. More important, AKI was associated with increased adverse discharge (discharge to long-term care or nursing homes). Although exact reasons are unclear, we can speculate that postoperative renal dysfunction precludes early rehabilitation, impeding desired functional outcome and disposition.28,29 Given the projected increases in primary and revision hip and knee arthroplasties,5 these data predict that the impact of AKI on health outcomes will increase alarmingly in coming years.
There are limitations to our study. First, it was based on administrative data and lacked patient-level and laboratory data. As reported, the sensitivity of AKI codes remains moderate,30 so the true burden may be higher than indicated here. As the definition of AKI was based on administrative coding, we also could not estimate severity, though previous studies have found that administrative codes typically capture a more severe form of disease.31 Another limitation is that, because the data were deidentified, we could not delineate the risk for recurrent AKI in repeated surgical procedures, though this cohort unlikely was large enough to qualitatively affect our results. The third limitation is that, though we used CCI to adjust for the comorbidity burden, we were unable to account for other unmeasured confounders associated with increased AKI incidence, such as specific medication use. In addition, given the lack of patient-level data, we could not analyze the specific factors responsible for AKI in the perioperative period. Nevertheless, the strengths of a nationally representative sample, such as large sample size and generalizability, outweigh these limitations.
Conclusion
AKI is potentially an important quality indicator of elective joint surgery, and reducing its incidence is therefore essential for quality improvement. Given that hip and knee arthroplasties are projected to increase exponentially, as is the burden of comorbid conditions in this population, postoperative AKI will continue to have an incremental impact on health and health care resources. Thus, a carefully planned approach of interdisciplinary perioperative care is warranted to reduce both the risk and the consequences of this devastating condition.
1. Reginster JY. The prevalence and burden of arthritis. Rheumatology. 2002;41(supp 1):3-6.
2. Kullenberg B, Runesson R, Tuvhag R, Olsson C, Resch S. Intraarticular corticosteroid injection: pain relief in osteoarthritis of the hip? J Rheumatol. 2004;31(11):2265-2268.
3. Kawasaki M, Hasegawa Y, Sakano S, Torii Y, Warashina H. Quality of life after several treatments for osteoarthritis of the hip. J Orthop Sci. 2003;8(1):32-35.
4. Ethgen O, Bruyère O, Richy F, Dardennes C, Reginster JY. Health-related quality of life in total hip and total knee arthroplasty. A qualitative and systematic review of the literature. J Bone Joint Surg Am. 2004;86(5):963-974.
5. Kurtz S, Ong K, Lau E, Mowat F, Halpern M. Projections of primary and revision hip and knee arthroplasty in the United States from 2005 to 2030. J Bone Joint Surg Am. 2007;89(4):780-785.
6. Matlock D, Earnest M, Epstein A. Utilization of elective hip and knee arthroplasty by age and payer. Clin Orthop Relat Res. 2008;466(4):914-919.
7. Parvizi J, Holiday AD, Ereth MH, Lewallen DG. The Frank Stinchfield Award. Sudden death during primary hip arthroplasty. Clin Orthop Relat Res. 1999;(369):39-48.
8. Parvizi J, Mui A, Purtill JJ, Sharkey PF, Hozack WJ, Rothman RH. Total joint arthroplasty: when do fatal or near-fatal complications occur? J Bone Joint Surg Am. 2007;89(1):27-32.
9. Parvizi J, Sullivan TA, Trousdale RT, Lewallen DG. Thirty-day mortality after total knee arthroplasty. J Bone Joint Surg Am. 2001;83(8):1157-1161.
10. Uchino S, Kellum JA, Bellomo R, et al; Beginning and Ending Supportive Therapy for the Kidney (BEST Kidney) Investigators. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA. 2005;294(7):813-818.
11. Thakar CV. Perioperative acute kidney injury. Adv Chronic Kidney Dis. 2013;20(1):67-75.
12. Hsu CY, Chertow GM, McCulloch CE, Fan D, Ordoñez JD, Go AS. Nonrecovery of kidney function and death after acute on chronic renal failure. Clin J Am Soc Nephrol. 2009;4(5):891-898.
13. Rewa O, Bagshaw SM. Acute kidney injury—epidemiology, outcomes and economics. Nat Rev Nephrol. 2014;10(4):193-207.
14. Thakar CV, Worley S, Arrigain S, Yared JP, Paganini EP. Influence of renal dysfunction on mortality after cardiac surgery: modifying effect of preoperative renal function. Kidney Int. 2005;67(3):1112-1119.
15. Zeng X, McMahon GM, Brunelli SM, Bates DW, Waikar SS. Incidence, outcomes, and comparisons across definitions of AKI in hospitalized individuals. Clin J Am Soc Nephrol. 2014;9(1):12-20.
16. Kimmel LA, Wilson S, Janardan JD, Liew SM, Walker RG. Incidence of acute kidney injury following total joint arthroplasty: a retrospective review by RIFLE criteria. Clin Kidney J. 2014;7(6):546-551.
17. Agency for Healthcare Research and Quality. Healthcare Cost and Utilization Project (HCUP) databases, 2002–2012. Rockville, MD: Agency for Healthcare Research and Quality.
18. Bjorgul K, Novicoff WM, Saleh KJ. Evaluating comorbidities in total hip and knee arthroplasty: available instruments. J Orthop Traumatol. 2010;11(4):203-209.
19. Voskuijl T, Hageman M, Ring D. Higher Charlson Comorbidity Index Scores are associated with readmission after orthopaedic surgery. Clin Orthop Relat Res. 2014;472(5):1638-1644.
20. Chertow GM, Burdick E, Honour M, Bonventre JV, Bates DW. Acute kidney injury, mortality, length of stay, and costs in hospitalized patients. J Am Soc Nephrol. 2005;16(11):3365-3370.
21. Lenihan CR, Montez-Rath ME, Mora Mangano CT, Chertow GM, Winkelmayer WC. Trends in acute kidney injury, associated use of dialysis, and mortality after cardiac surgery, 1999 to 2008. Ann Thorac Surg. 2013;95(1):20-28.
22. Belmont PJ Jr, Goodman GP, Waterman BR, Bader JO, Schoenfeld AJ. Thirty-day postoperative complications and mortality following total knee arthroplasty: incidence and risk factors among a national sample of 15,321 patients. J Bone Joint Surg Am. 2014;96(1):20-26.
23. Bennet SJ, Berry OM, Goddard J, Keating JF. Acute renal dysfunction following hip fracture. Injury. 2010;41(4):335-338.
24. Kateros K, Doulgerakis C, Galanakos SP, Sakellariou VI, Papadakis SA, Macheras GA. Analysis of kidney dysfunction in orthopaedic patients. BMC Nephrol. 2012;13:101.
25. Huang YM, Wang CM, Wang CT, Lin WP, Horng LC, Jiang CC. Perioperative celecoxib administration for pain management after total knee arthroplasty—a randomized, controlled study. BMC Musculoskelet Disord. 2008;9:77.
26. Kelley TC, Adams MJ, Mulliken BD, Dalury DF. Efficacy of multimodal perioperative analgesia protocol with periarticular medication injection in total knee arthroplasty: a randomized, double-blinded study. J Arthroplasty. 2013;28(8):1274-1277.
27. Lamplot JD, Wagner ER, Manning DW. Multimodal pain management in total knee arthroplasty: a prospective randomized controlled trial. J Arthroplasty. 2014;29(2):329-334.
28. Munin MC, Rudy TE, Glynn NW, Crossett LS, Rubash HE. Early inpatient rehabilitation after elective hip and knee arthroplasty. JAMA. 1998;279(11):847-852.
29. Pua YH, Ong PH. Association of early ambulation with length of stay and costs in total knee arthroplasty: retrospective cohort study. Am J Phys Med Rehabil. 2014;93(11):962-970.
30. Waikar SS, Wald R, Chertow GM, et al. Validity of International Classification of Diseases, Ninth Revision, Clinical Modification codes for acute renal failure. J Am Soc Nephrol. 2006;17(6):1688-1694.
31. Grams ME, Waikar SS, MacMahon B, Whelton S, Ballew SH, Coresh J. Performance and limitations of administrative data in the identification of AKI. Clin J Am Soc Nephrol. 2014;9(4):682-689.
Degenerative arthritis is a widespread chronic condition with an incidence of almost 43 million and annual health care costs of $60 billion in the United States alone.1 Although many cases can be managed symptomatically with medical therapy and intra-articular injections,2 many patients experience disease progression resulting in decreased ambulatory ability and work productivity. For these patients, elective hip and knee arthroplasties can drastically improve quality of life and functionality.3,4 Over the past decade, there has been a marked increase in the number of primary and revision total hip and knee arthroplasties performed in the United States. By 2030, the demand for primary total hip arthroplasties will grow an estimated 174%, to 572,000 procedures. Likewise, the demand for primary total knee arthroplasties is projected to grow by 673%, to 3.48 million procedures.5 However, though better surgical techniques and technology have led to improved functional outcomes, there is still substantial risk for complications in the perioperative period, especially in the geriatric population, in which substantial comorbidities are common.6-9
Acute kidney injury (AKI) is a common public health problem in hospitalized patients and in patients undergoing procedures. More than one-third of all AKI cases occur in surgical settings.10,11 Over the past decade, both community-acquired and in-hospital AKIs rapidly increased in incidence in all major clinical settings.12-14 Patients with AKI have high rates of adverse outcomes during hospitalization and discharge.11,15 Sequelae of AKIs include worsening chronic kidney disease (CKD) and progression to end-stage renal disease, necessitating either long-term dialysis or transplantation.12 This in turn leads to exacerbated disability, diminished quality of life, and disproportionate burden on health care resources.
Much of our knowledge about postoperative AKI has been derived from cardiovascular, thoracic, and abdominal surgery settings. However, there is a paucity of data on epidemiology and trends for either AKI or associated outcomes in patients undergoing major orthopedic surgery. The few studies to date either were single-center or had inadequate sample sizes for appropriately powered analysis of the risk factors and outcomes related to AKI.16
In the study reported here, we analyzed a large cohort of patients from a nationwide multicenter database to determine the incidence of and risk factors for AKI. We also examined the mortality and adverse discharges associated with AKI after major joint surgery. Lastly, we assessed temporal trends in both incidence and outcomes of AKI, including the death risk attributable to AKI.
Methods
Database
We extracted our study cohort from the Nationwide Inpatient Sample (NIS) and the National Inpatient Sample of Healthcare Cost and Utilization Project (HCUP) compiled by the Agency for Healthcare Research and Quality.17 NIS, the largest inpatient care database in the United States, stores data from almost 8 million stays in about 1000 hospitals across the country each year. Its participating hospital pool consists of about 20% of US community hospitals, resulting in a sampling frame comprising about 90% of all hospital discharges in the United States. This allows for calculation of precise, weighted nationwide estimates. Data elements within NIS are drawn from hospital discharge abstracts that indicate all procedures performed. NIS also stores information on patient characteristics, length of stay (LOS), discharge disposition, postoperative morbidity, and observed in-hospital mortality. However, it stores no information on long-term follow-up or complications after discharge.
Data Analysis
For the period 2002–2012, we queried the NIS database for hip and knee arthroplasties with primary diagnosis codes for osteoarthritis and secondary codes for AKI. We excluded patients under age 18 years and patients with diagnosis codes for hip and knee fracture/necrosis, inflammatory/infectious arthritis, or bone neoplasms (Table 1). We then extracted baseline characteristics of the study population. Patient-level characteristics included age, sex, race, quartile classification of median household income according to postal (ZIP) code, and primary payer (Medicare/Medicaid, private insurance, self-pay, no charge). Hospital-level characteristics included hospital location (urban, rural), hospital bed size (small, medium, large), region (Northeast, Midwest/North Central, South, West), and teaching status. We defined illness severity and likelihood of death using Deyo’s modification of the Charlson Comorbidity Index (CCI), which draws on principal and secondary ICD-9-CM (International Classification of Diseases, Ninth Revision-Clinical Modification) diagnosis codes, procedure codes, and patient demographics to estimate a patient’s mortality risk. This method reliably predicts mortality and readmission in the orthopedic population.18,19 We assessed the effect of AKI on 4 outcomes, including in-hospital mortality, discharge disposition, LOS, and cost of stay. Discharge disposition was grouped by either (a) home or short-term facility or (b) adverse discharge. Home or short-term facility covered routine, short-term hospital, against medical advice, home intravenous provider, another rehabilitation facility, another institution for outpatient services, institution for outpatient services, discharged alive, and destination unknown; adverse discharge covered skilled nursing facility, intermediate care, hospice home, hospice medical facility, long-term care hospital, and certified nursing facility. This dichotomization of discharge disposition is often used in studies of NIS data.20
Statistical Analyses
We compared the baseline characteristics of hospitalized patients with and without AKI. To test for significance, we used the χ2 test for categorical variables, the Student t test for normally distributed continuous variables, the Wilcoxon rank sum test for non-normally distributed continuous variables, and the Cochran-Armitage test for trends in AKI incidence. We used survey logistic regression models to calculate adjusted odds ratios (ORs) with 95% confidence intervals (95% CIs) in order to estimate the predictors of AKI and the impact of AKI on hospital outcomes. We constructed final models after adjusting for confounders, testing for potential interactions, and ensuring no multicolinearity between covariates. Last, we computed the risk proportion of death attributable to AKI, indicating the proportion of deaths that could potentially be avoided if AKI and its complications were abrogated.21
We performed all statistical analyses with SAS Version 9.3 (SAS Institute) using designated weight values to produce weighted national estimates. The threshold for statistical significance was set at P < .01 (with ORs and 95% CIs that excluded 1).
Results
AKI Incidence, Risk Factors, and Trends
We identified 7,235,251 patients who underwent elective hip or knee arthroplasty for osteoarthritis between 2002 and 2012—an estimate consistent with data from the Centers for Disease Control and Prevention.22 Of that total, 94,367 (1.3%) had AKI. The proportion of discharges diagnosed with AKI increased rapidly over the decade, from 0.5% in 2002 to 1.8% to 1.9% in the period 2010–2012. This upward trend was highly significant (Ptrend < .001) (Figure 1). Patients with AKI (vs patients without AKI) were more likely to be older (mean age, 70 vs 66 years; P < .001), male (50.8% vs 38.4%; P < .001), and black (10.07% vs 5.15%; P<. 001). They were also found to have a significantly higher comorbidity score (mean CCI, 2.8 vs 1.5; P < .001) and higher proportions of comorbidities, including hypertension, CKD, atrial fibrillation, diabetes mellitus (DM), congestive heart failure, chronic liver disease, and hepatitis C virus infection. In addition, AKI was associated with perioperative myocardial infarction (MI), sepsis, cardiac catheterization, and blood transfusion. Regarding socioeconomic characteristics, patients with AKI were more likely to have Medicare/Medicaid insurance (72.26% vs 58.06%; P < .001) and to belong to the extremes of income categories (Table 2).
Using multivariable logistic regression, we found that increased age (1.11 increase in adjusted OR for every year older; 95% CI, 1.09-1.14; P < .001), male sex (adjusted OR, 1.65; 95% CI, 1.60-1.71; P < .001), and black race (adjusted OR, 1.57; 95% CI, 1.45-1.69; P < .001) were significantly associated with postoperative AKI. Regarding comorbidities, baseline CKD (adjusted OR, 8.64; 95% CI, 8.14-9.18; P < .001) and congestive heart failure (adjusted OR, 2.74; 95% CI, 2.57-2.92; P< .0001) were most significantly associated with AKI. Perioperative events, including sepsis (adjusted OR, 35.64; 95% CI, 30.28-41.96; P < .0001), MI (adjusted OR, 6.14; 95% CI, 5.17-7.28; P < .0001), and blood transfusion (adjusted OR, 2.28; 95% CI, 2.15-2.42; P < .0001), were also strongly associated with postoperative AKI. Last, compared with urban hospitals and small hospital bed size, rural hospitals (adjusted OR, 0.70; 95% CI, 0.60-0.81; P< .001) and large bed size (adjusted OR, 0.82; 95% CI, 0.70-0.93; P = .003) were associated with lower probability of developing AKI (Table 3).
Figure 2 elucidates the frequency of AKI based on a combination of key preoperative comorbid conditions and postoperative complications—demonstrating that the proportion of AKI cases associated with other postoperative complications is significantly higher in the CKD and concomitant DM/CKD patient populations. Patients hospitalized with CKD exhibited higher rates of AKI in cases involving blood transfusion (20.9% vs 1.8%; P < .001), acute MI (48.9% vs 13.8%; P < .001), and sepsis (74.7% vs 36.3%;P< .001) relative to patients without CKD. Similarly, patients with concomitant DM/CKD exhibited higher rates of AKI in cases involving blood transfusion (23% vs 1.9%; P< .001), acute MI (51.1% vs 12.1%; P< .001), and sepsis (75% vs 38.2%; P < .001) relative to patients without either condition. However, patients hospitalized with DM alone exhibited only marginally higher rates of AKI in cases involving blood transfusion (4.7% vs 2%; P < .01) and acute MI (19.2% vs 16.7%; P< .01) and a lower rate in cases involving sepsis (38.2% vs 41.7%; P < .01) relative to patients without DM. These data suggest that CKD is the most significant clinically relevant risk factor for AKI and that CKD may synergize with DM to raise the risk for AKI.
Outcomes
We then analyzed the impact of AKI on hospital outcomes, including in-hospital mortality, discharge disposition, LOS, and cost of care. Mortality was significantly higher in patients with AKI than in patients without it (2.08% vs 0.06%; P < .001). Even after adjusting for confounders (eg, demographics, comorbidity burden, perioperative sepsis, hospital-level characteristics), AKI was still associated with strikingly higher odds of in-hospital death (adjusted OR, 11.32; 95% CI, 9.34-13.74; P < .001). However, analysis of temporal trends indicated that the odds for adjusted mortality associated with AKI decreased from 18.09 to 9.45 (Ptrend = .01) over the period 2002–2012 (Figure 3). This decrease in odds of death was countered by an increase in incidence of AKI, resulting in a stable attributable risk proportion (97.9% in 2002 to 97.3% in 2012; Ptrend = .90) (Table 4). Regarding discharge disposition, patients with AKI were much less likely to be discharged home (41.35% vs 62.59%; P < .001) and more likely to be discharged to long-term care (56.37% vs 37.03%; P< .001). After adjustment for confounders, AKI was associated with significantly increased odds of adverse discharge (adjusted OR, 2.24; 95% CI, 2.12-2.36; P< .001). Analysis of temporal trends revealed no appreciable decrease in the adjusted odds of adverse discharge between 2002 (adjusted OR, 1.87; 95% CI, 1.37-2.55; P < .001) and 2012 (adjusted OR, 1.93; 95% CI, 1.76-2.11; P < .001) (Figure 4, Table 5). Last, both mean LOS (5 days vs 3 days; P < .001) and mean cost of hospitalization (US $22,269 vs $15,757; P < .001) were significantly higher in patients with AKI.
Discussion
In this study, we found that the incidence of AKI among hospitalized patients increased 4-fold between 2002 and 2012. Moreover, we identified numerous patient-specific, hospital-specific, perioperative risk factors for AKI. Most important, we found that AKI was associated with a strikingly higher risk of in-hospital death, and surviving patients were more likely to experience adverse discharge. Although the adjusted mortality rate associated with AKI decreased over that decade, the attributable risk proportion remained stable.
Few studies have addressed this significant public health concern. In one recent study in Australia, Kimmel and colleagues16 identified risk factors for AKI but lacked data on AKI outcomes. In a study of complications and mortality occurring after orthopedic surgery, Belmont and colleagues22 categorized complications as either local or systemic but did not examine renal complications. Only 2 other major studies have been conducted on renal outcomes associated with major joint surgery, and both were limited to patients with acute hip fractures. The first included acute fracture surgery patients and omitted elective joint surgery patients, and it evaluated admission renal function but not postoperative AKI.22 The second study had a sample size of only 170 patients.23 Thus, the literature leaves us with a crucial knowledge gap in renal outcomes and their postoperative impact in elective arthroplasties.
The present study filled this information gap by examining the incidence, risk factors, outcomes, and temporal trends of AKI after elective hip and knee arthroplasties. The increasing incidence of AKI in this surgical setting is similar to that of AKI in other surgical settings (cardiac and noncardiac).21 Although our analysis was limited by lack of perioperative management data, patients undergoing elective joint arthroplasty can experience kidney dysfunction for several reasons, including volume depletion, postoperative sepsis, and influence of medications, such as nonsteroidal anti-inflammatory drugs (NSAIDs), especially in older patients with more comorbidities and a higher burden of CKD. Each of these factors can cause renal dysfunction in patients having orthopedic procedures.24 Moreover, NSAID use among elective joint arthroplasty patients is likely higher because of an emphasis on multimodal analgesia, as recent randomized controlled trials have demonstrated the efficacy of NSAID use in controlling pain without increasing bleeding.25-27 Our results also demonstrated that the absolute incidence of AKI after orthopedic surgery is relatively low. One possible explanation for this phenomenon is that the definitions used were based on ICD-9-CM codes that underestimate the true incidence of AKI.
Consistent with other studies, we found that certain key preoperative comorbid conditions and postoperative events were associated with higher AKI risk. We stratified the rate of AKI associated with each postoperative event (sepsis, acute MI, cardiac catheterization, need for transfusion) by DM/CKD comorbidity. CKD was associated with significantly higher AKI risk across all postoperative complications. This information may provide clinicians with bedside information that can be used to determine which patients may be at higher or lower risk for AKI.
Our analysis of patient outcomes revealed that, though AKI was relatively uncommon, it increased the risk for death during hospitalization more than 10-fold between 2002 and 2012. Although the adjusted OR of in-hospital mortality decreased over the decade studied, the concurrent increase in AKI incidence caused the attributable risk of death associated with AKI to essentially remain the same. This observation is consistent with recent reports from cardiac surgery settings.21 These data together suggest that ameliorating occurrences of AKI would decrease mortality and increase quality of care for patients undergoing elective joint surgeries.
We also examined the effect of AKI on resource use by studying LOS, costs, and risk for adverse discharge. Much as in other surgical settings, AKI increased both LOS and overall hospitalization costs. More important, AKI was associated with increased adverse discharge (discharge to long-term care or nursing homes). Although exact reasons are unclear, we can speculate that postoperative renal dysfunction precludes early rehabilitation, impeding desired functional outcome and disposition.28,29 Given the projected increases in primary and revision hip and knee arthroplasties,5 these data predict that the impact of AKI on health outcomes will increase alarmingly in coming years.
There are limitations to our study. First, it was based on administrative data and lacked patient-level and laboratory data. As reported, the sensitivity of AKI codes remains moderate,30 so the true burden may be higher than indicated here. As the definition of AKI was based on administrative coding, we also could not estimate severity, though previous studies have found that administrative codes typically capture a more severe form of disease.31 Another limitation is that, because the data were deidentified, we could not delineate the risk for recurrent AKI in repeated surgical procedures, though this cohort unlikely was large enough to qualitatively affect our results. The third limitation is that, though we used CCI to adjust for the comorbidity burden, we were unable to account for other unmeasured confounders associated with increased AKI incidence, such as specific medication use. In addition, given the lack of patient-level data, we could not analyze the specific factors responsible for AKI in the perioperative period. Nevertheless, the strengths of a nationally representative sample, such as large sample size and generalizability, outweigh these limitations.
Conclusion
AKI is potentially an important quality indicator of elective joint surgery, and reducing its incidence is therefore essential for quality improvement. Given that hip and knee arthroplasties are projected to increase exponentially, as is the burden of comorbid conditions in this population, postoperative AKI will continue to have an incremental impact on health and health care resources. Thus, a carefully planned approach of interdisciplinary perioperative care is warranted to reduce both the risk and the consequences of this devastating condition.
Degenerative arthritis is a widespread chronic condition with an incidence of almost 43 million and annual health care costs of $60 billion in the United States alone.1 Although many cases can be managed symptomatically with medical therapy and intra-articular injections,2 many patients experience disease progression resulting in decreased ambulatory ability and work productivity. For these patients, elective hip and knee arthroplasties can drastically improve quality of life and functionality.3,4 Over the past decade, there has been a marked increase in the number of primary and revision total hip and knee arthroplasties performed in the United States. By 2030, the demand for primary total hip arthroplasties will grow an estimated 174%, to 572,000 procedures. Likewise, the demand for primary total knee arthroplasties is projected to grow by 673%, to 3.48 million procedures.5 However, though better surgical techniques and technology have led to improved functional outcomes, there is still substantial risk for complications in the perioperative period, especially in the geriatric population, in which substantial comorbidities are common.6-9
Acute kidney injury (AKI) is a common public health problem in hospitalized patients and in patients undergoing procedures. More than one-third of all AKI cases occur in surgical settings.10,11 Over the past decade, both community-acquired and in-hospital AKIs rapidly increased in incidence in all major clinical settings.12-14 Patients with AKI have high rates of adverse outcomes during hospitalization and discharge.11,15 Sequelae of AKIs include worsening chronic kidney disease (CKD) and progression to end-stage renal disease, necessitating either long-term dialysis or transplantation.12 This in turn leads to exacerbated disability, diminished quality of life, and disproportionate burden on health care resources.
Much of our knowledge about postoperative AKI has been derived from cardiovascular, thoracic, and abdominal surgery settings. However, there is a paucity of data on epidemiology and trends for either AKI or associated outcomes in patients undergoing major orthopedic surgery. The few studies to date either were single-center or had inadequate sample sizes for appropriately powered analysis of the risk factors and outcomes related to AKI.16
In the study reported here, we analyzed a large cohort of patients from a nationwide multicenter database to determine the incidence of and risk factors for AKI. We also examined the mortality and adverse discharges associated with AKI after major joint surgery. Lastly, we assessed temporal trends in both incidence and outcomes of AKI, including the death risk attributable to AKI.
Methods
Database
We extracted our study cohort from the Nationwide Inpatient Sample (NIS) and the National Inpatient Sample of Healthcare Cost and Utilization Project (HCUP) compiled by the Agency for Healthcare Research and Quality.17 NIS, the largest inpatient care database in the United States, stores data from almost 8 million stays in about 1000 hospitals across the country each year. Its participating hospital pool consists of about 20% of US community hospitals, resulting in a sampling frame comprising about 90% of all hospital discharges in the United States. This allows for calculation of precise, weighted nationwide estimates. Data elements within NIS are drawn from hospital discharge abstracts that indicate all procedures performed. NIS also stores information on patient characteristics, length of stay (LOS), discharge disposition, postoperative morbidity, and observed in-hospital mortality. However, it stores no information on long-term follow-up or complications after discharge.
Data Analysis
For the period 2002–2012, we queried the NIS database for hip and knee arthroplasties with primary diagnosis codes for osteoarthritis and secondary codes for AKI. We excluded patients under age 18 years and patients with diagnosis codes for hip and knee fracture/necrosis, inflammatory/infectious arthritis, or bone neoplasms (Table 1). We then extracted baseline characteristics of the study population. Patient-level characteristics included age, sex, race, quartile classification of median household income according to postal (ZIP) code, and primary payer (Medicare/Medicaid, private insurance, self-pay, no charge). Hospital-level characteristics included hospital location (urban, rural), hospital bed size (small, medium, large), region (Northeast, Midwest/North Central, South, West), and teaching status. We defined illness severity and likelihood of death using Deyo’s modification of the Charlson Comorbidity Index (CCI), which draws on principal and secondary ICD-9-CM (International Classification of Diseases, Ninth Revision-Clinical Modification) diagnosis codes, procedure codes, and patient demographics to estimate a patient’s mortality risk. This method reliably predicts mortality and readmission in the orthopedic population.18,19 We assessed the effect of AKI on 4 outcomes, including in-hospital mortality, discharge disposition, LOS, and cost of stay. Discharge disposition was grouped by either (a) home or short-term facility or (b) adverse discharge. Home or short-term facility covered routine, short-term hospital, against medical advice, home intravenous provider, another rehabilitation facility, another institution for outpatient services, institution for outpatient services, discharged alive, and destination unknown; adverse discharge covered skilled nursing facility, intermediate care, hospice home, hospice medical facility, long-term care hospital, and certified nursing facility. This dichotomization of discharge disposition is often used in studies of NIS data.20
Statistical Analyses
We compared the baseline characteristics of hospitalized patients with and without AKI. To test for significance, we used the χ2 test for categorical variables, the Student t test for normally distributed continuous variables, the Wilcoxon rank sum test for non-normally distributed continuous variables, and the Cochran-Armitage test for trends in AKI incidence. We used survey logistic regression models to calculate adjusted odds ratios (ORs) with 95% confidence intervals (95% CIs) in order to estimate the predictors of AKI and the impact of AKI on hospital outcomes. We constructed final models after adjusting for confounders, testing for potential interactions, and ensuring no multicolinearity between covariates. Last, we computed the risk proportion of death attributable to AKI, indicating the proportion of deaths that could potentially be avoided if AKI and its complications were abrogated.21
We performed all statistical analyses with SAS Version 9.3 (SAS Institute) using designated weight values to produce weighted national estimates. The threshold for statistical significance was set at P < .01 (with ORs and 95% CIs that excluded 1).
Results
AKI Incidence, Risk Factors, and Trends
We identified 7,235,251 patients who underwent elective hip or knee arthroplasty for osteoarthritis between 2002 and 2012—an estimate consistent with data from the Centers for Disease Control and Prevention.22 Of that total, 94,367 (1.3%) had AKI. The proportion of discharges diagnosed with AKI increased rapidly over the decade, from 0.5% in 2002 to 1.8% to 1.9% in the period 2010–2012. This upward trend was highly significant (Ptrend < .001) (Figure 1). Patients with AKI (vs patients without AKI) were more likely to be older (mean age, 70 vs 66 years; P < .001), male (50.8% vs 38.4%; P < .001), and black (10.07% vs 5.15%; P<. 001). They were also found to have a significantly higher comorbidity score (mean CCI, 2.8 vs 1.5; P < .001) and higher proportions of comorbidities, including hypertension, CKD, atrial fibrillation, diabetes mellitus (DM), congestive heart failure, chronic liver disease, and hepatitis C virus infection. In addition, AKI was associated with perioperative myocardial infarction (MI), sepsis, cardiac catheterization, and blood transfusion. Regarding socioeconomic characteristics, patients with AKI were more likely to have Medicare/Medicaid insurance (72.26% vs 58.06%; P < .001) and to belong to the extremes of income categories (Table 2).
Using multivariable logistic regression, we found that increased age (1.11 increase in adjusted OR for every year older; 95% CI, 1.09-1.14; P < .001), male sex (adjusted OR, 1.65; 95% CI, 1.60-1.71; P < .001), and black race (adjusted OR, 1.57; 95% CI, 1.45-1.69; P < .001) were significantly associated with postoperative AKI. Regarding comorbidities, baseline CKD (adjusted OR, 8.64; 95% CI, 8.14-9.18; P < .001) and congestive heart failure (adjusted OR, 2.74; 95% CI, 2.57-2.92; P< .0001) were most significantly associated with AKI. Perioperative events, including sepsis (adjusted OR, 35.64; 95% CI, 30.28-41.96; P < .0001), MI (adjusted OR, 6.14; 95% CI, 5.17-7.28; P < .0001), and blood transfusion (adjusted OR, 2.28; 95% CI, 2.15-2.42; P < .0001), were also strongly associated with postoperative AKI. Last, compared with urban hospitals and small hospital bed size, rural hospitals (adjusted OR, 0.70; 95% CI, 0.60-0.81; P< .001) and large bed size (adjusted OR, 0.82; 95% CI, 0.70-0.93; P = .003) were associated with lower probability of developing AKI (Table 3).
Figure 2 elucidates the frequency of AKI based on a combination of key preoperative comorbid conditions and postoperative complications—demonstrating that the proportion of AKI cases associated with other postoperative complications is significantly higher in the CKD and concomitant DM/CKD patient populations. Patients hospitalized with CKD exhibited higher rates of AKI in cases involving blood transfusion (20.9% vs 1.8%; P < .001), acute MI (48.9% vs 13.8%; P < .001), and sepsis (74.7% vs 36.3%;P< .001) relative to patients without CKD. Similarly, patients with concomitant DM/CKD exhibited higher rates of AKI in cases involving blood transfusion (23% vs 1.9%; P< .001), acute MI (51.1% vs 12.1%; P< .001), and sepsis (75% vs 38.2%; P < .001) relative to patients without either condition. However, patients hospitalized with DM alone exhibited only marginally higher rates of AKI in cases involving blood transfusion (4.7% vs 2%; P < .01) and acute MI (19.2% vs 16.7%; P< .01) and a lower rate in cases involving sepsis (38.2% vs 41.7%; P < .01) relative to patients without DM. These data suggest that CKD is the most significant clinically relevant risk factor for AKI and that CKD may synergize with DM to raise the risk for AKI.
Outcomes
We then analyzed the impact of AKI on hospital outcomes, including in-hospital mortality, discharge disposition, LOS, and cost of care. Mortality was significantly higher in patients with AKI than in patients without it (2.08% vs 0.06%; P < .001). Even after adjusting for confounders (eg, demographics, comorbidity burden, perioperative sepsis, hospital-level characteristics), AKI was still associated with strikingly higher odds of in-hospital death (adjusted OR, 11.32; 95% CI, 9.34-13.74; P < .001). However, analysis of temporal trends indicated that the odds for adjusted mortality associated with AKI decreased from 18.09 to 9.45 (Ptrend = .01) over the period 2002–2012 (Figure 3). This decrease in odds of death was countered by an increase in incidence of AKI, resulting in a stable attributable risk proportion (97.9% in 2002 to 97.3% in 2012; Ptrend = .90) (Table 4). Regarding discharge disposition, patients with AKI were much less likely to be discharged home (41.35% vs 62.59%; P < .001) and more likely to be discharged to long-term care (56.37% vs 37.03%; P< .001). After adjustment for confounders, AKI was associated with significantly increased odds of adverse discharge (adjusted OR, 2.24; 95% CI, 2.12-2.36; P< .001). Analysis of temporal trends revealed no appreciable decrease in the adjusted odds of adverse discharge between 2002 (adjusted OR, 1.87; 95% CI, 1.37-2.55; P < .001) and 2012 (adjusted OR, 1.93; 95% CI, 1.76-2.11; P < .001) (Figure 4, Table 5). Last, both mean LOS (5 days vs 3 days; P < .001) and mean cost of hospitalization (US $22,269 vs $15,757; P < .001) were significantly higher in patients with AKI.
Discussion
In this study, we found that the incidence of AKI among hospitalized patients increased 4-fold between 2002 and 2012. Moreover, we identified numerous patient-specific, hospital-specific, perioperative risk factors for AKI. Most important, we found that AKI was associated with a strikingly higher risk of in-hospital death, and surviving patients were more likely to experience adverse discharge. Although the adjusted mortality rate associated with AKI decreased over that decade, the attributable risk proportion remained stable.
Few studies have addressed this significant public health concern. In one recent study in Australia, Kimmel and colleagues16 identified risk factors for AKI but lacked data on AKI outcomes. In a study of complications and mortality occurring after orthopedic surgery, Belmont and colleagues22 categorized complications as either local or systemic but did not examine renal complications. Only 2 other major studies have been conducted on renal outcomes associated with major joint surgery, and both were limited to patients with acute hip fractures. The first included acute fracture surgery patients and omitted elective joint surgery patients, and it evaluated admission renal function but not postoperative AKI.22 The second study had a sample size of only 170 patients.23 Thus, the literature leaves us with a crucial knowledge gap in renal outcomes and their postoperative impact in elective arthroplasties.
The present study filled this information gap by examining the incidence, risk factors, outcomes, and temporal trends of AKI after elective hip and knee arthroplasties. The increasing incidence of AKI in this surgical setting is similar to that of AKI in other surgical settings (cardiac and noncardiac).21 Although our analysis was limited by lack of perioperative management data, patients undergoing elective joint arthroplasty can experience kidney dysfunction for several reasons, including volume depletion, postoperative sepsis, and influence of medications, such as nonsteroidal anti-inflammatory drugs (NSAIDs), especially in older patients with more comorbidities and a higher burden of CKD. Each of these factors can cause renal dysfunction in patients having orthopedic procedures.24 Moreover, NSAID use among elective joint arthroplasty patients is likely higher because of an emphasis on multimodal analgesia, as recent randomized controlled trials have demonstrated the efficacy of NSAID use in controlling pain without increasing bleeding.25-27 Our results also demonstrated that the absolute incidence of AKI after orthopedic surgery is relatively low. One possible explanation for this phenomenon is that the definitions used were based on ICD-9-CM codes that underestimate the true incidence of AKI.
Consistent with other studies, we found that certain key preoperative comorbid conditions and postoperative events were associated with higher AKI risk. We stratified the rate of AKI associated with each postoperative event (sepsis, acute MI, cardiac catheterization, need for transfusion) by DM/CKD comorbidity. CKD was associated with significantly higher AKI risk across all postoperative complications. This information may provide clinicians with bedside information that can be used to determine which patients may be at higher or lower risk for AKI.
Our analysis of patient outcomes revealed that, though AKI was relatively uncommon, it increased the risk for death during hospitalization more than 10-fold between 2002 and 2012. Although the adjusted OR of in-hospital mortality decreased over the decade studied, the concurrent increase in AKI incidence caused the attributable risk of death associated with AKI to essentially remain the same. This observation is consistent with recent reports from cardiac surgery settings.21 These data together suggest that ameliorating occurrences of AKI would decrease mortality and increase quality of care for patients undergoing elective joint surgeries.
We also examined the effect of AKI on resource use by studying LOS, costs, and risk for adverse discharge. Much as in other surgical settings, AKI increased both LOS and overall hospitalization costs. More important, AKI was associated with increased adverse discharge (discharge to long-term care or nursing homes). Although exact reasons are unclear, we can speculate that postoperative renal dysfunction precludes early rehabilitation, impeding desired functional outcome and disposition.28,29 Given the projected increases in primary and revision hip and knee arthroplasties,5 these data predict that the impact of AKI on health outcomes will increase alarmingly in coming years.
There are limitations to our study. First, it was based on administrative data and lacked patient-level and laboratory data. As reported, the sensitivity of AKI codes remains moderate,30 so the true burden may be higher than indicated here. As the definition of AKI was based on administrative coding, we also could not estimate severity, though previous studies have found that administrative codes typically capture a more severe form of disease.31 Another limitation is that, because the data were deidentified, we could not delineate the risk for recurrent AKI in repeated surgical procedures, though this cohort unlikely was large enough to qualitatively affect our results. The third limitation is that, though we used CCI to adjust for the comorbidity burden, we were unable to account for other unmeasured confounders associated with increased AKI incidence, such as specific medication use. In addition, given the lack of patient-level data, we could not analyze the specific factors responsible for AKI in the perioperative period. Nevertheless, the strengths of a nationally representative sample, such as large sample size and generalizability, outweigh these limitations.
Conclusion
AKI is potentially an important quality indicator of elective joint surgery, and reducing its incidence is therefore essential for quality improvement. Given that hip and knee arthroplasties are projected to increase exponentially, as is the burden of comorbid conditions in this population, postoperative AKI will continue to have an incremental impact on health and health care resources. Thus, a carefully planned approach of interdisciplinary perioperative care is warranted to reduce both the risk and the consequences of this devastating condition.
1. Reginster JY. The prevalence and burden of arthritis. Rheumatology. 2002;41(supp 1):3-6.
2. Kullenberg B, Runesson R, Tuvhag R, Olsson C, Resch S. Intraarticular corticosteroid injection: pain relief in osteoarthritis of the hip? J Rheumatol. 2004;31(11):2265-2268.
3. Kawasaki M, Hasegawa Y, Sakano S, Torii Y, Warashina H. Quality of life after several treatments for osteoarthritis of the hip. J Orthop Sci. 2003;8(1):32-35.
4. Ethgen O, Bruyère O, Richy F, Dardennes C, Reginster JY. Health-related quality of life in total hip and total knee arthroplasty. A qualitative and systematic review of the literature. J Bone Joint Surg Am. 2004;86(5):963-974.
5. Kurtz S, Ong K, Lau E, Mowat F, Halpern M. Projections of primary and revision hip and knee arthroplasty in the United States from 2005 to 2030. J Bone Joint Surg Am. 2007;89(4):780-785.
6. Matlock D, Earnest M, Epstein A. Utilization of elective hip and knee arthroplasty by age and payer. Clin Orthop Relat Res. 2008;466(4):914-919.
7. Parvizi J, Holiday AD, Ereth MH, Lewallen DG. The Frank Stinchfield Award. Sudden death during primary hip arthroplasty. Clin Orthop Relat Res. 1999;(369):39-48.
8. Parvizi J, Mui A, Purtill JJ, Sharkey PF, Hozack WJ, Rothman RH. Total joint arthroplasty: when do fatal or near-fatal complications occur? J Bone Joint Surg Am. 2007;89(1):27-32.
9. Parvizi J, Sullivan TA, Trousdale RT, Lewallen DG. Thirty-day mortality after total knee arthroplasty. J Bone Joint Surg Am. 2001;83(8):1157-1161.
10. Uchino S, Kellum JA, Bellomo R, et al; Beginning and Ending Supportive Therapy for the Kidney (BEST Kidney) Investigators. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA. 2005;294(7):813-818.
11. Thakar CV. Perioperative acute kidney injury. Adv Chronic Kidney Dis. 2013;20(1):67-75.
12. Hsu CY, Chertow GM, McCulloch CE, Fan D, Ordoñez JD, Go AS. Nonrecovery of kidney function and death after acute on chronic renal failure. Clin J Am Soc Nephrol. 2009;4(5):891-898.
13. Rewa O, Bagshaw SM. Acute kidney injury—epidemiology, outcomes and economics. Nat Rev Nephrol. 2014;10(4):193-207.
14. Thakar CV, Worley S, Arrigain S, Yared JP, Paganini EP. Influence of renal dysfunction on mortality after cardiac surgery: modifying effect of preoperative renal function. Kidney Int. 2005;67(3):1112-1119.
15. Zeng X, McMahon GM, Brunelli SM, Bates DW, Waikar SS. Incidence, outcomes, and comparisons across definitions of AKI in hospitalized individuals. Clin J Am Soc Nephrol. 2014;9(1):12-20.
16. Kimmel LA, Wilson S, Janardan JD, Liew SM, Walker RG. Incidence of acute kidney injury following total joint arthroplasty: a retrospective review by RIFLE criteria. Clin Kidney J. 2014;7(6):546-551.
17. Agency for Healthcare Research and Quality. Healthcare Cost and Utilization Project (HCUP) databases, 2002–2012. Rockville, MD: Agency for Healthcare Research and Quality.
18. Bjorgul K, Novicoff WM, Saleh KJ. Evaluating comorbidities in total hip and knee arthroplasty: available instruments. J Orthop Traumatol. 2010;11(4):203-209.
19. Voskuijl T, Hageman M, Ring D. Higher Charlson Comorbidity Index Scores are associated with readmission after orthopaedic surgery. Clin Orthop Relat Res. 2014;472(5):1638-1644.
20. Chertow GM, Burdick E, Honour M, Bonventre JV, Bates DW. Acute kidney injury, mortality, length of stay, and costs in hospitalized patients. J Am Soc Nephrol. 2005;16(11):3365-3370.
21. Lenihan CR, Montez-Rath ME, Mora Mangano CT, Chertow GM, Winkelmayer WC. Trends in acute kidney injury, associated use of dialysis, and mortality after cardiac surgery, 1999 to 2008. Ann Thorac Surg. 2013;95(1):20-28.
22. Belmont PJ Jr, Goodman GP, Waterman BR, Bader JO, Schoenfeld AJ. Thirty-day postoperative complications and mortality following total knee arthroplasty: incidence and risk factors among a national sample of 15,321 patients. J Bone Joint Surg Am. 2014;96(1):20-26.
23. Bennet SJ, Berry OM, Goddard J, Keating JF. Acute renal dysfunction following hip fracture. Injury. 2010;41(4):335-338.
24. Kateros K, Doulgerakis C, Galanakos SP, Sakellariou VI, Papadakis SA, Macheras GA. Analysis of kidney dysfunction in orthopaedic patients. BMC Nephrol. 2012;13:101.
25. Huang YM, Wang CM, Wang CT, Lin WP, Horng LC, Jiang CC. Perioperative celecoxib administration for pain management after total knee arthroplasty—a randomized, controlled study. BMC Musculoskelet Disord. 2008;9:77.
26. Kelley TC, Adams MJ, Mulliken BD, Dalury DF. Efficacy of multimodal perioperative analgesia protocol with periarticular medication injection in total knee arthroplasty: a randomized, double-blinded study. J Arthroplasty. 2013;28(8):1274-1277.
27. Lamplot JD, Wagner ER, Manning DW. Multimodal pain management in total knee arthroplasty: a prospective randomized controlled trial. J Arthroplasty. 2014;29(2):329-334.
28. Munin MC, Rudy TE, Glynn NW, Crossett LS, Rubash HE. Early inpatient rehabilitation after elective hip and knee arthroplasty. JAMA. 1998;279(11):847-852.
29. Pua YH, Ong PH. Association of early ambulation with length of stay and costs in total knee arthroplasty: retrospective cohort study. Am J Phys Med Rehabil. 2014;93(11):962-970.
30. Waikar SS, Wald R, Chertow GM, et al. Validity of International Classification of Diseases, Ninth Revision, Clinical Modification codes for acute renal failure. J Am Soc Nephrol. 2006;17(6):1688-1694.
31. Grams ME, Waikar SS, MacMahon B, Whelton S, Ballew SH, Coresh J. Performance and limitations of administrative data in the identification of AKI. Clin J Am Soc Nephrol. 2014;9(4):682-689.
1. Reginster JY. The prevalence and burden of arthritis. Rheumatology. 2002;41(supp 1):3-6.
2. Kullenberg B, Runesson R, Tuvhag R, Olsson C, Resch S. Intraarticular corticosteroid injection: pain relief in osteoarthritis of the hip? J Rheumatol. 2004;31(11):2265-2268.
3. Kawasaki M, Hasegawa Y, Sakano S, Torii Y, Warashina H. Quality of life after several treatments for osteoarthritis of the hip. J Orthop Sci. 2003;8(1):32-35.
4. Ethgen O, Bruyère O, Richy F, Dardennes C, Reginster JY. Health-related quality of life in total hip and total knee arthroplasty. A qualitative and systematic review of the literature. J Bone Joint Surg Am. 2004;86(5):963-974.
5. Kurtz S, Ong K, Lau E, Mowat F, Halpern M. Projections of primary and revision hip and knee arthroplasty in the United States from 2005 to 2030. J Bone Joint Surg Am. 2007;89(4):780-785.
6. Matlock D, Earnest M, Epstein A. Utilization of elective hip and knee arthroplasty by age and payer. Clin Orthop Relat Res. 2008;466(4):914-919.
7. Parvizi J, Holiday AD, Ereth MH, Lewallen DG. The Frank Stinchfield Award. Sudden death during primary hip arthroplasty. Clin Orthop Relat Res. 1999;(369):39-48.
8. Parvizi J, Mui A, Purtill JJ, Sharkey PF, Hozack WJ, Rothman RH. Total joint arthroplasty: when do fatal or near-fatal complications occur? J Bone Joint Surg Am. 2007;89(1):27-32.
9. Parvizi J, Sullivan TA, Trousdale RT, Lewallen DG. Thirty-day mortality after total knee arthroplasty. J Bone Joint Surg Am. 2001;83(8):1157-1161.
10. Uchino S, Kellum JA, Bellomo R, et al; Beginning and Ending Supportive Therapy for the Kidney (BEST Kidney) Investigators. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA. 2005;294(7):813-818.
11. Thakar CV. Perioperative acute kidney injury. Adv Chronic Kidney Dis. 2013;20(1):67-75.
12. Hsu CY, Chertow GM, McCulloch CE, Fan D, Ordoñez JD, Go AS. Nonrecovery of kidney function and death after acute on chronic renal failure. Clin J Am Soc Nephrol. 2009;4(5):891-898.
13. Rewa O, Bagshaw SM. Acute kidney injury—epidemiology, outcomes and economics. Nat Rev Nephrol. 2014;10(4):193-207.
14. Thakar CV, Worley S, Arrigain S, Yared JP, Paganini EP. Influence of renal dysfunction on mortality after cardiac surgery: modifying effect of preoperative renal function. Kidney Int. 2005;67(3):1112-1119.
15. Zeng X, McMahon GM, Brunelli SM, Bates DW, Waikar SS. Incidence, outcomes, and comparisons across definitions of AKI in hospitalized individuals. Clin J Am Soc Nephrol. 2014;9(1):12-20.
16. Kimmel LA, Wilson S, Janardan JD, Liew SM, Walker RG. Incidence of acute kidney injury following total joint arthroplasty: a retrospective review by RIFLE criteria. Clin Kidney J. 2014;7(6):546-551.
17. Agency for Healthcare Research and Quality. Healthcare Cost and Utilization Project (HCUP) databases, 2002–2012. Rockville, MD: Agency for Healthcare Research and Quality.
18. Bjorgul K, Novicoff WM, Saleh KJ. Evaluating comorbidities in total hip and knee arthroplasty: available instruments. J Orthop Traumatol. 2010;11(4):203-209.
19. Voskuijl T, Hageman M, Ring D. Higher Charlson Comorbidity Index Scores are associated with readmission after orthopaedic surgery. Clin Orthop Relat Res. 2014;472(5):1638-1644.
20. Chertow GM, Burdick E, Honour M, Bonventre JV, Bates DW. Acute kidney injury, mortality, length of stay, and costs in hospitalized patients. J Am Soc Nephrol. 2005;16(11):3365-3370.
21. Lenihan CR, Montez-Rath ME, Mora Mangano CT, Chertow GM, Winkelmayer WC. Trends in acute kidney injury, associated use of dialysis, and mortality after cardiac surgery, 1999 to 2008. Ann Thorac Surg. 2013;95(1):20-28.
22. Belmont PJ Jr, Goodman GP, Waterman BR, Bader JO, Schoenfeld AJ. Thirty-day postoperative complications and mortality following total knee arthroplasty: incidence and risk factors among a national sample of 15,321 patients. J Bone Joint Surg Am. 2014;96(1):20-26.
23. Bennet SJ, Berry OM, Goddard J, Keating JF. Acute renal dysfunction following hip fracture. Injury. 2010;41(4):335-338.
24. Kateros K, Doulgerakis C, Galanakos SP, Sakellariou VI, Papadakis SA, Macheras GA. Analysis of kidney dysfunction in orthopaedic patients. BMC Nephrol. 2012;13:101.
25. Huang YM, Wang CM, Wang CT, Lin WP, Horng LC, Jiang CC. Perioperative celecoxib administration for pain management after total knee arthroplasty—a randomized, controlled study. BMC Musculoskelet Disord. 2008;9:77.
26. Kelley TC, Adams MJ, Mulliken BD, Dalury DF. Efficacy of multimodal perioperative analgesia protocol with periarticular medication injection in total knee arthroplasty: a randomized, double-blinded study. J Arthroplasty. 2013;28(8):1274-1277.
27. Lamplot JD, Wagner ER, Manning DW. Multimodal pain management in total knee arthroplasty: a prospective randomized controlled trial. J Arthroplasty. 2014;29(2):329-334.
28. Munin MC, Rudy TE, Glynn NW, Crossett LS, Rubash HE. Early inpatient rehabilitation after elective hip and knee arthroplasty. JAMA. 1998;279(11):847-852.
29. Pua YH, Ong PH. Association of early ambulation with length of stay and costs in total knee arthroplasty: retrospective cohort study. Am J Phys Med Rehabil. 2014;93(11):962-970.
30. Waikar SS, Wald R, Chertow GM, et al. Validity of International Classification of Diseases, Ninth Revision, Clinical Modification codes for acute renal failure. J Am Soc Nephrol. 2006;17(6):1688-1694.
31. Grams ME, Waikar SS, MacMahon B, Whelton S, Ballew SH, Coresh J. Performance and limitations of administrative data in the identification of AKI. Clin J Am Soc Nephrol. 2014;9(4):682-689.
Analysis of Direct Costs of Outpatient Arthroscopic Rotator Cuff Repair
Musculoskeletal disorders, the leading cause of disability in the United States,1 account for more than half of all persons reporting missing a workday because of a medical condition.2 Shoulder disorders in particular play a significant role in the burden of musculoskeletal disorders and cost of care. In 2008, 18.9 million adults (8.2% of the US adult population) reported chronic shoulder pain.1 Among shoulder disorders, rotator cuff pathology is the most common cause of shoulder-related disability found by orthopedic surgeons.3 Rotator cuff surgery (RCS) is one of the most commonly performed orthopedic surgical procedures, and surgery volume is on the rise. One study found a 141% increase in rotator cuff repairs between the years 1996 (~41 per 100,000 population) and 2006 (~98 per 100,000 population).4
US health care costs are also increasing. In 2011, $2.7 trillion was spent on health care, representing 17.9% of the national gross domestic product (GDP). According to projections, costs will rise to $4.6 trillion by 2020.5 In particular, as patients continue to live longer and remain more active into their later years, the costs of treating and managing musculoskeletal disorders become more important from a public policy standpoint. In 2006, the cost of treating musculoskeletal disorders alone was $576 billion, representing 4.5% of that year’s GDP.2
Paramount in this era of rising costs is the idea of maximizing the value of health care dollars. Health care economists Porter and Teisberg6 defined value as patient health outcomes achieved per dollar of cost expended in a care cycle (diagnosis, treatment, ongoing management) for a particular disease or disorder. For proper management of value, outcomes and costs for an entire cycle of care must be determined. From a practical standpoint, this first requires determining the true cost of a care cycle—dollars spent on personnel, equipment, materials, and other resources required to deliver a particular service—rather than the amount charged or reimbursed for providing the service in question.7
Kaplan and Anderson8,9 described the TDABC (time-driven activity-based costing) algorithm for calculating the cost of delivering a service based on 2 parameters: unit cost of a particular resource, and time required to supply it. These parameters apply to material costs and labor costs. In the medical setting, the TDABC algorithm can be applied by defining a care delivery value chain for each aspect of patient care and then multiplying incremental cost per unit time by time required to deliver that resource (Figure 1). Tabulating the overall unit cost for each resource then yields the overall cost of the care cycle. Clinical outcomes data can then be determined and used to calculate overall value for the patient care cycle.
In the study reported here, we used the TDABC algorithm to calculate the direct financial costs of surgical treatment of rotator cuff tears confirmed by magnetic resonance imaging (MRI) in an academic medical center.
Methods
Per our institution’s Office for the Protection of Research Subjects, institutional review board (IRB) approval is required only for projects using “human subjects” as defined by federal policy. In the present study, no private information could be identified, and all data were obtained from hospital billing records without intervention or interaction with individual patients. Accordingly, IRB approval was deemed unnecessary for our economic cost analysis.
Billing records of a single academic fellowship-trained sports surgeon were reviewed to identify patients who underwent primary repair of an MRI-confirmed rotator cuff tear between April 1, 2009, and July 31, 2012. Patients who had undergone prior shoulder surgery of any type were excluded from the study. Operative reports were reviewed, and exact surgical procedures performed were noted. The operating surgeon selected the specific repair techniques, including single- or double-row repair, with emphasis on restoring footprint coverage and avoiding overtensioning.
All surgeries were performed in an outpatient surgical center owned and operated by the surgeon’s home university. Surgeries were performed by the attending physician assisted by a senior orthopedic resident. The RCS care cycle was divided into 3 phases (Figure 2):
1. Preoperative. Patient’s interaction with receptionist in surgery center, time with preoperative nurse and circulating nurse in preoperative area, resident check-in time, and time placing preoperative nerve block and consumable materials used during block placement.
2. Operative. Time in operating room with surgical team for RCS, consumable materials used during surgery (eg, anchors, shavers, drapes), anesthetic medications, shoulder abduction pillow placed on completion of surgery, and cost of instrument processing.
3. Postoperative. Time in postoperative recovery area with recovery room nursing staff.
Time in each portion of the care cycle was directly observed and tabulated by hospital volunteers in the surgery center. Institutional billing data were used to identify material resources consumed, and the actual cost paid by the hospital for these resources was obtained from internal records. Mean hourly salary data and standard benefit rates were obtained for surgery center staff. Attending physician salary was extrapolated from published mean market salary data for academic physicians and mean hours worked,10,11 and resident physician costs were tabulated from publically available institutional payroll data and average resident work hours at our institution. These cost data and times were then used to tabulate total cost for the RCS care cycle using the TDABC algorithm.
Results
We identified 28 shoulders in 26 patients (mean age, 54.5 years) who met the inclusion criteria. Of these 28 shoulders, 18 (64.3%) had an isolated supraspinatus tear, 8 (28.6%) had combined supraspinatus and infraspinatus tears, 1 (3.6%) had combined supraspinatus and subscapularis tears, and 1 (3.6%) had an isolated infraspinatus tear. Demographic data are listed in Table 1.
All patients received an interscalene nerve block in the preoperative area before being brought into the operating room. In our analysis, we included nerve block supply costs and the anesthesiologist’s mean time placing the nerve block.
In all cases, primary rotator cuff repair was performed with suture anchors (Parcus Medical) with the patient in the lateral decubitus position. In 13 (46%) of the 28 shoulders, this repair was described as “complex,” requiring double-row technique. Subacromial decompression and bursectomy were performed in addition to the rotator cuff repair. Labral débridement was performed in 23 patients, synovectomy in 10, biceps tenodesis with anchor (Smith & Nephew) in 1, and biceps tenotomy in 1. Mean time in operating room was 148 minutes; mean time in postoperative recovery unit was 105 minutes.
Directly observing the care cycle, hospital volunteers found that patients spent a mean of 15 minutes with the receptionist when they arrived in the outpatient surgical center, 25 minutes with nurses for check-in in the preoperative holding area, and 10 minutes with the anesthesiology resident and 15 minutes with the orthopedic surgery resident for preoperative evaluation and paperwork. Mean nerve block time was 20 minutes. Mean electrocardiogram (ECG) time (12 patients) was 15 minutes. The surgical technician spent a mean time of 20 minutes setting up the operating room before the patient was brought in and 15 minutes cleaning up after the patient was transferred to the recovery room. Costs of postoperative care in the recovery room were based on a 2:1 patient-to-nurse ratio, as is the standard practice in our outpatient surgery center.
Using the times mentioned and our hospital’s salary data—including standard hospital benefits rates of 33.5% for nonphysicians and 17.65% for physicians—we determined, using the TDABC algorithm, a direct cost of $5904.21 for this process cycle, excluding hospital overhead and indirect costs. Table 2 provides the overall cost breakdown. Compared with the direct economic cost, the mean hospital charge to insurers for the procedure was $31,459.35. Mean reimbursement from insurers was $9679.08.
Overall attending and resident physician costs were $1077.75, which consisted of $623.66 for the surgeon and $454.09 for the anesthesiologist (included placement of nerve block and administration of anesthesia during surgery). Preoperative bloodwork was obtained in 23 cases, adding a mean cost of $111.04 after adjusting for standard hospital markup. Preoperative ECG was performed in 12 cases, for an added mean cost of $7.30 based on the TDABC algorithm.
We also broke down costs by care cycle phase. The preoperative phase, excluding the preoperative laboratory studies and ECGs (not performed in all cases), cost $134.34 (2.3% of total costs); the operative phase cost $5718.01 (96.8% of total costs); and the postoperative phase cost $51.86 (0.9% of total costs). Within the operative phase, the cost of consumables (specifically, suture anchors) was the main cost driver. Mean anchor cost per case was $3432.67. “Complex” tears involving a double-row repair averaged $4570.25 in anchor cost per patient, as compared with $2522.60 in anchor costs for simple repairs.
Discussion
US health care costs continue to increase unsustainably, with rising pressure on hospitals and providers to deliver the highest value for each health care dollar. The present study is the first to calculate (using the TDABC algorithm) the direct economic cost ($5904.21) of the entire RCS care cycle at a university-based outpatient surgery center. Rent, utility costs, administrative costs, overhead, and other indirect costs at the surgery center were not included in this cost analysis, as they would be incurred irrespective of type of surgery performed. As such, our data isolate the procedure-specific costs of rotator cuff repair in order to provide a more meaningful comparison for other institutions, where indirect costs may be different.
In the literature, rigorous economic analysis of shoulder pathology is sparse. Kuye and colleagues12 systematically reviewed economic evaluations in shoulder surgery for the period 1980–2010 and noted more than 50% of the papers were published between 2005 and 2010.12 They also noted the poor quality of these studies and concluded more rigorous economic evaluations are needed to help justify the rising costs of shoulder-related treatments.
Several studies have directly evaluated costs associated with RCS. Cordasco and colleagues13 detailed the success of open rotator cuff repair as an outpatient procedure—noting its 43% cost savings ($4300 for outpatient vs $7500 for inpatient) and high patient satisfaction—using hospital charge data for operating room time, supplies, instruments, and postoperative slings. Churchill and Ghorai14 evaluated costs of mini-open and arthroscopic rotator cuff repairs in a statewide database and estimated the arthroscopic repair cost at $8985, compared with $7841 for the mini-open repair. They used reported hospital charge data, which were not itemized and did not include physician professional fees. Adla and colleagues,15 in a similar analysis of open versus arthroscopic cuff repair, estimated direct material costs of $1609.50 (arthroscopic) and $360.75 (open); these figures were converted from 2005 UK currency using the exchange rate cited in their paper. Salaries of surgeon, anesthesiologist, and other operating room personnel were said to be included in the operating room cost, but the authors’ paper did not include these data.
Two studies directly estimated the costs of arthroscopic rotator cuff repair. Hearnden and Tennent16 calculated the cost of RCS at their UK institution to be £2672, which included cost of operating room consumable materials, medication, and salaries of operating room personnel, including surgeon and anesthesiologist. Using online currency conversion from 2008 exchange rates and adjusting for inflation gave a corresponding US cost of $5449.63.17 Vitale and colleagues18 prospectively calculated costs of arthroscopic rotator cuff repair over a 1-year period using a cost-to-charge ratio from tabulated inpatient charges, procedure charges, and physician fees and payments abstracted from medical records, hospital billing, and administrative databases. Mean total cost for this cycle was $10,605.20, which included several costs (physical therapy, radiologist fees) not included in the present study. These studies, though more comprehensive than prior work, did not capture the entire cycle of surgical care.
Our study was designed to provide initial data on the direct costs of arthroscopic repair of the rotator cuff for the entire process cycle. Our overall cost estimate of $5904.21 differs significantly from prior work—not unexpected given the completely different cost methodology used.
Our study had several limitations. First, it was a single-surgeon evaluation, and a number of operating room variables (eg, use of adjunct instrumentation such as radiofrequency probes, differences in draping preferences) as well as surgeon volume in performing rotator cuff repairs might have substantially affected the reproducibility and generalizability of our data. Similarly, the large number of adjunctive procedures (eg, subacromial decompression, labral débridement) performed in conjunction with the rotator cuff repairs added operative time and therefore increased overall cost. Double-row repairs added operative time and increased the cost of consumable materials as well. Differences in surgeon preference for suture anchors may also be important, as anchors are a major cost driver and can vary significantly between vendors and institutions. Tear-related variables (eg, tear size, tear chronicity, degree of fatty cuff degeneration) were not controlled for and might have significantly affected operative time and associated cost. Resident involvement in the surgical procedure and anesthesia process in an academic setting prolongs surgical time and thus directly impacts costs.
In addition, we used the patient’s time in the operating room as a proxy for actual surgical time, as this was the only reliable and reproducible data point available in our electronic medical record. As such, an unquantifiable amount of surgeon time may have been overallocated to our cost estimate for time spent inducing anesthesia, positioning, helping take the patient off the operating table, and so on. However, as typical surgeon practice is to be involved in these tasks in the operating room, the possible overestimate of surgeon cost is likely minimal.
Our salary data for the TDABC algorithm were based on national averages for work hours and gross income for physicians and on hospital-based wage structure and may not be generalizable to other institutions. There may also be regional differences in work hours and salaries, which in turn would factor into a different per-minute cost for surgeon and anesthesiologist, depending on the exact geographic area where the surgery is performed. Costs may be higher at institutions that use certified nurse anesthetists rather than resident physicians because of the salary differences between these practitioners.
Moreover, the time that patients spend in the holding area—waiting to go into surgery and, after surgery, waiting for their ride home, for their prescriptions to be ready, and so forth—is an important variable to consider from a cost standpoint. However, as this time varied significantly and involved minimal contact with hospital personnel, we excluded its associated costs from our analysis. Similarly, and as already noted, hospital overhead and other indirect costs were excluded from analysis as well.
Conclusion
Using the TDABC algorithm, we found a direct economic cost of $5904.21 for RCS at our academic outpatient surgical center, with anchor cost the main cost driver. Judicious use of consumable resources is a key focus for cost containment in arthroscopic shoulder surgery, particularly with respect to implantable suture anchors. However, in the setting of more complex tears that require multiple anchors in a double-row repair construct, our pilot data may be useful to hospitals and surgery centers negotiating procedural reimbursement for the increased cost of complex repairs. Use of the TDABC algorithm for RCS and other procedures may also help in identifying opportunities to deliver more cost-effective health care.
1. American Academy of Orthopaedic Surgeons. The Burden of Musculoskeletal Diseases in the United States: Prevalence, Societal and Economic Cost. Rosemont, IL: American Academy of Orthopaedic Surgeons; 2011.
2. National health expenditure data. Centers for Medicare & Medicare Services website. https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/NationalHealthExpendData/index.html. Updated May 5, 2014. Accessed December 1, 2015.
3. Tashjian RZ. Epidemiology, natural history, and indications for treatment of rotator cuff tears. Clin Sports Med. 2012;31(4):589-604.
4. Colvin AC, Egorova N, Harrison AK, Moskowitz A, Flatow EL. National trends in rotator cuff repair. J Bone Joint Surg Am. 2012;94(3):227-233.
5. Black EM, Higgins LD, Warner JJ. Value-based shoulder surgery: practicing outcomes-driven, cost-conscious care. J Shoulder Elbow Surg. 2013;22(7):1000-1009.
6. Porter ME, Teisberg EO. Redefining Health Care: Creating Value-Based Competition on Results. Boston, MA: Harvard Business School Press; 2006.
7. Kaplan RS, Porter ME. How to solve the cost crisis in health care. Harv Bus Rev. 2011;89(9):46-52, 54, 56-61 passim.
8. Kaplan RS, Anderson SR. Time-driven activity-based costing. Harv Bus Rev. 2004;82(11):131-138, 150.
9. Kaplan RS, Anderson SR. Time-Driven Activity-Based Costing: A Simpler and More Powerful Path to Higher Profits. Boston, MA: Harvard Business Review Press; 2007.
10. American Academy of Orthopaedic Surgeons. Orthopaedic Practice in the U.S. 2012. Rosemont, IL: American Academy of Orthopaedic Surgeons; 2012.
11. Medical Group Management Association. Physician Compensation and Production Survey: 2012 Report Based on 2011 Data. Englewood, CO: Medical Group Management Association; 2012.
12. Kuye IO, Jain NB, Warner L, Herndon JH, Warner JJ. Economic evaluations in shoulder pathologies: a systematic review of the literature. J Shoulder Elbow Surg. 2012;21(3):367-375.
13. Cordasco FA, McGinley BJ, Charlton T. Rotator cuff repair as an outpatient procedure. J Shoulder Elbow Surg. 2000;9(1):27-30.
14. Churchill RS, Ghorai JK. Total cost and operating room time comparison of rotator cuff repair techniques at low, intermediate, and high volume centers: mini-open versus all-arthroscopic. J Shoulder Elbow Surg. 2010;19(5):716-721.
15. Adla DN, Rowsell M, Pandey R. Cost-effectiveness of open versus arthroscopic rotator cuff repair. J Shoulder Elbow Surg. 2010;19(2):258-261.
16. Hearnden A, Tennent D. The cost of shoulder arthroscopy: a comparison with national tariff. Ann R Coll Surg Engl. 2008;90(7):587-591.
17. Xrates currency conversion. http://www.x-rates.com/historical/?from=GBP&amount=1&date=2015-12-03. Accessed December 13, 2015.
18. Vitale MA, Vitale MG, Zivin JG, Braman JP, Bigliani LU, Flatow EL. Rotator cuff repair: an analysis of utility scores and cost-effectiveness. J Shoulder Elbow Surg. 2007;16(2):181-187.
Musculoskeletal disorders, the leading cause of disability in the United States,1 account for more than half of all persons reporting missing a workday because of a medical condition.2 Shoulder disorders in particular play a significant role in the burden of musculoskeletal disorders and cost of care. In 2008, 18.9 million adults (8.2% of the US adult population) reported chronic shoulder pain.1 Among shoulder disorders, rotator cuff pathology is the most common cause of shoulder-related disability found by orthopedic surgeons.3 Rotator cuff surgery (RCS) is one of the most commonly performed orthopedic surgical procedures, and surgery volume is on the rise. One study found a 141% increase in rotator cuff repairs between the years 1996 (~41 per 100,000 population) and 2006 (~98 per 100,000 population).4
US health care costs are also increasing. In 2011, $2.7 trillion was spent on health care, representing 17.9% of the national gross domestic product (GDP). According to projections, costs will rise to $4.6 trillion by 2020.5 In particular, as patients continue to live longer and remain more active into their later years, the costs of treating and managing musculoskeletal disorders become more important from a public policy standpoint. In 2006, the cost of treating musculoskeletal disorders alone was $576 billion, representing 4.5% of that year’s GDP.2
Paramount in this era of rising costs is the idea of maximizing the value of health care dollars. Health care economists Porter and Teisberg6 defined value as patient health outcomes achieved per dollar of cost expended in a care cycle (diagnosis, treatment, ongoing management) for a particular disease or disorder. For proper management of value, outcomes and costs for an entire cycle of care must be determined. From a practical standpoint, this first requires determining the true cost of a care cycle—dollars spent on personnel, equipment, materials, and other resources required to deliver a particular service—rather than the amount charged or reimbursed for providing the service in question.7
Kaplan and Anderson8,9 described the TDABC (time-driven activity-based costing) algorithm for calculating the cost of delivering a service based on 2 parameters: unit cost of a particular resource, and time required to supply it. These parameters apply to material costs and labor costs. In the medical setting, the TDABC algorithm can be applied by defining a care delivery value chain for each aspect of patient care and then multiplying incremental cost per unit time by time required to deliver that resource (Figure 1). Tabulating the overall unit cost for each resource then yields the overall cost of the care cycle. Clinical outcomes data can then be determined and used to calculate overall value for the patient care cycle.
In the study reported here, we used the TDABC algorithm to calculate the direct financial costs of surgical treatment of rotator cuff tears confirmed by magnetic resonance imaging (MRI) in an academic medical center.
Methods
Per our institution’s Office for the Protection of Research Subjects, institutional review board (IRB) approval is required only for projects using “human subjects” as defined by federal policy. In the present study, no private information could be identified, and all data were obtained from hospital billing records without intervention or interaction with individual patients. Accordingly, IRB approval was deemed unnecessary for our economic cost analysis.
Billing records of a single academic fellowship-trained sports surgeon were reviewed to identify patients who underwent primary repair of an MRI-confirmed rotator cuff tear between April 1, 2009, and July 31, 2012. Patients who had undergone prior shoulder surgery of any type were excluded from the study. Operative reports were reviewed, and exact surgical procedures performed were noted. The operating surgeon selected the specific repair techniques, including single- or double-row repair, with emphasis on restoring footprint coverage and avoiding overtensioning.
All surgeries were performed in an outpatient surgical center owned and operated by the surgeon’s home university. Surgeries were performed by the attending physician assisted by a senior orthopedic resident. The RCS care cycle was divided into 3 phases (Figure 2):
1. Preoperative. Patient’s interaction with receptionist in surgery center, time with preoperative nurse and circulating nurse in preoperative area, resident check-in time, and time placing preoperative nerve block and consumable materials used during block placement.
2. Operative. Time in operating room with surgical team for RCS, consumable materials used during surgery (eg, anchors, shavers, drapes), anesthetic medications, shoulder abduction pillow placed on completion of surgery, and cost of instrument processing.
3. Postoperative. Time in postoperative recovery area with recovery room nursing staff.
Time in each portion of the care cycle was directly observed and tabulated by hospital volunteers in the surgery center. Institutional billing data were used to identify material resources consumed, and the actual cost paid by the hospital for these resources was obtained from internal records. Mean hourly salary data and standard benefit rates were obtained for surgery center staff. Attending physician salary was extrapolated from published mean market salary data for academic physicians and mean hours worked,10,11 and resident physician costs were tabulated from publically available institutional payroll data and average resident work hours at our institution. These cost data and times were then used to tabulate total cost for the RCS care cycle using the TDABC algorithm.
Results
We identified 28 shoulders in 26 patients (mean age, 54.5 years) who met the inclusion criteria. Of these 28 shoulders, 18 (64.3%) had an isolated supraspinatus tear, 8 (28.6%) had combined supraspinatus and infraspinatus tears, 1 (3.6%) had combined supraspinatus and subscapularis tears, and 1 (3.6%) had an isolated infraspinatus tear. Demographic data are listed in Table 1.
All patients received an interscalene nerve block in the preoperative area before being brought into the operating room. In our analysis, we included nerve block supply costs and the anesthesiologist’s mean time placing the nerve block.
In all cases, primary rotator cuff repair was performed with suture anchors (Parcus Medical) with the patient in the lateral decubitus position. In 13 (46%) of the 28 shoulders, this repair was described as “complex,” requiring double-row technique. Subacromial decompression and bursectomy were performed in addition to the rotator cuff repair. Labral débridement was performed in 23 patients, synovectomy in 10, biceps tenodesis with anchor (Smith & Nephew) in 1, and biceps tenotomy in 1. Mean time in operating room was 148 minutes; mean time in postoperative recovery unit was 105 minutes.
Directly observing the care cycle, hospital volunteers found that patients spent a mean of 15 minutes with the receptionist when they arrived in the outpatient surgical center, 25 minutes with nurses for check-in in the preoperative holding area, and 10 minutes with the anesthesiology resident and 15 minutes with the orthopedic surgery resident for preoperative evaluation and paperwork. Mean nerve block time was 20 minutes. Mean electrocardiogram (ECG) time (12 patients) was 15 minutes. The surgical technician spent a mean time of 20 minutes setting up the operating room before the patient was brought in and 15 minutes cleaning up after the patient was transferred to the recovery room. Costs of postoperative care in the recovery room were based on a 2:1 patient-to-nurse ratio, as is the standard practice in our outpatient surgery center.
Using the times mentioned and our hospital’s salary data—including standard hospital benefits rates of 33.5% for nonphysicians and 17.65% for physicians—we determined, using the TDABC algorithm, a direct cost of $5904.21 for this process cycle, excluding hospital overhead and indirect costs. Table 2 provides the overall cost breakdown. Compared with the direct economic cost, the mean hospital charge to insurers for the procedure was $31,459.35. Mean reimbursement from insurers was $9679.08.
Overall attending and resident physician costs were $1077.75, which consisted of $623.66 for the surgeon and $454.09 for the anesthesiologist (included placement of nerve block and administration of anesthesia during surgery). Preoperative bloodwork was obtained in 23 cases, adding a mean cost of $111.04 after adjusting for standard hospital markup. Preoperative ECG was performed in 12 cases, for an added mean cost of $7.30 based on the TDABC algorithm.
We also broke down costs by care cycle phase. The preoperative phase, excluding the preoperative laboratory studies and ECGs (not performed in all cases), cost $134.34 (2.3% of total costs); the operative phase cost $5718.01 (96.8% of total costs); and the postoperative phase cost $51.86 (0.9% of total costs). Within the operative phase, the cost of consumables (specifically, suture anchors) was the main cost driver. Mean anchor cost per case was $3432.67. “Complex” tears involving a double-row repair averaged $4570.25 in anchor cost per patient, as compared with $2522.60 in anchor costs for simple repairs.
Discussion
US health care costs continue to increase unsustainably, with rising pressure on hospitals and providers to deliver the highest value for each health care dollar. The present study is the first to calculate (using the TDABC algorithm) the direct economic cost ($5904.21) of the entire RCS care cycle at a university-based outpatient surgery center. Rent, utility costs, administrative costs, overhead, and other indirect costs at the surgery center were not included in this cost analysis, as they would be incurred irrespective of type of surgery performed. As such, our data isolate the procedure-specific costs of rotator cuff repair in order to provide a more meaningful comparison for other institutions, where indirect costs may be different.
In the literature, rigorous economic analysis of shoulder pathology is sparse. Kuye and colleagues12 systematically reviewed economic evaluations in shoulder surgery for the period 1980–2010 and noted more than 50% of the papers were published between 2005 and 2010.12 They also noted the poor quality of these studies and concluded more rigorous economic evaluations are needed to help justify the rising costs of shoulder-related treatments.
Several studies have directly evaluated costs associated with RCS. Cordasco and colleagues13 detailed the success of open rotator cuff repair as an outpatient procedure—noting its 43% cost savings ($4300 for outpatient vs $7500 for inpatient) and high patient satisfaction—using hospital charge data for operating room time, supplies, instruments, and postoperative slings. Churchill and Ghorai14 evaluated costs of mini-open and arthroscopic rotator cuff repairs in a statewide database and estimated the arthroscopic repair cost at $8985, compared with $7841 for the mini-open repair. They used reported hospital charge data, which were not itemized and did not include physician professional fees. Adla and colleagues,15 in a similar analysis of open versus arthroscopic cuff repair, estimated direct material costs of $1609.50 (arthroscopic) and $360.75 (open); these figures were converted from 2005 UK currency using the exchange rate cited in their paper. Salaries of surgeon, anesthesiologist, and other operating room personnel were said to be included in the operating room cost, but the authors’ paper did not include these data.
Two studies directly estimated the costs of arthroscopic rotator cuff repair. Hearnden and Tennent16 calculated the cost of RCS at their UK institution to be £2672, which included cost of operating room consumable materials, medication, and salaries of operating room personnel, including surgeon and anesthesiologist. Using online currency conversion from 2008 exchange rates and adjusting for inflation gave a corresponding US cost of $5449.63.17 Vitale and colleagues18 prospectively calculated costs of arthroscopic rotator cuff repair over a 1-year period using a cost-to-charge ratio from tabulated inpatient charges, procedure charges, and physician fees and payments abstracted from medical records, hospital billing, and administrative databases. Mean total cost for this cycle was $10,605.20, which included several costs (physical therapy, radiologist fees) not included in the present study. These studies, though more comprehensive than prior work, did not capture the entire cycle of surgical care.
Our study was designed to provide initial data on the direct costs of arthroscopic repair of the rotator cuff for the entire process cycle. Our overall cost estimate of $5904.21 differs significantly from prior work—not unexpected given the completely different cost methodology used.
Our study had several limitations. First, it was a single-surgeon evaluation, and a number of operating room variables (eg, use of adjunct instrumentation such as radiofrequency probes, differences in draping preferences) as well as surgeon volume in performing rotator cuff repairs might have substantially affected the reproducibility and generalizability of our data. Similarly, the large number of adjunctive procedures (eg, subacromial decompression, labral débridement) performed in conjunction with the rotator cuff repairs added operative time and therefore increased overall cost. Double-row repairs added operative time and increased the cost of consumable materials as well. Differences in surgeon preference for suture anchors may also be important, as anchors are a major cost driver and can vary significantly between vendors and institutions. Tear-related variables (eg, tear size, tear chronicity, degree of fatty cuff degeneration) were not controlled for and might have significantly affected operative time and associated cost. Resident involvement in the surgical procedure and anesthesia process in an academic setting prolongs surgical time and thus directly impacts costs.
In addition, we used the patient’s time in the operating room as a proxy for actual surgical time, as this was the only reliable and reproducible data point available in our electronic medical record. As such, an unquantifiable amount of surgeon time may have been overallocated to our cost estimate for time spent inducing anesthesia, positioning, helping take the patient off the operating table, and so on. However, as typical surgeon practice is to be involved in these tasks in the operating room, the possible overestimate of surgeon cost is likely minimal.
Our salary data for the TDABC algorithm were based on national averages for work hours and gross income for physicians and on hospital-based wage structure and may not be generalizable to other institutions. There may also be regional differences in work hours and salaries, which in turn would factor into a different per-minute cost for surgeon and anesthesiologist, depending on the exact geographic area where the surgery is performed. Costs may be higher at institutions that use certified nurse anesthetists rather than resident physicians because of the salary differences between these practitioners.
Moreover, the time that patients spend in the holding area—waiting to go into surgery and, after surgery, waiting for their ride home, for their prescriptions to be ready, and so forth—is an important variable to consider from a cost standpoint. However, as this time varied significantly and involved minimal contact with hospital personnel, we excluded its associated costs from our analysis. Similarly, and as already noted, hospital overhead and other indirect costs were excluded from analysis as well.
Conclusion
Using the TDABC algorithm, we found a direct economic cost of $5904.21 for RCS at our academic outpatient surgical center, with anchor cost the main cost driver. Judicious use of consumable resources is a key focus for cost containment in arthroscopic shoulder surgery, particularly with respect to implantable suture anchors. However, in the setting of more complex tears that require multiple anchors in a double-row repair construct, our pilot data may be useful to hospitals and surgery centers negotiating procedural reimbursement for the increased cost of complex repairs. Use of the TDABC algorithm for RCS and other procedures may also help in identifying opportunities to deliver more cost-effective health care.
Musculoskeletal disorders, the leading cause of disability in the United States,1 account for more than half of all persons reporting missing a workday because of a medical condition.2 Shoulder disorders in particular play a significant role in the burden of musculoskeletal disorders and cost of care. In 2008, 18.9 million adults (8.2% of the US adult population) reported chronic shoulder pain.1 Among shoulder disorders, rotator cuff pathology is the most common cause of shoulder-related disability found by orthopedic surgeons.3 Rotator cuff surgery (RCS) is one of the most commonly performed orthopedic surgical procedures, and surgery volume is on the rise. One study found a 141% increase in rotator cuff repairs between the years 1996 (~41 per 100,000 population) and 2006 (~98 per 100,000 population).4
US health care costs are also increasing. In 2011, $2.7 trillion was spent on health care, representing 17.9% of the national gross domestic product (GDP). According to projections, costs will rise to $4.6 trillion by 2020.5 In particular, as patients continue to live longer and remain more active into their later years, the costs of treating and managing musculoskeletal disorders become more important from a public policy standpoint. In 2006, the cost of treating musculoskeletal disorders alone was $576 billion, representing 4.5% of that year’s GDP.2
Paramount in this era of rising costs is the idea of maximizing the value of health care dollars. Health care economists Porter and Teisberg6 defined value as patient health outcomes achieved per dollar of cost expended in a care cycle (diagnosis, treatment, ongoing management) for a particular disease or disorder. For proper management of value, outcomes and costs for an entire cycle of care must be determined. From a practical standpoint, this first requires determining the true cost of a care cycle—dollars spent on personnel, equipment, materials, and other resources required to deliver a particular service—rather than the amount charged or reimbursed for providing the service in question.7
Kaplan and Anderson8,9 described the TDABC (time-driven activity-based costing) algorithm for calculating the cost of delivering a service based on 2 parameters: unit cost of a particular resource, and time required to supply it. These parameters apply to material costs and labor costs. In the medical setting, the TDABC algorithm can be applied by defining a care delivery value chain for each aspect of patient care and then multiplying incremental cost per unit time by time required to deliver that resource (Figure 1). Tabulating the overall unit cost for each resource then yields the overall cost of the care cycle. Clinical outcomes data can then be determined and used to calculate overall value for the patient care cycle.
In the study reported here, we used the TDABC algorithm to calculate the direct financial costs of surgical treatment of rotator cuff tears confirmed by magnetic resonance imaging (MRI) in an academic medical center.
Methods
Per our institution’s Office for the Protection of Research Subjects, institutional review board (IRB) approval is required only for projects using “human subjects” as defined by federal policy. In the present study, no private information could be identified, and all data were obtained from hospital billing records without intervention or interaction with individual patients. Accordingly, IRB approval was deemed unnecessary for our economic cost analysis.
Billing records of a single academic fellowship-trained sports surgeon were reviewed to identify patients who underwent primary repair of an MRI-confirmed rotator cuff tear between April 1, 2009, and July 31, 2012. Patients who had undergone prior shoulder surgery of any type were excluded from the study. Operative reports were reviewed, and exact surgical procedures performed were noted. The operating surgeon selected the specific repair techniques, including single- or double-row repair, with emphasis on restoring footprint coverage and avoiding overtensioning.
All surgeries were performed in an outpatient surgical center owned and operated by the surgeon’s home university. Surgeries were performed by the attending physician assisted by a senior orthopedic resident. The RCS care cycle was divided into 3 phases (Figure 2):
1. Preoperative. Patient’s interaction with receptionist in surgery center, time with preoperative nurse and circulating nurse in preoperative area, resident check-in time, and time placing preoperative nerve block and consumable materials used during block placement.
2. Operative. Time in operating room with surgical team for RCS, consumable materials used during surgery (eg, anchors, shavers, drapes), anesthetic medications, shoulder abduction pillow placed on completion of surgery, and cost of instrument processing.
3. Postoperative. Time in postoperative recovery area with recovery room nursing staff.
Time in each portion of the care cycle was directly observed and tabulated by hospital volunteers in the surgery center. Institutional billing data were used to identify material resources consumed, and the actual cost paid by the hospital for these resources was obtained from internal records. Mean hourly salary data and standard benefit rates were obtained for surgery center staff. Attending physician salary was extrapolated from published mean market salary data for academic physicians and mean hours worked,10,11 and resident physician costs were tabulated from publically available institutional payroll data and average resident work hours at our institution. These cost data and times were then used to tabulate total cost for the RCS care cycle using the TDABC algorithm.
Results
We identified 28 shoulders in 26 patients (mean age, 54.5 years) who met the inclusion criteria. Of these 28 shoulders, 18 (64.3%) had an isolated supraspinatus tear, 8 (28.6%) had combined supraspinatus and infraspinatus tears, 1 (3.6%) had combined supraspinatus and subscapularis tears, and 1 (3.6%) had an isolated infraspinatus tear. Demographic data are listed in Table 1.
All patients received an interscalene nerve block in the preoperative area before being brought into the operating room. In our analysis, we included nerve block supply costs and the anesthesiologist’s mean time placing the nerve block.
In all cases, primary rotator cuff repair was performed with suture anchors (Parcus Medical) with the patient in the lateral decubitus position. In 13 (46%) of the 28 shoulders, this repair was described as “complex,” requiring double-row technique. Subacromial decompression and bursectomy were performed in addition to the rotator cuff repair. Labral débridement was performed in 23 patients, synovectomy in 10, biceps tenodesis with anchor (Smith & Nephew) in 1, and biceps tenotomy in 1. Mean time in operating room was 148 minutes; mean time in postoperative recovery unit was 105 minutes.
Directly observing the care cycle, hospital volunteers found that patients spent a mean of 15 minutes with the receptionist when they arrived in the outpatient surgical center, 25 minutes with nurses for check-in in the preoperative holding area, and 10 minutes with the anesthesiology resident and 15 minutes with the orthopedic surgery resident for preoperative evaluation and paperwork. Mean nerve block time was 20 minutes. Mean electrocardiogram (ECG) time (12 patients) was 15 minutes. The surgical technician spent a mean time of 20 minutes setting up the operating room before the patient was brought in and 15 minutes cleaning up after the patient was transferred to the recovery room. Costs of postoperative care in the recovery room were based on a 2:1 patient-to-nurse ratio, as is the standard practice in our outpatient surgery center.
Using the times mentioned and our hospital’s salary data—including standard hospital benefits rates of 33.5% for nonphysicians and 17.65% for physicians—we determined, using the TDABC algorithm, a direct cost of $5904.21 for this process cycle, excluding hospital overhead and indirect costs. Table 2 provides the overall cost breakdown. Compared with the direct economic cost, the mean hospital charge to insurers for the procedure was $31,459.35. Mean reimbursement from insurers was $9679.08.
Overall attending and resident physician costs were $1077.75, which consisted of $623.66 for the surgeon and $454.09 for the anesthesiologist (included placement of nerve block and administration of anesthesia during surgery). Preoperative bloodwork was obtained in 23 cases, adding a mean cost of $111.04 after adjusting for standard hospital markup. Preoperative ECG was performed in 12 cases, for an added mean cost of $7.30 based on the TDABC algorithm.
We also broke down costs by care cycle phase. The preoperative phase, excluding the preoperative laboratory studies and ECGs (not performed in all cases), cost $134.34 (2.3% of total costs); the operative phase cost $5718.01 (96.8% of total costs); and the postoperative phase cost $51.86 (0.9% of total costs). Within the operative phase, the cost of consumables (specifically, suture anchors) was the main cost driver. Mean anchor cost per case was $3432.67. “Complex” tears involving a double-row repair averaged $4570.25 in anchor cost per patient, as compared with $2522.60 in anchor costs for simple repairs.
Discussion
US health care costs continue to increase unsustainably, with rising pressure on hospitals and providers to deliver the highest value for each health care dollar. The present study is the first to calculate (using the TDABC algorithm) the direct economic cost ($5904.21) of the entire RCS care cycle at a university-based outpatient surgery center. Rent, utility costs, administrative costs, overhead, and other indirect costs at the surgery center were not included in this cost analysis, as they would be incurred irrespective of type of surgery performed. As such, our data isolate the procedure-specific costs of rotator cuff repair in order to provide a more meaningful comparison for other institutions, where indirect costs may be different.
In the literature, rigorous economic analysis of shoulder pathology is sparse. Kuye and colleagues12 systematically reviewed economic evaluations in shoulder surgery for the period 1980–2010 and noted more than 50% of the papers were published between 2005 and 2010.12 They also noted the poor quality of these studies and concluded more rigorous economic evaluations are needed to help justify the rising costs of shoulder-related treatments.
Several studies have directly evaluated costs associated with RCS. Cordasco and colleagues13 detailed the success of open rotator cuff repair as an outpatient procedure—noting its 43% cost savings ($4300 for outpatient vs $7500 for inpatient) and high patient satisfaction—using hospital charge data for operating room time, supplies, instruments, and postoperative slings. Churchill and Ghorai14 evaluated costs of mini-open and arthroscopic rotator cuff repairs in a statewide database and estimated the arthroscopic repair cost at $8985, compared with $7841 for the mini-open repair. They used reported hospital charge data, which were not itemized and did not include physician professional fees. Adla and colleagues,15 in a similar analysis of open versus arthroscopic cuff repair, estimated direct material costs of $1609.50 (arthroscopic) and $360.75 (open); these figures were converted from 2005 UK currency using the exchange rate cited in their paper. Salaries of surgeon, anesthesiologist, and other operating room personnel were said to be included in the operating room cost, but the authors’ paper did not include these data.
Two studies directly estimated the costs of arthroscopic rotator cuff repair. Hearnden and Tennent16 calculated the cost of RCS at their UK institution to be £2672, which included cost of operating room consumable materials, medication, and salaries of operating room personnel, including surgeon and anesthesiologist. Using online currency conversion from 2008 exchange rates and adjusting for inflation gave a corresponding US cost of $5449.63.17 Vitale and colleagues18 prospectively calculated costs of arthroscopic rotator cuff repair over a 1-year period using a cost-to-charge ratio from tabulated inpatient charges, procedure charges, and physician fees and payments abstracted from medical records, hospital billing, and administrative databases. Mean total cost for this cycle was $10,605.20, which included several costs (physical therapy, radiologist fees) not included in the present study. These studies, though more comprehensive than prior work, did not capture the entire cycle of surgical care.
Our study was designed to provide initial data on the direct costs of arthroscopic repair of the rotator cuff for the entire process cycle. Our overall cost estimate of $5904.21 differs significantly from prior work—not unexpected given the completely different cost methodology used.
Our study had several limitations. First, it was a single-surgeon evaluation, and a number of operating room variables (eg, use of adjunct instrumentation such as radiofrequency probes, differences in draping preferences) as well as surgeon volume in performing rotator cuff repairs might have substantially affected the reproducibility and generalizability of our data. Similarly, the large number of adjunctive procedures (eg, subacromial decompression, labral débridement) performed in conjunction with the rotator cuff repairs added operative time and therefore increased overall cost. Double-row repairs added operative time and increased the cost of consumable materials as well. Differences in surgeon preference for suture anchors may also be important, as anchors are a major cost driver and can vary significantly between vendors and institutions. Tear-related variables (eg, tear size, tear chronicity, degree of fatty cuff degeneration) were not controlled for and might have significantly affected operative time and associated cost. Resident involvement in the surgical procedure and anesthesia process in an academic setting prolongs surgical time and thus directly impacts costs.
In addition, we used the patient’s time in the operating room as a proxy for actual surgical time, as this was the only reliable and reproducible data point available in our electronic medical record. As such, an unquantifiable amount of surgeon time may have been overallocated to our cost estimate for time spent inducing anesthesia, positioning, helping take the patient off the operating table, and so on. However, as typical surgeon practice is to be involved in these tasks in the operating room, the possible overestimate of surgeon cost is likely minimal.
Our salary data for the TDABC algorithm were based on national averages for work hours and gross income for physicians and on hospital-based wage structure and may not be generalizable to other institutions. There may also be regional differences in work hours and salaries, which in turn would factor into a different per-minute cost for surgeon and anesthesiologist, depending on the exact geographic area where the surgery is performed. Costs may be higher at institutions that use certified nurse anesthetists rather than resident physicians because of the salary differences between these practitioners.
Moreover, the time that patients spend in the holding area—waiting to go into surgery and, after surgery, waiting for their ride home, for their prescriptions to be ready, and so forth—is an important variable to consider from a cost standpoint. However, as this time varied significantly and involved minimal contact with hospital personnel, we excluded its associated costs from our analysis. Similarly, and as already noted, hospital overhead and other indirect costs were excluded from analysis as well.
Conclusion
Using the TDABC algorithm, we found a direct economic cost of $5904.21 for RCS at our academic outpatient surgical center, with anchor cost the main cost driver. Judicious use of consumable resources is a key focus for cost containment in arthroscopic shoulder surgery, particularly with respect to implantable suture anchors. However, in the setting of more complex tears that require multiple anchors in a double-row repair construct, our pilot data may be useful to hospitals and surgery centers negotiating procedural reimbursement for the increased cost of complex repairs. Use of the TDABC algorithm for RCS and other procedures may also help in identifying opportunities to deliver more cost-effective health care.
1. American Academy of Orthopaedic Surgeons. The Burden of Musculoskeletal Diseases in the United States: Prevalence, Societal and Economic Cost. Rosemont, IL: American Academy of Orthopaedic Surgeons; 2011.
2. National health expenditure data. Centers for Medicare & Medicare Services website. https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/NationalHealthExpendData/index.html. Updated May 5, 2014. Accessed December 1, 2015.
3. Tashjian RZ. Epidemiology, natural history, and indications for treatment of rotator cuff tears. Clin Sports Med. 2012;31(4):589-604.
4. Colvin AC, Egorova N, Harrison AK, Moskowitz A, Flatow EL. National trends in rotator cuff repair. J Bone Joint Surg Am. 2012;94(3):227-233.
5. Black EM, Higgins LD, Warner JJ. Value-based shoulder surgery: practicing outcomes-driven, cost-conscious care. J Shoulder Elbow Surg. 2013;22(7):1000-1009.
6. Porter ME, Teisberg EO. Redefining Health Care: Creating Value-Based Competition on Results. Boston, MA: Harvard Business School Press; 2006.
7. Kaplan RS, Porter ME. How to solve the cost crisis in health care. Harv Bus Rev. 2011;89(9):46-52, 54, 56-61 passim.
8. Kaplan RS, Anderson SR. Time-driven activity-based costing. Harv Bus Rev. 2004;82(11):131-138, 150.
9. Kaplan RS, Anderson SR. Time-Driven Activity-Based Costing: A Simpler and More Powerful Path to Higher Profits. Boston, MA: Harvard Business Review Press; 2007.
10. American Academy of Orthopaedic Surgeons. Orthopaedic Practice in the U.S. 2012. Rosemont, IL: American Academy of Orthopaedic Surgeons; 2012.
11. Medical Group Management Association. Physician Compensation and Production Survey: 2012 Report Based on 2011 Data. Englewood, CO: Medical Group Management Association; 2012.
12. Kuye IO, Jain NB, Warner L, Herndon JH, Warner JJ. Economic evaluations in shoulder pathologies: a systematic review of the literature. J Shoulder Elbow Surg. 2012;21(3):367-375.
13. Cordasco FA, McGinley BJ, Charlton T. Rotator cuff repair as an outpatient procedure. J Shoulder Elbow Surg. 2000;9(1):27-30.
14. Churchill RS, Ghorai JK. Total cost and operating room time comparison of rotator cuff repair techniques at low, intermediate, and high volume centers: mini-open versus all-arthroscopic. J Shoulder Elbow Surg. 2010;19(5):716-721.
15. Adla DN, Rowsell M, Pandey R. Cost-effectiveness of open versus arthroscopic rotator cuff repair. J Shoulder Elbow Surg. 2010;19(2):258-261.
16. Hearnden A, Tennent D. The cost of shoulder arthroscopy: a comparison with national tariff. Ann R Coll Surg Engl. 2008;90(7):587-591.
17. Xrates currency conversion. http://www.x-rates.com/historical/?from=GBP&amount=1&date=2015-12-03. Accessed December 13, 2015.
18. Vitale MA, Vitale MG, Zivin JG, Braman JP, Bigliani LU, Flatow EL. Rotator cuff repair: an analysis of utility scores and cost-effectiveness. J Shoulder Elbow Surg. 2007;16(2):181-187.
1. American Academy of Orthopaedic Surgeons. The Burden of Musculoskeletal Diseases in the United States: Prevalence, Societal and Economic Cost. Rosemont, IL: American Academy of Orthopaedic Surgeons; 2011.
2. National health expenditure data. Centers for Medicare & Medicare Services website. https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/NationalHealthExpendData/index.html. Updated May 5, 2014. Accessed December 1, 2015.
3. Tashjian RZ. Epidemiology, natural history, and indications for treatment of rotator cuff tears. Clin Sports Med. 2012;31(4):589-604.
4. Colvin AC, Egorova N, Harrison AK, Moskowitz A, Flatow EL. National trends in rotator cuff repair. J Bone Joint Surg Am. 2012;94(3):227-233.
5. Black EM, Higgins LD, Warner JJ. Value-based shoulder surgery: practicing outcomes-driven, cost-conscious care. J Shoulder Elbow Surg. 2013;22(7):1000-1009.
6. Porter ME, Teisberg EO. Redefining Health Care: Creating Value-Based Competition on Results. Boston, MA: Harvard Business School Press; 2006.
7. Kaplan RS, Porter ME. How to solve the cost crisis in health care. Harv Bus Rev. 2011;89(9):46-52, 54, 56-61 passim.
8. Kaplan RS, Anderson SR. Time-driven activity-based costing. Harv Bus Rev. 2004;82(11):131-138, 150.
9. Kaplan RS, Anderson SR. Time-Driven Activity-Based Costing: A Simpler and More Powerful Path to Higher Profits. Boston, MA: Harvard Business Review Press; 2007.
10. American Academy of Orthopaedic Surgeons. Orthopaedic Practice in the U.S. 2012. Rosemont, IL: American Academy of Orthopaedic Surgeons; 2012.
11. Medical Group Management Association. Physician Compensation and Production Survey: 2012 Report Based on 2011 Data. Englewood, CO: Medical Group Management Association; 2012.
12. Kuye IO, Jain NB, Warner L, Herndon JH, Warner JJ. Economic evaluations in shoulder pathologies: a systematic review of the literature. J Shoulder Elbow Surg. 2012;21(3):367-375.
13. Cordasco FA, McGinley BJ, Charlton T. Rotator cuff repair as an outpatient procedure. J Shoulder Elbow Surg. 2000;9(1):27-30.
14. Churchill RS, Ghorai JK. Total cost and operating room time comparison of rotator cuff repair techniques at low, intermediate, and high volume centers: mini-open versus all-arthroscopic. J Shoulder Elbow Surg. 2010;19(5):716-721.
15. Adla DN, Rowsell M, Pandey R. Cost-effectiveness of open versus arthroscopic rotator cuff repair. J Shoulder Elbow Surg. 2010;19(2):258-261.
16. Hearnden A, Tennent D. The cost of shoulder arthroscopy: a comparison with national tariff. Ann R Coll Surg Engl. 2008;90(7):587-591.
17. Xrates currency conversion. http://www.x-rates.com/historical/?from=GBP&amount=1&date=2015-12-03. Accessed December 13, 2015.
18. Vitale MA, Vitale MG, Zivin JG, Braman JP, Bigliani LU, Flatow EL. Rotator cuff repair: an analysis of utility scores and cost-effectiveness. J Shoulder Elbow Surg. 2007;16(2):181-187.
Attending Workload, Teaching, and Safety
Teaching attending physicians must balance clinical workload and resident education simultaneously while supervising inpatient services. The workload of teaching attendings has been increasing due to many factors. As patient complexity has increased, length of stay has decreased, creating higher turnover and higher acuity of hospitalized patients.[1, 2, 3, 4, 5] The rising burden of clinical documentation has increased demands on inpatient attending physicians' time.[6] Additionally, resident duty hour restrictions have shifted the responsibility for patient care to the teaching attending.[7] These factors contribute to the perception of unsafe workloads among attending physicians[8] and could impact the ability to teach well.
Teaching effectiveness is an important facet of the graduate medical education (GME) learning environment.[9] Residents perceive that education suffers when their own workload increases,[10, 11, 12, 13, 14] and higher on‐call workload is associated with lower likelihood of participation in educational activities.[15] More contact between resident trainees and supervisory staff may improve the clinical value of inpatient rotations.[16] Program directors have expressed concern about the educational ramifications of work compression.[17, 18, 19, 20] Higher workload for attending physicians can negatively impact patient safety and quality of care,[21, 22] and perception of higher attending workload is associated with less time for teaching.[23] However, the impact of objective measures of attending physician workload on educational outcomes has not been explored. When attending physicians are responsible for increasingly complex clinical care in addition to resident education, teaching effectiveness may suffer. With growing emphasis on the educational environment's effect on healthcare quality and safety,[24] it is imperative to consider the influence of attending workload on patient care and resident education.
The combination of increasing clinical demands, fewer hours in‐house for residents, and less time for teaching has the potential to decrease attending physician teaching effectiveness. In this study, we aimed to evaluate relationships among objective measures of attending physician workload, resident perception of teaching effectiveness, and patient outcomes. We hypothesized that higher workload for attending physicians would be associated with lower ratings of teaching effectiveness and poorer outcomes for patients.
METHODS
We performed a retrospective study of attending physicians who supervised inpatient internal medicine teaching services at Mayo ClinicRochester from July 2005 through June 2011 (6 full academic years). The team structure for each service was 1 attending physician, 1 senior resident, and 3 interns. Senior residents were on call every fourth night, and interns were on call every sixth night. Up to 2 admissions per service were received during the daytime short call, and up to 5 admissions per service were received during the overnight long call. Attending physicians included all supervising physicians in appointment categories of attending/consultant, senior associate consultant, and chief medical resident at the Mayo Clinic. Maximum continuous on‐call time for residents during the study period was restricted to 30 hours continuously. The timeframe of this study was chosen to minimize variability in resident work schedules; effective July 1, 2011, duty hours for postgraduate year 1 residents were further restricted to a maximum of 16 hours in duration.[25]
Measures of Attending Physician Workload
To measure attending physician workload, we examined mean service census as reported at midnight, mean patient length of stay, mean number of daily admissions, and mean number of daily discharges. We also calculated mean daily outpatient relative value units (RVUs) generated as a measure of outpatient workload while the attending was supervising the inpatient service. Similar measures of workload have been used in previous research.[26] Attending physicians in this study functioned as hospitalists during their time supervising the teaching services; that is, they were not routinely assigned to any outpatient responsibilities. The only way for an outpatient RVU to be generated during their time supervising the hospital service was for the attending physician to specifically request to see an outpatient in the clinic. Attending physicians only supervised 1 teaching service at a time and had no concurrent nonteaching service obligations. Admissions were received on a rotating basis. Because patient illness severity may impact workload, we also examined mean expected mortality (per 1000 patients) for all patients on the attending physicians' hospital services.[27]
The above workload variables were measured in the specific timeframe that corresponded to the number of days an attending physician was supervising a particular team; for example, mean census was the mean number of patients on the attending physician's hospital service during his or her time supervising that resident team.
Teaching Effectiveness Outcome Measures
Teaching effectiveness was measured using residents' evaluations of their attending physicians with a 5‐point scale (1 = needs improvement, 3 = average, 5 = top 10% of attending physicians) that has been previously validated in similar contexts.[28, 29, 30, 31, 32] The evaluation questions are shown in Supporting Information, Appendix A, in the online version of this article.
Patient Outcome Measures
Patient outcomes included applicable patient safety indicators (PSIs) as defined by the Agency for Healthcare Research and Quality[33] (see Supporting Information, Appendix B, in the online version of this article), patient transfers to the intensive care unit (ICU), calls to the rapid response team/cardiopulmonary resuscitation team, and patient deaths. Each indicator and event was summarized as occurred or did not occur at the service‐team level. For example, for a particular attendingresident team, the occurrence of each of these events at any point during the time they worked together was recorded as occurred (1) or did not occur (0). Similar measures of patient outcomes have been used in previous research.[32]
Statistical Analysis
Mixed linear models with variance components covariance structure (including random effects to account for repeated ratings by residents and of faculty) were fit using restricted maximum likelihood to examine associations of attending workload and demographics with teaching scores. Generalized linear regression models, estimated via generalized estimating equations, were used to examine associations of attending workload and demographics with patient outcomes. Due to the binary nature of the outcomes, the binomial distribution and logit link function were used, producing odds ratios (ORs) for covariates akin to those found in standard logistic regression. Multivariate models were used to adjust for physician demographics including age, gender, teaching appointment (consultant, senior associate consultant/temporary clinical appointment, or chief medical resident) and academic rank (professor, associate professor, assistant professor, instructor/none).
To account for multiple comparisons, a significance level of P < 0.01 was used. All analyses were performed using SAS statistical software (version 9.3; SAS Institute Inc., Cary, NC). This study was deemed minimal risk after review by the Mayo Clinic Institutional Review Board.
RESULTS
Over the 6‐year study period, 107 attending physicians supervised internal medicine teaching services. Twenty‐three percent of teaching attending physicians were female. Mean attending age was 42.6 years. Attendings supervised a given service for between 2 and 19 days (mean [standard deviation] = 10.1 [4.1] days). There were 542 internal medicine residents on these teaching services who completed at least 1 teaching evaluation. A total of 69,386 teaching evaluation items were submitted by these residents during the study period.
In a multivariate analysis adjusted for faculty demographics and workload measures, teaching evaluation scores were significantly higher for attending physicians who had an academic rank of professor when compared to attendings who were assistant professors ( = 0.12, P = 0.007), or instructors/no academic rank ( = 0.23, P < 0.0001). The number of days an attending physician spent with the team showed a positive association with teaching evaluations ( = +0.015, P < 0.0001).
Associations between measures of attending physician workload and teaching evaluation scores are shown in Table 1. Mean midnight census and mean number of daily discharges were associated with lower teaching evaluation scores (both = 0.026, P < 0.0001). Mean number of daily admissions was associated with higher teaching scores ( = +0.021, P = 0.001). The mean expected mortality among hospitalized patients on the services supervised by teaching attendings and the outpatient RVUs generated by these attendings during the time they were supervising the hospital service showed no association with teaching scores. The average number of RVUs generated during an attending's entire time supervising hospital service was <1.
| Attending Physician Workload Measure | Mean (SD) | Multivariate Analysis* | |||
|---|---|---|---|---|---|
| SE | 99% CI | P | |||
| |||||
| Midnight census | 8.86 (1.8) | 0.026 | 0.002 | (0.03, 0.02) | <0.0001 |
| Length of stay, d | 6.91 (3.0) | +0.006 | 0.001 | (0.002, 0.009) | <0.0001 |
| Expected mortality (per 1,000 patients) | 51.94 (27.4) | 0.0001 | 0.0001 | (0.0004, 0.0001) | 0.19 |
| Daily admissions | 2.23 (0.54) | +0.021 | 0.006 | (0.004, 0.037) | 0.001 |
| Daily discharges | 2.13 (0.56) | 0.026 | 0.006 | (0.041, 0.010) | <0.0001 |
| Daily outpatient relative value units | 0.69 (1.2) | +0.004 | 0.003 | (0.002, 0.011) | 0.10 |
Table 2 shows relationships between attending physician workload and patient outcomes for the patients on hospital services supervised by 107 attending physicians during the study period. Patient outcome data showed positive associations between measures of higher workload and PSIs. Specifically, for each 1‐patient increase in the average number of daily admissions to the attending and resident services, the cohort of patients under the team's care was 1.8 times more likely to include at least 1 patient with a PSI event (OR = 1.81, 99% confidence interval [CI]: 1.21, 2.71, P = 0.0001). Likewise, for each 1‐day increase in average length of stay, the cohort of patients under the team's care was 1.16 times more likely to have at least 1 patient with a PSI (OR = 1.16, 99% CI: 1.07, 1.26, P < 0.0001). As anticipated, mean expected mortality was associated with actual mortality, cardiopulmonary resuscitation/rapid response team calls, and ICU transfers. There were no associations between patient outcomes and workload measures of midnight census and outpatient RVUs.
| Patient Outcomes, Multivariate Analysis* | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Patient Safety Indicators, n = 513 | Deaths, n = 352 | CPR/RRT Calls, n = 409 | ICU Transfers, n = 737 | |||||||||||||
| Workload measures | OR | SE | P | 99% CI | OR | SE | P | 99% CI | OR | SE | P | 99% CI | OR | SE | P | 99% CI |
| ||||||||||||||||
| Midnight census | 1.10 | 0.05 | 0.04 | (0.98, 1.24) | 0.91 | 0.04 | 0.03 | (0.81, 1.02) | 0.95 | 0.04 | 0.16 | (0.86, 1.05) | 1.06 | 0.04 | 0.16 | (0.96, 1.17) |
| Length of stay | 1.16 | 0.04 | <0.0001 | (1.07, 1.26) | 1.03 | 0.03 | 0.39 | (0.95, 1.12) | 0.99 | 0.03 | 0.63 | (0.92, 1.05) | 1.10 | 0.03 | 0.0001 | (1.03, 1.18) |
| Expected mortality (per 1,000 patients) | 1.00 | 0.003 | 0.24 | (0.99, 1.01) | 1.01 | 0.00 | 0.002 | (1.00, 1.02) | 1.02 | 0.00 | <0.0001 | (1.01, 1.02) | 1.01 | 0.00 | 0.003 | (1.00, 1.01) |
| Daily admissions | 1.81 | 0.28 | 0.0001 | (1.21, 2.71) | 0.78 | 0.14 | 0.16 | (0.49, 1.24) | 1.11 | 0.20 | 0.57 | (0.69, 1.77) | 1.34 | 0.24 | 0.09 | (0.85, 2.11) |
| Daily discharges | 1.06 | 0.13 | 0.61 | (0.78, 1.45) | 2.36 | 0.38 | <0.0001 | (1.56, 3.57) | 0.94 | 0.16 | 0.70 | (0.60, 1.46) | 1.09 | 0.16 | 0.53 | (0.75, 1.60) |
| Daily outpatient relative value units | 0.81 | 0.07 | 0.01 | (0.65, 1.00) | 1.02 | 0.04 | 0.56 | (0.92, 1.13) | 1.05 | 0.04 | 0.23 | (0.95, 1.17) | 0.92 | 0.06 | 0.23 | (0.77, 1.09) |
DISCUSSION
This study of internal medicine attending physician workload and resident education demonstrates that higher workload among attending physicians is associated with slightly lower teaching evaluation scores from residents as well as increased risks to patient safety.
The prior literature examining relationships between workload and teaching effectiveness is largely survey‐based and reliant upon physicians' self‐reported perceptions of workload.[10, 13, 23] The present study strengthens this evidence by using multiple objective measures of workload, objective measures of patient safety, and a large sample of teaching evaluations.
An interesting finding in this study was that the number of patient dismissals per day was associated with a significant decrease in teaching scores, whereas the number of admissions per day was associated with increased teaching scores. These findings may seem contradictory, because the number of admissions and discharges both measure physician workload. However, a likely explanation for this apparent inconsistency is that on internal medicine inpatient teaching services, much of the teaching of residents occurs at the time of a patient admission as residents are presenting cases to the attending physician, exploring differential diagnoses, and discussing management plans. By contrast, a patient dismissal tends to consist mainly of patient interaction, paperwork, and phone calls by the resident with less input required from the attending physician. Our findings suggest that although patient admissions remain a rich opportunity for resident education, patient dismissals may increase workload without improving teaching evaluations. As the inpatient hospital environment evolves, exploring options for nonphysician providers to assist with or complete patient dismissals may have a beneficial effect on resident education.[34] In addition, exploring more efficient teaching strategies may be beneficial in the fast‐paced inpatient learning milieu.[35]
There was a statistically significant positive association between the number of days an attending physician spent with the team and teaching evaluations. Although prior work has examined advantages and disadvantages of various resident schedules,[36, 37, 38] our results suggest scheduling models that emphasize continuity of the teaching attending and residents may be preferred to enhance teaching effectiveness. Further study would help elucidate potential implications of this finding for the scheduling of supervisory attendings to optimize education.
In this analysis, patient outcome measures were largely independent of attending physician workload, with the exception of PSIs. PSIs have been associated with longer stays in the hospital,[39, 40] which is consistent with our findings. However, mean daily admissions were also associated with PSIs. It could be expected that the more patients on a hospital service, the more PSIs will result. However, there was not a significant association between midnight census and PSIs when other variables were accounted for. Because new patient admissions are time consuming and contribute to the workload of both residents and attending physicians, it is possible that safety of the service's hospitalized patients is compromised when the team is putting time and effort toward new patients. Previous research has shown variability in PSI trends with changes in the workload environment.[41] Further studies are needed to fully explore relationships between admission volume and PSIs on teaching services.
It is worthwhile to note that attending physicians have specific responsibilities of supervision and documentation for new admissions. Although it could be argued that new admissions raise the workload for the entire team, and the higher team workload may impact teaching evaluations, previous research has demonstrated that resident burnout and well‐being, which are influenced by workload, do not impact residents' assessments of teachers.[42] In addition, metrics that could arguably be more apt to measure the workload of the team as a whole (eg, team census) did not show a significant association with patient outcomes.
This study has important limitations. First, the cohort of attending physicians, residents, and patients was from a large single institution and may not be generalizable to all settings. Second, most attending physicians in this sample were experienced teachers, so consequences of increased workload may have been managed effectively without a major impact on resident education in some cases. Third, the magnitude of change in teaching effectiveness, although statistically significant, was small and might call into question the educational significance of these findings. Fourth, although resident satisfaction does not influence teaching scores, it is possible that residents' perception of their own workload may have impacted teaching evaluations. Finally, data collection was intentionally closed at the end of the 2011 academic year because accreditation standards for resident duty hours changed again at that time.[43] Thus, these data may not directly reflect the evolving hospital learning environment but serve as a useful benchmark for future studies of workload and teaching effectiveness in the inpatient setting. Once hospitals have had sufficient time and experience with the new duty hour standards, additional studies exploring relationships between workload, teaching effectiveness, and patient outcomes may be warranted.
Limitations notwithstanding, this study shows that attending physician workload may adversely impact teaching and patient safety on internal medicine hospital services. Ongoing efforts by residency programs to optimize the learning environment should include strategies to manage the workload of supervising attendings.
Disclosures
This publication was made possible in part by Clinical and Translational Science Award grant number UL1 TR000135 from the National Center for Advancing Translational Sciences, a component of the National Institutes of Health (NIH). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of NIH. Authors also acknowledge support for the Mayo Clinic Department of Medicine Write‐up and Publish grant. In addition, this study was supported in part by the Mayo Clinic Internal Medicine Residency Office of Education Innovations as part of the Accreditation Council for Graduate Medical Education Educational Innovations Project. The information contained in this article was based in part on the performance package data maintained by the University HealthSystem Consortium. Copyright 2015 UHC. All rights reserved.
- , , . The future of residents' education in internal medicine. Am J Med. 2004;116(9):648–650.
- , , , , . Redesigning residency education in internal medicine: a position paper from the Association of Program Directors in Internal Medicine. Ann Intern Med. 2006;144(12):920–926.
- , , . Residency training in the modern era: the pipe dream of less time to learn more, care better, and be more professional. Arch Intern Med. 2005;165(22):2561–2562.
- , , , et al. Trends in Hospitalizations Among Medicare Survivors of Aortic Valve Replacement in the United States From 1999 to 2010. Ann Thorac Surg. 2015;99(2):509–517.
- , , . Restructuring an inpatient resident service to improve outcomes for residents, students, and patients. Acad Med. 2011;86(12):1500–1507.
- , , , . Clinical documentation in the 21st century: executive summary of a policy position paper from the American College of Physicians. Ann Intern Med. 2015;162(4):301–303.
- , . Effect of ACGME duty hours on attending physician teaching and satisfaction. Arch Intern Med. 2008;168(11):1226–1228.
- , , , , . Identifying potential predictors of a safe attending physician workload: a survey of hospitalists. J Hosp Med. 2013;8(11):644–646.
- , , . The clinical learning environment: the foundation of graduate medical education. JAMA. 2013;309(16):1687–1688.
- , , , , , . Better rested, but more stressed? Evidence of the effects of resident work hour restrictions. Acad Pediatr. 2012;12(4):335–343.
- , , , . Multifaceted longitudinal study of surgical resident education, quality of life, and patient care before and after July 2011. J Surg Educ. 2013;70(6):769–776.
- , , . Impact of the new 16‐hour duty period on pediatric interns' neonatal education. Clin Pediatr (Phila). 2014;53(1):51–59.
- , , , , , . Relationship between resident workload and self‐perceived learning on inpatient medicine wards: a longitudinal study. BMC Med Educ. 2006;6:35.
- , , , . Perceptions of educational experience and inpatient workload among pediatric residents. Hosp Pediatr. 2013;3(3):276–284.
- , , , et al. Association of workload of on‐call medical interns with on‐call sleep duration, shift duration, and participation in educational activities. JAMA. 2008;300(10):1146–1153.
- , , , , , . Effects of increased overnight supervision on resident education, decision‐making, and autonomy. J Hosp Med. 2012;7(8):606–610.
- , , , , . Approval and perceived impact of duty hour regulations: survey of pediatric program directors. Pediatrics. 2013;132(5):819–824.
- , , , et al. Anticipated consequences of the 2011 duty hours standards: views of internal medicine and surgery program directors. Acad Med. 2012;87(7):895–903.
- , , , , . Training on the clock: family medicine residency directors' responses to resident duty hours reform. Acad Med. 2006;81(12):1032–1037.
- , , , et al. Duty hour recommendations and implications for meeting the ACGME core competencies: views of residency directors. Mayo Clin Proc. 2011;86(3):185–191.
- , , , et al. Does surgeon workload per day affect outcomes after pulmonary lobectomies? Ann Thorac Surg. 2012;94(3):966–973.
- , , , . Impact of attending physician workload on patient care: a survey of hospitalists. JAMA Intern Med. 2013;173(5):375–377.
- , , , et al. No time for teaching? Inpatient attending physicians' workload and teaching before and after the implementation of the 2003 duty hours regulations. Acad Med. 2013;88(9):1293–1298.
- Accreditation Council for Graduate Medical Education. Clinical Learning Environment Review (CLER) Program. Available at: http://www.acgme.org/acgmeweb/tabid/436/ProgramandInstitutionalAccreditation/NextAccreditationSystem/ClinicalLearningEnvironmentReviewProgram.aspx. Accessed April 27, 2015.
- Accreditation Council for Graduate Medical Education. Frequently Asked Questions: A ACGME common duty hour requirements. Available at: https://www.acgme.org/acgmeweb/Portals/0/PDFs/dh‐faqs 2011.pdf. Accessed April 27, 2015.
- , , , , . Effect of hospitalist workload on the quality and efficiency of care. JAMA Intern Med. 2014;174(5):786–793.
- University HealthSystem Consortium. UHC clinical database/resource manager for Mayo Clinic. Available at: http://www.uhc.edu. Data accessed August 25, 2011.
- , . The interpersonal, cognitive and efficiency domains of clinical teaching: construct validity of a multi‐dimensional scale. Med Educ. 2005;39(12):1221–1229.
- , , . Factor instability of clinical teaching assessment scores among general internists and cardiologists. Med Educ. 2006;40(12):1209–1216.
- , , , . Determining reliability of clinical assessment scores in real time. Teach Learn Med. 2009;21(3):188–194.
- , , , , , . Behaviors of highly professional resident physicians. JAMA. 2008;300(11):1326–1333.
- , , , et al. Service census caps and unit‐based admissions: resident workload, conference attendance, duty hour compliance, and patient safety. Mayo Clin Proc. 2012;87(4):320–327.
- Agency for Healthcare Research and Quality. Patient safety indicators technical specifications updates—Version 5.0, March 2015. Available at: http://www.qualityindicators.ahrq.gov/Modules/PSI_TechSpec.aspx. Accessed May 29, 2015.
- , , , , , . The impact of nonphysician clinicians: do they improve the quality and cost‐effectiveness of health care services? Med Care Res Rev. 2009;66(6 suppl):36S–89S.
- , , . Maximizing teaching on the wards: review and application of the One‐Minute Preceptor and SNAPPS models. J Hosp Med. 2015;10(2):125–130.
- , , , . Resident perceptions of the educational value of night float rotations. Teach Learn Med. 2010;22(3):196–201.
- , , , , , . An evaluation of internal medicine residency continuity clinic redesign to a 50/50 outpatient‐inpatient model. J Gen Intern Med. 2013;28(8):1014–1019.
- , , , et al. Revisiting the rotating call schedule in less than 80 hours per week. J Surg Educ. 2009;66(6):357–360.
- , . Excess length of stay, charges, and mortality attributable to medical injuries during hospitalization. JAMA. 2003;290(14):1868–1874.
- , , , . Agency for Healthcare Research and Quality patient safety indicators and mortality in surgical patients. Am Surg. 2014;80(8):801–804.
- , , , et al. Patient safety in the era of the 80‐hour workweek. J Surg Educ. 2014;71(4):551–559.
- , , , . Impact of resident well‐being and empathy on assessments of faculty physicians. J Gen Intern Med. 2010;25(1):52–56.
- , , , et al. Stress management training for surgeons‐a randomized, controlled, intervention study. Ann Surg. 2011;253(3):488–494.
Teaching attending physicians must balance clinical workload and resident education simultaneously while supervising inpatient services. The workload of teaching attendings has been increasing due to many factors. As patient complexity has increased, length of stay has decreased, creating higher turnover and higher acuity of hospitalized patients.[1, 2, 3, 4, 5] The rising burden of clinical documentation has increased demands on inpatient attending physicians' time.[6] Additionally, resident duty hour restrictions have shifted the responsibility for patient care to the teaching attending.[7] These factors contribute to the perception of unsafe workloads among attending physicians[8] and could impact the ability to teach well.
Teaching effectiveness is an important facet of the graduate medical education (GME) learning environment.[9] Residents perceive that education suffers when their own workload increases,[10, 11, 12, 13, 14] and higher on‐call workload is associated with lower likelihood of participation in educational activities.[15] More contact between resident trainees and supervisory staff may improve the clinical value of inpatient rotations.[16] Program directors have expressed concern about the educational ramifications of work compression.[17, 18, 19, 20] Higher workload for attending physicians can negatively impact patient safety and quality of care,[21, 22] and perception of higher attending workload is associated with less time for teaching.[23] However, the impact of objective measures of attending physician workload on educational outcomes has not been explored. When attending physicians are responsible for increasingly complex clinical care in addition to resident education, teaching effectiveness may suffer. With growing emphasis on the educational environment's effect on healthcare quality and safety,[24] it is imperative to consider the influence of attending workload on patient care and resident education.
The combination of increasing clinical demands, fewer hours in‐house for residents, and less time for teaching has the potential to decrease attending physician teaching effectiveness. In this study, we aimed to evaluate relationships among objective measures of attending physician workload, resident perception of teaching effectiveness, and patient outcomes. We hypothesized that higher workload for attending physicians would be associated with lower ratings of teaching effectiveness and poorer outcomes for patients.
METHODS
We performed a retrospective study of attending physicians who supervised inpatient internal medicine teaching services at Mayo ClinicRochester from July 2005 through June 2011 (6 full academic years). The team structure for each service was 1 attending physician, 1 senior resident, and 3 interns. Senior residents were on call every fourth night, and interns were on call every sixth night. Up to 2 admissions per service were received during the daytime short call, and up to 5 admissions per service were received during the overnight long call. Attending physicians included all supervising physicians in appointment categories of attending/consultant, senior associate consultant, and chief medical resident at the Mayo Clinic. Maximum continuous on‐call time for residents during the study period was restricted to 30 hours continuously. The timeframe of this study was chosen to minimize variability in resident work schedules; effective July 1, 2011, duty hours for postgraduate year 1 residents were further restricted to a maximum of 16 hours in duration.[25]
Measures of Attending Physician Workload
To measure attending physician workload, we examined mean service census as reported at midnight, mean patient length of stay, mean number of daily admissions, and mean number of daily discharges. We also calculated mean daily outpatient relative value units (RVUs) generated as a measure of outpatient workload while the attending was supervising the inpatient service. Similar measures of workload have been used in previous research.[26] Attending physicians in this study functioned as hospitalists during their time supervising the teaching services; that is, they were not routinely assigned to any outpatient responsibilities. The only way for an outpatient RVU to be generated during their time supervising the hospital service was for the attending physician to specifically request to see an outpatient in the clinic. Attending physicians only supervised 1 teaching service at a time and had no concurrent nonteaching service obligations. Admissions were received on a rotating basis. Because patient illness severity may impact workload, we also examined mean expected mortality (per 1000 patients) for all patients on the attending physicians' hospital services.[27]
The above workload variables were measured in the specific timeframe that corresponded to the number of days an attending physician was supervising a particular team; for example, mean census was the mean number of patients on the attending physician's hospital service during his or her time supervising that resident team.
Teaching Effectiveness Outcome Measures
Teaching effectiveness was measured using residents' evaluations of their attending physicians with a 5‐point scale (1 = needs improvement, 3 = average, 5 = top 10% of attending physicians) that has been previously validated in similar contexts.[28, 29, 30, 31, 32] The evaluation questions are shown in Supporting Information, Appendix A, in the online version of this article.
Patient Outcome Measures
Patient outcomes included applicable patient safety indicators (PSIs) as defined by the Agency for Healthcare Research and Quality[33] (see Supporting Information, Appendix B, in the online version of this article), patient transfers to the intensive care unit (ICU), calls to the rapid response team/cardiopulmonary resuscitation team, and patient deaths. Each indicator and event was summarized as occurred or did not occur at the service‐team level. For example, for a particular attendingresident team, the occurrence of each of these events at any point during the time they worked together was recorded as occurred (1) or did not occur (0). Similar measures of patient outcomes have been used in previous research.[32]
Statistical Analysis
Mixed linear models with variance components covariance structure (including random effects to account for repeated ratings by residents and of faculty) were fit using restricted maximum likelihood to examine associations of attending workload and demographics with teaching scores. Generalized linear regression models, estimated via generalized estimating equations, were used to examine associations of attending workload and demographics with patient outcomes. Due to the binary nature of the outcomes, the binomial distribution and logit link function were used, producing odds ratios (ORs) for covariates akin to those found in standard logistic regression. Multivariate models were used to adjust for physician demographics including age, gender, teaching appointment (consultant, senior associate consultant/temporary clinical appointment, or chief medical resident) and academic rank (professor, associate professor, assistant professor, instructor/none).
To account for multiple comparisons, a significance level of P < 0.01 was used. All analyses were performed using SAS statistical software (version 9.3; SAS Institute Inc., Cary, NC). This study was deemed minimal risk after review by the Mayo Clinic Institutional Review Board.
RESULTS
Over the 6‐year study period, 107 attending physicians supervised internal medicine teaching services. Twenty‐three percent of teaching attending physicians were female. Mean attending age was 42.6 years. Attendings supervised a given service for between 2 and 19 days (mean [standard deviation] = 10.1 [4.1] days). There were 542 internal medicine residents on these teaching services who completed at least 1 teaching evaluation. A total of 69,386 teaching evaluation items were submitted by these residents during the study period.
In a multivariate analysis adjusted for faculty demographics and workload measures, teaching evaluation scores were significantly higher for attending physicians who had an academic rank of professor when compared to attendings who were assistant professors ( = 0.12, P = 0.007), or instructors/no academic rank ( = 0.23, P < 0.0001). The number of days an attending physician spent with the team showed a positive association with teaching evaluations ( = +0.015, P < 0.0001).
Associations between measures of attending physician workload and teaching evaluation scores are shown in Table 1. Mean midnight census and mean number of daily discharges were associated with lower teaching evaluation scores (both = 0.026, P < 0.0001). Mean number of daily admissions was associated with higher teaching scores ( = +0.021, P = 0.001). The mean expected mortality among hospitalized patients on the services supervised by teaching attendings and the outpatient RVUs generated by these attendings during the time they were supervising the hospital service showed no association with teaching scores. The average number of RVUs generated during an attending's entire time supervising hospital service was <1.
| Attending Physician Workload Measure | Mean (SD) | Multivariate Analysis* | |||
|---|---|---|---|---|---|
| SE | 99% CI | P | |||
| |||||
| Midnight census | 8.86 (1.8) | 0.026 | 0.002 | (0.03, 0.02) | <0.0001 |
| Length of stay, d | 6.91 (3.0) | +0.006 | 0.001 | (0.002, 0.009) | <0.0001 |
| Expected mortality (per 1,000 patients) | 51.94 (27.4) | 0.0001 | 0.0001 | (0.0004, 0.0001) | 0.19 |
| Daily admissions | 2.23 (0.54) | +0.021 | 0.006 | (0.004, 0.037) | 0.001 |
| Daily discharges | 2.13 (0.56) | 0.026 | 0.006 | (0.041, 0.010) | <0.0001 |
| Daily outpatient relative value units | 0.69 (1.2) | +0.004 | 0.003 | (0.002, 0.011) | 0.10 |
Table 2 shows relationships between attending physician workload and patient outcomes for the patients on hospital services supervised by 107 attending physicians during the study period. Patient outcome data showed positive associations between measures of higher workload and PSIs. Specifically, for each 1‐patient increase in the average number of daily admissions to the attending and resident services, the cohort of patients under the team's care was 1.8 times more likely to include at least 1 patient with a PSI event (OR = 1.81, 99% confidence interval [CI]: 1.21, 2.71, P = 0.0001). Likewise, for each 1‐day increase in average length of stay, the cohort of patients under the team's care was 1.16 times more likely to have at least 1 patient with a PSI (OR = 1.16, 99% CI: 1.07, 1.26, P < 0.0001). As anticipated, mean expected mortality was associated with actual mortality, cardiopulmonary resuscitation/rapid response team calls, and ICU transfers. There were no associations between patient outcomes and workload measures of midnight census and outpatient RVUs.
| Patient Outcomes, Multivariate Analysis* | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Patient Safety Indicators, n = 513 | Deaths, n = 352 | CPR/RRT Calls, n = 409 | ICU Transfers, n = 737 | |||||||||||||
| Workload measures | OR | SE | P | 99% CI | OR | SE | P | 99% CI | OR | SE | P | 99% CI | OR | SE | P | 99% CI |
| ||||||||||||||||
| Midnight census | 1.10 | 0.05 | 0.04 | (0.98, 1.24) | 0.91 | 0.04 | 0.03 | (0.81, 1.02) | 0.95 | 0.04 | 0.16 | (0.86, 1.05) | 1.06 | 0.04 | 0.16 | (0.96, 1.17) |
| Length of stay | 1.16 | 0.04 | <0.0001 | (1.07, 1.26) | 1.03 | 0.03 | 0.39 | (0.95, 1.12) | 0.99 | 0.03 | 0.63 | (0.92, 1.05) | 1.10 | 0.03 | 0.0001 | (1.03, 1.18) |
| Expected mortality (per 1,000 patients) | 1.00 | 0.003 | 0.24 | (0.99, 1.01) | 1.01 | 0.00 | 0.002 | (1.00, 1.02) | 1.02 | 0.00 | <0.0001 | (1.01, 1.02) | 1.01 | 0.00 | 0.003 | (1.00, 1.01) |
| Daily admissions | 1.81 | 0.28 | 0.0001 | (1.21, 2.71) | 0.78 | 0.14 | 0.16 | (0.49, 1.24) | 1.11 | 0.20 | 0.57 | (0.69, 1.77) | 1.34 | 0.24 | 0.09 | (0.85, 2.11) |
| Daily discharges | 1.06 | 0.13 | 0.61 | (0.78, 1.45) | 2.36 | 0.38 | <0.0001 | (1.56, 3.57) | 0.94 | 0.16 | 0.70 | (0.60, 1.46) | 1.09 | 0.16 | 0.53 | (0.75, 1.60) |
| Daily outpatient relative value units | 0.81 | 0.07 | 0.01 | (0.65, 1.00) | 1.02 | 0.04 | 0.56 | (0.92, 1.13) | 1.05 | 0.04 | 0.23 | (0.95, 1.17) | 0.92 | 0.06 | 0.23 | (0.77, 1.09) |
DISCUSSION
This study of internal medicine attending physician workload and resident education demonstrates that higher workload among attending physicians is associated with slightly lower teaching evaluation scores from residents as well as increased risks to patient safety.
The prior literature examining relationships between workload and teaching effectiveness is largely survey‐based and reliant upon physicians' self‐reported perceptions of workload.[10, 13, 23] The present study strengthens this evidence by using multiple objective measures of workload, objective measures of patient safety, and a large sample of teaching evaluations.
An interesting finding in this study was that the number of patient dismissals per day was associated with a significant decrease in teaching scores, whereas the number of admissions per day was associated with increased teaching scores. These findings may seem contradictory, because the number of admissions and discharges both measure physician workload. However, a likely explanation for this apparent inconsistency is that on internal medicine inpatient teaching services, much of the teaching of residents occurs at the time of a patient admission as residents are presenting cases to the attending physician, exploring differential diagnoses, and discussing management plans. By contrast, a patient dismissal tends to consist mainly of patient interaction, paperwork, and phone calls by the resident with less input required from the attending physician. Our findings suggest that although patient admissions remain a rich opportunity for resident education, patient dismissals may increase workload without improving teaching evaluations. As the inpatient hospital environment evolves, exploring options for nonphysician providers to assist with or complete patient dismissals may have a beneficial effect on resident education.[34] In addition, exploring more efficient teaching strategies may be beneficial in the fast‐paced inpatient learning milieu.[35]
There was a statistically significant positive association between the number of days an attending physician spent with the team and teaching evaluations. Although prior work has examined advantages and disadvantages of various resident schedules,[36, 37, 38] our results suggest scheduling models that emphasize continuity of the teaching attending and residents may be preferred to enhance teaching effectiveness. Further study would help elucidate potential implications of this finding for the scheduling of supervisory attendings to optimize education.
In this analysis, patient outcome measures were largely independent of attending physician workload, with the exception of PSIs. PSIs have been associated with longer stays in the hospital,[39, 40] which is consistent with our findings. However, mean daily admissions were also associated with PSIs. It could be expected that the more patients on a hospital service, the more PSIs will result. However, there was not a significant association between midnight census and PSIs when other variables were accounted for. Because new patient admissions are time consuming and contribute to the workload of both residents and attending physicians, it is possible that safety of the service's hospitalized patients is compromised when the team is putting time and effort toward new patients. Previous research has shown variability in PSI trends with changes in the workload environment.[41] Further studies are needed to fully explore relationships between admission volume and PSIs on teaching services.
It is worthwhile to note that attending physicians have specific responsibilities of supervision and documentation for new admissions. Although it could be argued that new admissions raise the workload for the entire team, and the higher team workload may impact teaching evaluations, previous research has demonstrated that resident burnout and well‐being, which are influenced by workload, do not impact residents' assessments of teachers.[42] In addition, metrics that could arguably be more apt to measure the workload of the team as a whole (eg, team census) did not show a significant association with patient outcomes.
This study has important limitations. First, the cohort of attending physicians, residents, and patients was from a large single institution and may not be generalizable to all settings. Second, most attending physicians in this sample were experienced teachers, so consequences of increased workload may have been managed effectively without a major impact on resident education in some cases. Third, the magnitude of change in teaching effectiveness, although statistically significant, was small and might call into question the educational significance of these findings. Fourth, although resident satisfaction does not influence teaching scores, it is possible that residents' perception of their own workload may have impacted teaching evaluations. Finally, data collection was intentionally closed at the end of the 2011 academic year because accreditation standards for resident duty hours changed again at that time.[43] Thus, these data may not directly reflect the evolving hospital learning environment but serve as a useful benchmark for future studies of workload and teaching effectiveness in the inpatient setting. Once hospitals have had sufficient time and experience with the new duty hour standards, additional studies exploring relationships between workload, teaching effectiveness, and patient outcomes may be warranted.
Limitations notwithstanding, this study shows that attending physician workload may adversely impact teaching and patient safety on internal medicine hospital services. Ongoing efforts by residency programs to optimize the learning environment should include strategies to manage the workload of supervising attendings.
Disclosures
This publication was made possible in part by Clinical and Translational Science Award grant number UL1 TR000135 from the National Center for Advancing Translational Sciences, a component of the National Institutes of Health (NIH). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of NIH. Authors also acknowledge support for the Mayo Clinic Department of Medicine Write‐up and Publish grant. In addition, this study was supported in part by the Mayo Clinic Internal Medicine Residency Office of Education Innovations as part of the Accreditation Council for Graduate Medical Education Educational Innovations Project. The information contained in this article was based in part on the performance package data maintained by the University HealthSystem Consortium. Copyright 2015 UHC. All rights reserved.
Teaching attending physicians must balance clinical workload and resident education simultaneously while supervising inpatient services. The workload of teaching attendings has been increasing due to many factors. As patient complexity has increased, length of stay has decreased, creating higher turnover and higher acuity of hospitalized patients.[1, 2, 3, 4, 5] The rising burden of clinical documentation has increased demands on inpatient attending physicians' time.[6] Additionally, resident duty hour restrictions have shifted the responsibility for patient care to the teaching attending.[7] These factors contribute to the perception of unsafe workloads among attending physicians[8] and could impact the ability to teach well.
Teaching effectiveness is an important facet of the graduate medical education (GME) learning environment.[9] Residents perceive that education suffers when their own workload increases,[10, 11, 12, 13, 14] and higher on‐call workload is associated with lower likelihood of participation in educational activities.[15] More contact between resident trainees and supervisory staff may improve the clinical value of inpatient rotations.[16] Program directors have expressed concern about the educational ramifications of work compression.[17, 18, 19, 20] Higher workload for attending physicians can negatively impact patient safety and quality of care,[21, 22] and perception of higher attending workload is associated with less time for teaching.[23] However, the impact of objective measures of attending physician workload on educational outcomes has not been explored. When attending physicians are responsible for increasingly complex clinical care in addition to resident education, teaching effectiveness may suffer. With growing emphasis on the educational environment's effect on healthcare quality and safety,[24] it is imperative to consider the influence of attending workload on patient care and resident education.
The combination of increasing clinical demands, fewer hours in‐house for residents, and less time for teaching has the potential to decrease attending physician teaching effectiveness. In this study, we aimed to evaluate relationships among objective measures of attending physician workload, resident perception of teaching effectiveness, and patient outcomes. We hypothesized that higher workload for attending physicians would be associated with lower ratings of teaching effectiveness and poorer outcomes for patients.
METHODS
We performed a retrospective study of attending physicians who supervised inpatient internal medicine teaching services at Mayo ClinicRochester from July 2005 through June 2011 (6 full academic years). The team structure for each service was 1 attending physician, 1 senior resident, and 3 interns. Senior residents were on call every fourth night, and interns were on call every sixth night. Up to 2 admissions per service were received during the daytime short call, and up to 5 admissions per service were received during the overnight long call. Attending physicians included all supervising physicians in appointment categories of attending/consultant, senior associate consultant, and chief medical resident at the Mayo Clinic. Maximum continuous on‐call time for residents during the study period was restricted to 30 hours continuously. The timeframe of this study was chosen to minimize variability in resident work schedules; effective July 1, 2011, duty hours for postgraduate year 1 residents were further restricted to a maximum of 16 hours in duration.[25]
Measures of Attending Physician Workload
To measure attending physician workload, we examined mean service census as reported at midnight, mean patient length of stay, mean number of daily admissions, and mean number of daily discharges. We also calculated mean daily outpatient relative value units (RVUs) generated as a measure of outpatient workload while the attending was supervising the inpatient service. Similar measures of workload have been used in previous research.[26] Attending physicians in this study functioned as hospitalists during their time supervising the teaching services; that is, they were not routinely assigned to any outpatient responsibilities. The only way for an outpatient RVU to be generated during their time supervising the hospital service was for the attending physician to specifically request to see an outpatient in the clinic. Attending physicians only supervised 1 teaching service at a time and had no concurrent nonteaching service obligations. Admissions were received on a rotating basis. Because patient illness severity may impact workload, we also examined mean expected mortality (per 1000 patients) for all patients on the attending physicians' hospital services.[27]
The above workload variables were measured in the specific timeframe that corresponded to the number of days an attending physician was supervising a particular team; for example, mean census was the mean number of patients on the attending physician's hospital service during his or her time supervising that resident team.
Teaching Effectiveness Outcome Measures
Teaching effectiveness was measured using residents' evaluations of their attending physicians with a 5‐point scale (1 = needs improvement, 3 = average, 5 = top 10% of attending physicians) that has been previously validated in similar contexts.[28, 29, 30, 31, 32] The evaluation questions are shown in Supporting Information, Appendix A, in the online version of this article.
Patient Outcome Measures
Patient outcomes included applicable patient safety indicators (PSIs) as defined by the Agency for Healthcare Research and Quality[33] (see Supporting Information, Appendix B, in the online version of this article), patient transfers to the intensive care unit (ICU), calls to the rapid response team/cardiopulmonary resuscitation team, and patient deaths. Each indicator and event was summarized as occurred or did not occur at the service‐team level. For example, for a particular attendingresident team, the occurrence of each of these events at any point during the time they worked together was recorded as occurred (1) or did not occur (0). Similar measures of patient outcomes have been used in previous research.[32]
Statistical Analysis
Mixed linear models with variance components covariance structure (including random effects to account for repeated ratings by residents and of faculty) were fit using restricted maximum likelihood to examine associations of attending workload and demographics with teaching scores. Generalized linear regression models, estimated via generalized estimating equations, were used to examine associations of attending workload and demographics with patient outcomes. Due to the binary nature of the outcomes, the binomial distribution and logit link function were used, producing odds ratios (ORs) for covariates akin to those found in standard logistic regression. Multivariate models were used to adjust for physician demographics including age, gender, teaching appointment (consultant, senior associate consultant/temporary clinical appointment, or chief medical resident) and academic rank (professor, associate professor, assistant professor, instructor/none).
To account for multiple comparisons, a significance level of P < 0.01 was used. All analyses were performed using SAS statistical software (version 9.3; SAS Institute Inc., Cary, NC). This study was deemed minimal risk after review by the Mayo Clinic Institutional Review Board.
RESULTS
Over the 6‐year study period, 107 attending physicians supervised internal medicine teaching services. Twenty‐three percent of teaching attending physicians were female. Mean attending age was 42.6 years. Attendings supervised a given service for between 2 and 19 days (mean [standard deviation] = 10.1 [4.1] days). There were 542 internal medicine residents on these teaching services who completed at least 1 teaching evaluation. A total of 69,386 teaching evaluation items were submitted by these residents during the study period.
In a multivariate analysis adjusted for faculty demographics and workload measures, teaching evaluation scores were significantly higher for attending physicians who had an academic rank of professor when compared to attendings who were assistant professors ( = 0.12, P = 0.007), or instructors/no academic rank ( = 0.23, P < 0.0001). The number of days an attending physician spent with the team showed a positive association with teaching evaluations ( = +0.015, P < 0.0001).
Associations between measures of attending physician workload and teaching evaluation scores are shown in Table 1. Mean midnight census and mean number of daily discharges were associated with lower teaching evaluation scores (both = 0.026, P < 0.0001). Mean number of daily admissions was associated with higher teaching scores ( = +0.021, P = 0.001). The mean expected mortality among hospitalized patients on the services supervised by teaching attendings and the outpatient RVUs generated by these attendings during the time they were supervising the hospital service showed no association with teaching scores. The average number of RVUs generated during an attending's entire time supervising hospital service was <1.
| Attending Physician Workload Measure | Mean (SD) | Multivariate Analysis* | |||
|---|---|---|---|---|---|
| SE | 99% CI | P | |||
| |||||
| Midnight census | 8.86 (1.8) | 0.026 | 0.002 | (0.03, 0.02) | <0.0001 |
| Length of stay, d | 6.91 (3.0) | +0.006 | 0.001 | (0.002, 0.009) | <0.0001 |
| Expected mortality (per 1,000 patients) | 51.94 (27.4) | 0.0001 | 0.0001 | (0.0004, 0.0001) | 0.19 |
| Daily admissions | 2.23 (0.54) | +0.021 | 0.006 | (0.004, 0.037) | 0.001 |
| Daily discharges | 2.13 (0.56) | 0.026 | 0.006 | (0.041, 0.010) | <0.0001 |
| Daily outpatient relative value units | 0.69 (1.2) | +0.004 | 0.003 | (0.002, 0.011) | 0.10 |
Table 2 shows relationships between attending physician workload and patient outcomes for the patients on hospital services supervised by 107 attending physicians during the study period. Patient outcome data showed positive associations between measures of higher workload and PSIs. Specifically, for each 1‐patient increase in the average number of daily admissions to the attending and resident services, the cohort of patients under the team's care was 1.8 times more likely to include at least 1 patient with a PSI event (OR = 1.81, 99% confidence interval [CI]: 1.21, 2.71, P = 0.0001). Likewise, for each 1‐day increase in average length of stay, the cohort of patients under the team's care was 1.16 times more likely to have at least 1 patient with a PSI (OR = 1.16, 99% CI: 1.07, 1.26, P < 0.0001). As anticipated, mean expected mortality was associated with actual mortality, cardiopulmonary resuscitation/rapid response team calls, and ICU transfers. There were no associations between patient outcomes and workload measures of midnight census and outpatient RVUs.
| Patient Outcomes, Multivariate Analysis* | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Patient Safety Indicators, n = 513 | Deaths, n = 352 | CPR/RRT Calls, n = 409 | ICU Transfers, n = 737 | |||||||||||||
| Workload measures | OR | SE | P | 99% CI | OR | SE | P | 99% CI | OR | SE | P | 99% CI | OR | SE | P | 99% CI |
| ||||||||||||||||
| Midnight census | 1.10 | 0.05 | 0.04 | (0.98, 1.24) | 0.91 | 0.04 | 0.03 | (0.81, 1.02) | 0.95 | 0.04 | 0.16 | (0.86, 1.05) | 1.06 | 0.04 | 0.16 | (0.96, 1.17) |
| Length of stay | 1.16 | 0.04 | <0.0001 | (1.07, 1.26) | 1.03 | 0.03 | 0.39 | (0.95, 1.12) | 0.99 | 0.03 | 0.63 | (0.92, 1.05) | 1.10 | 0.03 | 0.0001 | (1.03, 1.18) |
| Expected mortality (per 1,000 patients) | 1.00 | 0.003 | 0.24 | (0.99, 1.01) | 1.01 | 0.00 | 0.002 | (1.00, 1.02) | 1.02 | 0.00 | <0.0001 | (1.01, 1.02) | 1.01 | 0.00 | 0.003 | (1.00, 1.01) |
| Daily admissions | 1.81 | 0.28 | 0.0001 | (1.21, 2.71) | 0.78 | 0.14 | 0.16 | (0.49, 1.24) | 1.11 | 0.20 | 0.57 | (0.69, 1.77) | 1.34 | 0.24 | 0.09 | (0.85, 2.11) |
| Daily discharges | 1.06 | 0.13 | 0.61 | (0.78, 1.45) | 2.36 | 0.38 | <0.0001 | (1.56, 3.57) | 0.94 | 0.16 | 0.70 | (0.60, 1.46) | 1.09 | 0.16 | 0.53 | (0.75, 1.60) |
| Daily outpatient relative value units | 0.81 | 0.07 | 0.01 | (0.65, 1.00) | 1.02 | 0.04 | 0.56 | (0.92, 1.13) | 1.05 | 0.04 | 0.23 | (0.95, 1.17) | 0.92 | 0.06 | 0.23 | (0.77, 1.09) |
DISCUSSION
This study of internal medicine attending physician workload and resident education demonstrates that higher workload among attending physicians is associated with slightly lower teaching evaluation scores from residents as well as increased risks to patient safety.
The prior literature examining relationships between workload and teaching effectiveness is largely survey‐based and reliant upon physicians' self‐reported perceptions of workload.[10, 13, 23] The present study strengthens this evidence by using multiple objective measures of workload, objective measures of patient safety, and a large sample of teaching evaluations.
An interesting finding in this study was that the number of patient dismissals per day was associated with a significant decrease in teaching scores, whereas the number of admissions per day was associated with increased teaching scores. These findings may seem contradictory, because the number of admissions and discharges both measure physician workload. However, a likely explanation for this apparent inconsistency is that on internal medicine inpatient teaching services, much of the teaching of residents occurs at the time of a patient admission as residents are presenting cases to the attending physician, exploring differential diagnoses, and discussing management plans. By contrast, a patient dismissal tends to consist mainly of patient interaction, paperwork, and phone calls by the resident with less input required from the attending physician. Our findings suggest that although patient admissions remain a rich opportunity for resident education, patient dismissals may increase workload without improving teaching evaluations. As the inpatient hospital environment evolves, exploring options for nonphysician providers to assist with or complete patient dismissals may have a beneficial effect on resident education.[34] In addition, exploring more efficient teaching strategies may be beneficial in the fast‐paced inpatient learning milieu.[35]
There was a statistically significant positive association between the number of days an attending physician spent with the team and teaching evaluations. Although prior work has examined advantages and disadvantages of various resident schedules,[36, 37, 38] our results suggest scheduling models that emphasize continuity of the teaching attending and residents may be preferred to enhance teaching effectiveness. Further study would help elucidate potential implications of this finding for the scheduling of supervisory attendings to optimize education.
In this analysis, patient outcome measures were largely independent of attending physician workload, with the exception of PSIs. PSIs have been associated with longer stays in the hospital,[39, 40] which is consistent with our findings. However, mean daily admissions were also associated with PSIs. It could be expected that the more patients on a hospital service, the more PSIs will result. However, there was not a significant association between midnight census and PSIs when other variables were accounted for. Because new patient admissions are time consuming and contribute to the workload of both residents and attending physicians, it is possible that safety of the service's hospitalized patients is compromised when the team is putting time and effort toward new patients. Previous research has shown variability in PSI trends with changes in the workload environment.[41] Further studies are needed to fully explore relationships between admission volume and PSIs on teaching services.
It is worthwhile to note that attending physicians have specific responsibilities of supervision and documentation for new admissions. Although it could be argued that new admissions raise the workload for the entire team, and the higher team workload may impact teaching evaluations, previous research has demonstrated that resident burnout and well‐being, which are influenced by workload, do not impact residents' assessments of teachers.[42] In addition, metrics that could arguably be more apt to measure the workload of the team as a whole (eg, team census) did not show a significant association with patient outcomes.
This study has important limitations. First, the cohort of attending physicians, residents, and patients was from a large single institution and may not be generalizable to all settings. Second, most attending physicians in this sample were experienced teachers, so consequences of increased workload may have been managed effectively without a major impact on resident education in some cases. Third, the magnitude of change in teaching effectiveness, although statistically significant, was small and might call into question the educational significance of these findings. Fourth, although resident satisfaction does not influence teaching scores, it is possible that residents' perception of their own workload may have impacted teaching evaluations. Finally, data collection was intentionally closed at the end of the 2011 academic year because accreditation standards for resident duty hours changed again at that time.[43] Thus, these data may not directly reflect the evolving hospital learning environment but serve as a useful benchmark for future studies of workload and teaching effectiveness in the inpatient setting. Once hospitals have had sufficient time and experience with the new duty hour standards, additional studies exploring relationships between workload, teaching effectiveness, and patient outcomes may be warranted.
Limitations notwithstanding, this study shows that attending physician workload may adversely impact teaching and patient safety on internal medicine hospital services. Ongoing efforts by residency programs to optimize the learning environment should include strategies to manage the workload of supervising attendings.
Disclosures
This publication was made possible in part by Clinical and Translational Science Award grant number UL1 TR000135 from the National Center for Advancing Translational Sciences, a component of the National Institutes of Health (NIH). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of NIH. Authors also acknowledge support for the Mayo Clinic Department of Medicine Write‐up and Publish grant. In addition, this study was supported in part by the Mayo Clinic Internal Medicine Residency Office of Education Innovations as part of the Accreditation Council for Graduate Medical Education Educational Innovations Project. The information contained in this article was based in part on the performance package data maintained by the University HealthSystem Consortium. Copyright 2015 UHC. All rights reserved.
- , , . The future of residents' education in internal medicine. Am J Med. 2004;116(9):648–650.
- , , , , . Redesigning residency education in internal medicine: a position paper from the Association of Program Directors in Internal Medicine. Ann Intern Med. 2006;144(12):920–926.
- , , . Residency training in the modern era: the pipe dream of less time to learn more, care better, and be more professional. Arch Intern Med. 2005;165(22):2561–2562.
- , , , et al. Trends in Hospitalizations Among Medicare Survivors of Aortic Valve Replacement in the United States From 1999 to 2010. Ann Thorac Surg. 2015;99(2):509–517.
- , , . Restructuring an inpatient resident service to improve outcomes for residents, students, and patients. Acad Med. 2011;86(12):1500–1507.
- , , , . Clinical documentation in the 21st century: executive summary of a policy position paper from the American College of Physicians. Ann Intern Med. 2015;162(4):301–303.
- , . Effect of ACGME duty hours on attending physician teaching and satisfaction. Arch Intern Med. 2008;168(11):1226–1228.
- , , , , . Identifying potential predictors of a safe attending physician workload: a survey of hospitalists. J Hosp Med. 2013;8(11):644–646.
- , , . The clinical learning environment: the foundation of graduate medical education. JAMA. 2013;309(16):1687–1688.
- , , , , , . Better rested, but more stressed? Evidence of the effects of resident work hour restrictions. Acad Pediatr. 2012;12(4):335–343.
- , , , . Multifaceted longitudinal study of surgical resident education, quality of life, and patient care before and after July 2011. J Surg Educ. 2013;70(6):769–776.
- , , . Impact of the new 16‐hour duty period on pediatric interns' neonatal education. Clin Pediatr (Phila). 2014;53(1):51–59.
- , , , , , . Relationship between resident workload and self‐perceived learning on inpatient medicine wards: a longitudinal study. BMC Med Educ. 2006;6:35.
- , , , . Perceptions of educational experience and inpatient workload among pediatric residents. Hosp Pediatr. 2013;3(3):276–284.
- , , , et al. Association of workload of on‐call medical interns with on‐call sleep duration, shift duration, and participation in educational activities. JAMA. 2008;300(10):1146–1153.
- , , , , , . Effects of increased overnight supervision on resident education, decision‐making, and autonomy. J Hosp Med. 2012;7(8):606–610.
- , , , , . Approval and perceived impact of duty hour regulations: survey of pediatric program directors. Pediatrics. 2013;132(5):819–824.
- , , , et al. Anticipated consequences of the 2011 duty hours standards: views of internal medicine and surgery program directors. Acad Med. 2012;87(7):895–903.
- , , , , . Training on the clock: family medicine residency directors' responses to resident duty hours reform. Acad Med. 2006;81(12):1032–1037.
- , , , et al. Duty hour recommendations and implications for meeting the ACGME core competencies: views of residency directors. Mayo Clin Proc. 2011;86(3):185–191.
- , , , et al. Does surgeon workload per day affect outcomes after pulmonary lobectomies? Ann Thorac Surg. 2012;94(3):966–973.
- , , , . Impact of attending physician workload on patient care: a survey of hospitalists. JAMA Intern Med. 2013;173(5):375–377.
- , , , et al. No time for teaching? Inpatient attending physicians' workload and teaching before and after the implementation of the 2003 duty hours regulations. Acad Med. 2013;88(9):1293–1298.
- Accreditation Council for Graduate Medical Education. Clinical Learning Environment Review (CLER) Program. Available at: http://www.acgme.org/acgmeweb/tabid/436/ProgramandInstitutionalAccreditation/NextAccreditationSystem/ClinicalLearningEnvironmentReviewProgram.aspx. Accessed April 27, 2015.
- Accreditation Council for Graduate Medical Education. Frequently Asked Questions: A ACGME common duty hour requirements. Available at: https://www.acgme.org/acgmeweb/Portals/0/PDFs/dh‐faqs 2011.pdf. Accessed April 27, 2015.
- , , , , . Effect of hospitalist workload on the quality and efficiency of care. JAMA Intern Med. 2014;174(5):786–793.
- University HealthSystem Consortium. UHC clinical database/resource manager for Mayo Clinic. Available at: http://www.uhc.edu. Data accessed August 25, 2011.
- , . The interpersonal, cognitive and efficiency domains of clinical teaching: construct validity of a multi‐dimensional scale. Med Educ. 2005;39(12):1221–1229.
- , , . Factor instability of clinical teaching assessment scores among general internists and cardiologists. Med Educ. 2006;40(12):1209–1216.
- , , , . Determining reliability of clinical assessment scores in real time. Teach Learn Med. 2009;21(3):188–194.
- , , , , , . Behaviors of highly professional resident physicians. JAMA. 2008;300(11):1326–1333.
- , , , et al. Service census caps and unit‐based admissions: resident workload, conference attendance, duty hour compliance, and patient safety. Mayo Clin Proc. 2012;87(4):320–327.
- Agency for Healthcare Research and Quality. Patient safety indicators technical specifications updates—Version 5.0, March 2015. Available at: http://www.qualityindicators.ahrq.gov/Modules/PSI_TechSpec.aspx. Accessed May 29, 2015.
- , , , , , . The impact of nonphysician clinicians: do they improve the quality and cost‐effectiveness of health care services? Med Care Res Rev. 2009;66(6 suppl):36S–89S.
- , , . Maximizing teaching on the wards: review and application of the One‐Minute Preceptor and SNAPPS models. J Hosp Med. 2015;10(2):125–130.
- , , , . Resident perceptions of the educational value of night float rotations. Teach Learn Med. 2010;22(3):196–201.
- , , , , , . An evaluation of internal medicine residency continuity clinic redesign to a 50/50 outpatient‐inpatient model. J Gen Intern Med. 2013;28(8):1014–1019.
- , , , et al. Revisiting the rotating call schedule in less than 80 hours per week. J Surg Educ. 2009;66(6):357–360.
- , . Excess length of stay, charges, and mortality attributable to medical injuries during hospitalization. JAMA. 2003;290(14):1868–1874.
- , , , . Agency for Healthcare Research and Quality patient safety indicators and mortality in surgical patients. Am Surg. 2014;80(8):801–804.
- , , , et al. Patient safety in the era of the 80‐hour workweek. J Surg Educ. 2014;71(4):551–559.
- , , , . Impact of resident well‐being and empathy on assessments of faculty physicians. J Gen Intern Med. 2010;25(1):52–56.
- , , , et al. Stress management training for surgeons‐a randomized, controlled, intervention study. Ann Surg. 2011;253(3):488–494.
- , , . The future of residents' education in internal medicine. Am J Med. 2004;116(9):648–650.
- , , , , . Redesigning residency education in internal medicine: a position paper from the Association of Program Directors in Internal Medicine. Ann Intern Med. 2006;144(12):920–926.
- , , . Residency training in the modern era: the pipe dream of less time to learn more, care better, and be more professional. Arch Intern Med. 2005;165(22):2561–2562.
- , , , et al. Trends in Hospitalizations Among Medicare Survivors of Aortic Valve Replacement in the United States From 1999 to 2010. Ann Thorac Surg. 2015;99(2):509–517.
- , , . Restructuring an inpatient resident service to improve outcomes for residents, students, and patients. Acad Med. 2011;86(12):1500–1507.
- , , , . Clinical documentation in the 21st century: executive summary of a policy position paper from the American College of Physicians. Ann Intern Med. 2015;162(4):301–303.
- , . Effect of ACGME duty hours on attending physician teaching and satisfaction. Arch Intern Med. 2008;168(11):1226–1228.
- , , , , . Identifying potential predictors of a safe attending physician workload: a survey of hospitalists. J Hosp Med. 2013;8(11):644–646.
- , , . The clinical learning environment: the foundation of graduate medical education. JAMA. 2013;309(16):1687–1688.
- , , , , , . Better rested, but more stressed? Evidence of the effects of resident work hour restrictions. Acad Pediatr. 2012;12(4):335–343.
- , , , . Multifaceted longitudinal study of surgical resident education, quality of life, and patient care before and after July 2011. J Surg Educ. 2013;70(6):769–776.
- , , . Impact of the new 16‐hour duty period on pediatric interns' neonatal education. Clin Pediatr (Phila). 2014;53(1):51–59.
- , , , , , . Relationship between resident workload and self‐perceived learning on inpatient medicine wards: a longitudinal study. BMC Med Educ. 2006;6:35.
- , , , . Perceptions of educational experience and inpatient workload among pediatric residents. Hosp Pediatr. 2013;3(3):276–284.
- , , , et al. Association of workload of on‐call medical interns with on‐call sleep duration, shift duration, and participation in educational activities. JAMA. 2008;300(10):1146–1153.
- , , , , , . Effects of increased overnight supervision on resident education, decision‐making, and autonomy. J Hosp Med. 2012;7(8):606–610.
- , , , , . Approval and perceived impact of duty hour regulations: survey of pediatric program directors. Pediatrics. 2013;132(5):819–824.
- , , , et al. Anticipated consequences of the 2011 duty hours standards: views of internal medicine and surgery program directors. Acad Med. 2012;87(7):895–903.
- , , , , . Training on the clock: family medicine residency directors' responses to resident duty hours reform. Acad Med. 2006;81(12):1032–1037.
- , , , et al. Duty hour recommendations and implications for meeting the ACGME core competencies: views of residency directors. Mayo Clin Proc. 2011;86(3):185–191.
- , , , et al. Does surgeon workload per day affect outcomes after pulmonary lobectomies? Ann Thorac Surg. 2012;94(3):966–973.
- , , , . Impact of attending physician workload on patient care: a survey of hospitalists. JAMA Intern Med. 2013;173(5):375–377.
- , , , et al. No time for teaching? Inpatient attending physicians' workload and teaching before and after the implementation of the 2003 duty hours regulations. Acad Med. 2013;88(9):1293–1298.
- Accreditation Council for Graduate Medical Education. Clinical Learning Environment Review (CLER) Program. Available at: http://www.acgme.org/acgmeweb/tabid/436/ProgramandInstitutionalAccreditation/NextAccreditationSystem/ClinicalLearningEnvironmentReviewProgram.aspx. Accessed April 27, 2015.
- Accreditation Council for Graduate Medical Education. Frequently Asked Questions: A ACGME common duty hour requirements. Available at: https://www.acgme.org/acgmeweb/Portals/0/PDFs/dh‐faqs 2011.pdf. Accessed April 27, 2015.
- , , , , . Effect of hospitalist workload on the quality and efficiency of care. JAMA Intern Med. 2014;174(5):786–793.
- University HealthSystem Consortium. UHC clinical database/resource manager for Mayo Clinic. Available at: http://www.uhc.edu. Data accessed August 25, 2011.
- , . The interpersonal, cognitive and efficiency domains of clinical teaching: construct validity of a multi‐dimensional scale. Med Educ. 2005;39(12):1221–1229.
- , , . Factor instability of clinical teaching assessment scores among general internists and cardiologists. Med Educ. 2006;40(12):1209–1216.
- , , , . Determining reliability of clinical assessment scores in real time. Teach Learn Med. 2009;21(3):188–194.
- , , , , , . Behaviors of highly professional resident physicians. JAMA. 2008;300(11):1326–1333.
- , , , et al. Service census caps and unit‐based admissions: resident workload, conference attendance, duty hour compliance, and patient safety. Mayo Clin Proc. 2012;87(4):320–327.
- Agency for Healthcare Research and Quality. Patient safety indicators technical specifications updates—Version 5.0, March 2015. Available at: http://www.qualityindicators.ahrq.gov/Modules/PSI_TechSpec.aspx. Accessed May 29, 2015.
- , , , , , . The impact of nonphysician clinicians: do they improve the quality and cost‐effectiveness of health care services? Med Care Res Rev. 2009;66(6 suppl):36S–89S.
- , , . Maximizing teaching on the wards: review and application of the One‐Minute Preceptor and SNAPPS models. J Hosp Med. 2015;10(2):125–130.
- , , , . Resident perceptions of the educational value of night float rotations. Teach Learn Med. 2010;22(3):196–201.
- , , , , , . An evaluation of internal medicine residency continuity clinic redesign to a 50/50 outpatient‐inpatient model. J Gen Intern Med. 2013;28(8):1014–1019.
- , , , et al. Revisiting the rotating call schedule in less than 80 hours per week. J Surg Educ. 2009;66(6):357–360.
- , . Excess length of stay, charges, and mortality attributable to medical injuries during hospitalization. JAMA. 2003;290(14):1868–1874.
- , , , . Agency for Healthcare Research and Quality patient safety indicators and mortality in surgical patients. Am Surg. 2014;80(8):801–804.
- , , , et al. Patient safety in the era of the 80‐hour workweek. J Surg Educ. 2014;71(4):551–559.
- , , , . Impact of resident well‐being and empathy on assessments of faculty physicians. J Gen Intern Med. 2010;25(1):52–56.
- , , , et al. Stress management training for surgeons‐a randomized, controlled, intervention study. Ann Surg. 2011;253(3):488–494.
© 2016 Society of Hospital Medicine
Association Between DCBN and LOS
Slow hospital throughputthe process whereby a patient is admitted, placed in a room, and eventually dischargedcan worsen outcomes if admitted patients are boarded in emergency rooms or postanesthesia units.[1] One potential method to improve throughput is to discharge patients earlier in the day,[2] freeing up available beds and conceivably reducing hospital length of stay (LOS).
To quantify throughput, hospitals are beginning to measure the proportion of patients discharged before noon (DCBN). One study, looking at discharges on a single medical floor in an urban academic medical center, suggested that increasing the percentage of patients discharged by noon decreased observed‐to‐expected LOS in hospitalized medicine patients,[3] and a follow‐up study demonstrated that it was associated with admissions from the emergency department occurring earlier in the day.[4] However, these studies did not adjust for changes in case mix index (CMI) and other patient‐level characteristics that may also have affected these outcomes. Concerns persist that more efforts to discharge patients by noon could inadvertently increase LOS if staff chose to keep patients overnight for an early discharge the following day.
We undertook a retrospective analysis of data from patients discharged from a large academic medical center where an institution‐wide emphasis was placed on discharging more patients by noon. Using these data, we examined the association between discharges before noon and LOS in medical and surgical inpatients.
METHODS
Site and Subjects
Our study was based at the University of California, San Francisco (UCSF) Medical Center, a 400‐bed academic hospital located in San Francisco, California. We examined adult medical and surgical discharges from July 2012 through April 2015. Patients who stayed less than 24 hours or more than 20 days were excluded. Discharges from the hospital medicine service and the following surgical services were included in the analysis: cardiac surgery, colorectal surgery, cardiothoracic surgery, general surgery, gynecologic oncology, gynecology, neurosurgery, orthopedics, otolaryngology, head and neck surgery, plastic surgery, thoracic surgery, urology, and vascular surgery. No exclusions were made based on patient status (eg, observation vs inpatient). UCSF's institutional review board approved our study.
During the time of our study, discharges before noon time became an institutional priority. To this end, rates of DCBN were tracked using retrospective data, and various units undertook efforts such as informal afternoon meetings to prompt planning for the next morning's discharges. These efforts did not differentially affect medical or surgical units or emergent or nonemergent admissions, and no financial incentives or other changes in workflow were in place to increase DCBN rates.
Data Sources
We used the cost accounting system at UCSF (Enterprise Performance System Inc. [EPSI], Chicago, IL) to collect demographic information about each patient, including age, sex, primary race, and primary ethnicity. This system was also used to collect characteristics of each hospitalization including LOS (calculated from admission date time and discharge date time), hospital service at discharge, the discharge attending, discharge disposition of the patient, and the CMI, a marker of the severity of illness of the patient during that hospitalization. EPSI was also used to collect data on the admission type of all patients, either emergent, urgent, or routine, and the insurance status of the patient during that hospitalization.
Data on time of discharge were entered by the discharging nurse or unit assistant to reflect the time the patient left the hospital. Using these data, we defined a before‐noon discharge as one taking place between 8:00 am and 12:00 pm.
Statistical Analysis
Wilcoxon rank sum test and 2 statistics were used to compare baseline characteristics of hospitalizations of patients discharged before and after noon.
We used generalized linear models to assess the association of a discharge before noon on the LOS with gamma models. We accounted for clustering of discharge attendings using generalized estimating equations with exchangeable working correlation and robust standard errors. After the initial unadjusted analyses, covariates were included in the adjusted analysis if they were associated with an LOS at P < 0.05 or reasons of face validity. These variables are shown in Table 1. Because an effort to increase the discharges before noon was started in the 2014 academic year, we added an interaction term between the date of discharge and whether a discharge occurred before noon. The interaction term was included by dividing the study period into time periods corresponding to sequential 6‐month intervals. A new variable was defined by a categorical variable that indicated in which of these time periods a discharge occurred.
| Discharged Before Noon | Discharged After Noon | P Value | |
|---|---|---|---|
| |||
| Median LOS (IQR) | 3.4 (2.25.9) | 3.7 (2.36.3) | <0.0005 |
| Median CMI (IQR) | 1.8 (1.12.4) | 1.7 (1.12.5) | 0.006 |
| Service type, N (%) | |||
| Hospital medicine | 1,919 (29.6) | 11,290 (35.4) | |
| Surgical services | 4,565 (70.4) | 20,591 (64.6) | <0.0005 |
| Discharged before noon, N (%) | 6,484 (16.9) | 31,881 (83.1) | |
| Discharged on weekend, N (%) | |||
| Yes | 1,543 (23.8) | 7,411 (23.3) | |
| No | 4,941 (76.2) | 24,470 (76.8) | 0.34 |
| Discharge disposition, N (%) | |||
| Home with home health | 748 (11.5) | 5,774 (18.1) | |
| Home without home health | 3,997 (61.6) | 17,862 (56.0) | |
| SNF | 837 (12.9) | 3,082 (9.7) | |
| Other | 902 (13.9) | 5,163 (16.2) | <0.0005 |
| 6‐month interval, N (%) | |||
| JulyDecember 2012 | 993 (15.3) | 5,596 (17.6) | |
| JanuaryJune 2013 | 980 (15.1) | 5,721 (17.9) | |
| JulyDecember 2013 | 1,088 (16.8) | 5,690 (17.9) | |
| JanuaryJune 2014 | 1,288 (19.9) | 5,441 (17.1) | |
| JulyDecember 2014 | 1,275 (19.7) | 5,656 (17.7) | |
| JanuaryApril 2015 | 860 (13.3) | 3,777 (11.9) | <0.0005 |
| Age category, N (%) | |||
| 1864 years | 4,177 (64.4) | 20,044 (62.9) | |
| 65+ years | 2,307 (35.6) | 11,837 (37.1) | 0.02 |
| Male, N (%) | 3,274 (50.5) | 15,596 (48.9) | |
| Female, N (%) | 3,210 (49.5) | 16,284 (51.1) | 0.06 |
| Race, N (%) | |||
| White or Caucasian | 4,133 (63.7) | 18,798 (59.0) | |
| African American | 518 (8.0) | 3,020 (9.5) | |
| Asian | 703 (10.8) | 4,052 (12.7) | |
| Other | 1,130 (17.4) | 6,011 (18.9) | <0.0005 |
| Ethnicity, N (%) | |||
| Hispanic or Latino | 691 (10.7) | 3,713 (11.7) | |
| Not Hispanic or Latino | 5,597 (86.3) | 27,209 (85.4) | |
| Unknown/declined | 196 (3.0) | 959 (3.0) | 0.07 |
| Admission type, N (%) | |||
| Elective | 3,494 (53.9) | 13,881 (43.5) | |
| Emergency | 2,047 (31.6) | 12,145 (38.1) | |
| Urgent | 889 (13.7) | 5,459 (17.1) | |
| Other | 54 (0.8) | 396 (1.2) | <0.0005 |
| Payor class, N (%) | |||
| Medicare | 2,648 (40.8) | 13,808 (43.3) | |
| Medi‐Cal | 1,060 (16.4) | 5,913 (18.6) | |
| Commercial | 2,633 (40.6) | 11,242 (35.3) | |
| Other | 143 (2.2) | 918 (2.9) | <0.0005 |
We conducted a sensitivity analysis using propensity scores. The propensity score was based on demographic and clinical variables (as listed in Table 1) that exhibited P < 0.2 in bivariate analysis between the variable and being discharged before noon. We then used the propensity score as a covariate in a generalized linear model of the LOS with a gamma distribution and with generalized estimating equations as described above.
Finally, we performed prespecified secondary subset analyses of patients admitted emergently and nonemergently.
Statistical modeling and analysis was completed using Stata version 13 (StataCorp, College Station, TX).
RESULTS
Patient Demographics and Discharge Before Noon
Our study population comprised 27,983 patients for a total of 38,365 hospitalizations with a median LOS of 3.7 days. We observed 6484 discharges before noon (16.9%) and 31,881 discharges after noon (83.1%). The characteristics of the hospitalizations are shown in Table 1.
Patients who were discharged before noon tended to be younger, white, and discharged with a disposition to home without home health. The median CMI was slightly higher in discharges before noon (1.81, P = 0.006), and elective admissions were more likely than emergent to be discharged before noon (53.9% vs 31.6%, P < 0.0005).
Multivariable Analysis
A discharge before noon was associated with a 4.3% increase in LOS (adjusted odds ratio [OR]: 1.043, 95% confidence interval [CI]: 1.003‐1.086), adjusting for CMI, the service type, discharge on the weekend, discharge disposition, age, sex, ethnicity, race, urgency of admission, payor class, and a full interaction with the date of discharge (in 6‐month intervals). In preplanned subset analyses, the association between longer LOS and DCBN was more pronounced in patients admitted emergently (adjusted OR: 1.14, 95% CI: 1.033‐1.249) and less pronounced for patients not admitted emergently (adjusted OR: 1.03, 95% CI: 0.988‐1.074), although the latter did not meet statistical significance. In patients admitted emergently, this corresponds to approximately a 12‐hour increase in LOS. The interaction term of discharge date and DCBN was significant in the model. In further subset analyses, the association between longer LOS and DCBN was more pronounced in medicine patients (adjusted OR: 1.116, 95% CI: 1.014‐1.228) than in surgical patients (adjusted OR: 1.030, 95% CI: 0.989‐1.074), although the relationship in surgical patients did not meet statistical significance.
We also undertook sensitivity analyses utilizing propensity scores as a covariate in our base multivariable models. Results from these analyses did not differ from the base models and are not presented here. Results also did not differ when comparing discharges before and after the initiation of an attending only service.
DISCUSSION AND CONCLUSION
In our retrospective study of patients discharged from an academic medical center, discharge before noon was associated with a longer LOS, with the effect more pronounced in patients admitted emergently in the hospital. Our results suggest that efforts to discharge patients earlier in the day may have varying degrees of success depending on patient characteristics. Conceivably, elective admissions recover according to predictable plans, allowing for discharges earlier in the day. In contrast, patients discharged from emergent hospitalizations may have ongoing evolution of their care plan, making plans for discharging before noon more challenging.
Our results differ from a previous study,[3] which suggested that increasing the proportion of before‐noon discharges was associated with a fall in observed‐to‐expected LOS. However, observational studies of DCBN are challenging, because the association between early discharge and LOS is potentially bidirectional. One interpretation, for example, is that patients were kept longer in order to be discharged by noon the following day, which for the subgroups of patients admitted emergently corresponded to a roughly 12‐hour increase in LOS. However, it is also plausible that patients who stayed longer also had more time to plan for an early discharge. In either scenario, the ability of managers to utilize LOS as a key metric of throughput efforts may be flawed, and suggests that alternatives (eg, number of patients waiting for beds off unit) may be a more reasonable measure of throughput. Our results have several limitations. As in any observational study, our results are vulnerable to biases from unmeasured covariates that confound the analysis. We caution that a causal relationship between a discharge before noon and LOS cannot be determined from the nature of the study. Our results are also limited in that we were unable to adjust for day‐to‐day hospital capacity and other variables that affect LOS including caregiver and transportation availability, bed capacity at receiving care facilities, and patient consent to discharge. Finally, as a single‐site study, our findings may not be applicable to nonacademic settings.
In conclusion, our observational study discerned an association between discharging patients before noon and longer LOS. We believe our findings suggest a rationale for alternate approaches to measuring an early discharge program's effectiveness, namely, that the evaluation of the success of an early discharge initiative should consider multiple evaluation metrics including the effect on emergency department wait times, intensive care unit or postanesthesia transitions, and on patient reported experiences of care transitions.
Disclosures
Andrew Auerbach, MD, is supported by a K24 grant from the National Heart, Lung, and Blood Institute: K24HL098372. The authors report no conflicts of interest.
- , , , et al. The effect of emergency department crowding on clinically oriented outcomes. Acad Emerg Med. 2009;16(1):1–10.
- Centers for Medicare 2013.
- , , , et al. Discharge before noon: an achievable hospital goal. J Hosp Med. 2014;9(4):210–214.
- , , , et al. Discharge before noon: effect on throughput and sustainability. J Hosp Med. 2015;10(10):664–669.
Slow hospital throughputthe process whereby a patient is admitted, placed in a room, and eventually dischargedcan worsen outcomes if admitted patients are boarded in emergency rooms or postanesthesia units.[1] One potential method to improve throughput is to discharge patients earlier in the day,[2] freeing up available beds and conceivably reducing hospital length of stay (LOS).
To quantify throughput, hospitals are beginning to measure the proportion of patients discharged before noon (DCBN). One study, looking at discharges on a single medical floor in an urban academic medical center, suggested that increasing the percentage of patients discharged by noon decreased observed‐to‐expected LOS in hospitalized medicine patients,[3] and a follow‐up study demonstrated that it was associated with admissions from the emergency department occurring earlier in the day.[4] However, these studies did not adjust for changes in case mix index (CMI) and other patient‐level characteristics that may also have affected these outcomes. Concerns persist that more efforts to discharge patients by noon could inadvertently increase LOS if staff chose to keep patients overnight for an early discharge the following day.
We undertook a retrospective analysis of data from patients discharged from a large academic medical center where an institution‐wide emphasis was placed on discharging more patients by noon. Using these data, we examined the association between discharges before noon and LOS in medical and surgical inpatients.
METHODS
Site and Subjects
Our study was based at the University of California, San Francisco (UCSF) Medical Center, a 400‐bed academic hospital located in San Francisco, California. We examined adult medical and surgical discharges from July 2012 through April 2015. Patients who stayed less than 24 hours or more than 20 days were excluded. Discharges from the hospital medicine service and the following surgical services were included in the analysis: cardiac surgery, colorectal surgery, cardiothoracic surgery, general surgery, gynecologic oncology, gynecology, neurosurgery, orthopedics, otolaryngology, head and neck surgery, plastic surgery, thoracic surgery, urology, and vascular surgery. No exclusions were made based on patient status (eg, observation vs inpatient). UCSF's institutional review board approved our study.
During the time of our study, discharges before noon time became an institutional priority. To this end, rates of DCBN were tracked using retrospective data, and various units undertook efforts such as informal afternoon meetings to prompt planning for the next morning's discharges. These efforts did not differentially affect medical or surgical units or emergent or nonemergent admissions, and no financial incentives or other changes in workflow were in place to increase DCBN rates.
Data Sources
We used the cost accounting system at UCSF (Enterprise Performance System Inc. [EPSI], Chicago, IL) to collect demographic information about each patient, including age, sex, primary race, and primary ethnicity. This system was also used to collect characteristics of each hospitalization including LOS (calculated from admission date time and discharge date time), hospital service at discharge, the discharge attending, discharge disposition of the patient, and the CMI, a marker of the severity of illness of the patient during that hospitalization. EPSI was also used to collect data on the admission type of all patients, either emergent, urgent, or routine, and the insurance status of the patient during that hospitalization.
Data on time of discharge were entered by the discharging nurse or unit assistant to reflect the time the patient left the hospital. Using these data, we defined a before‐noon discharge as one taking place between 8:00 am and 12:00 pm.
Statistical Analysis
Wilcoxon rank sum test and 2 statistics were used to compare baseline characteristics of hospitalizations of patients discharged before and after noon.
We used generalized linear models to assess the association of a discharge before noon on the LOS with gamma models. We accounted for clustering of discharge attendings using generalized estimating equations with exchangeable working correlation and robust standard errors. After the initial unadjusted analyses, covariates were included in the adjusted analysis if they were associated with an LOS at P < 0.05 or reasons of face validity. These variables are shown in Table 1. Because an effort to increase the discharges before noon was started in the 2014 academic year, we added an interaction term between the date of discharge and whether a discharge occurred before noon. The interaction term was included by dividing the study period into time periods corresponding to sequential 6‐month intervals. A new variable was defined by a categorical variable that indicated in which of these time periods a discharge occurred.
| Discharged Before Noon | Discharged After Noon | P Value | |
|---|---|---|---|
| |||
| Median LOS (IQR) | 3.4 (2.25.9) | 3.7 (2.36.3) | <0.0005 |
| Median CMI (IQR) | 1.8 (1.12.4) | 1.7 (1.12.5) | 0.006 |
| Service type, N (%) | |||
| Hospital medicine | 1,919 (29.6) | 11,290 (35.4) | |
| Surgical services | 4,565 (70.4) | 20,591 (64.6) | <0.0005 |
| Discharged before noon, N (%) | 6,484 (16.9) | 31,881 (83.1) | |
| Discharged on weekend, N (%) | |||
| Yes | 1,543 (23.8) | 7,411 (23.3) | |
| No | 4,941 (76.2) | 24,470 (76.8) | 0.34 |
| Discharge disposition, N (%) | |||
| Home with home health | 748 (11.5) | 5,774 (18.1) | |
| Home without home health | 3,997 (61.6) | 17,862 (56.0) | |
| SNF | 837 (12.9) | 3,082 (9.7) | |
| Other | 902 (13.9) | 5,163 (16.2) | <0.0005 |
| 6‐month interval, N (%) | |||
| JulyDecember 2012 | 993 (15.3) | 5,596 (17.6) | |
| JanuaryJune 2013 | 980 (15.1) | 5,721 (17.9) | |
| JulyDecember 2013 | 1,088 (16.8) | 5,690 (17.9) | |
| JanuaryJune 2014 | 1,288 (19.9) | 5,441 (17.1) | |
| JulyDecember 2014 | 1,275 (19.7) | 5,656 (17.7) | |
| JanuaryApril 2015 | 860 (13.3) | 3,777 (11.9) | <0.0005 |
| Age category, N (%) | |||
| 1864 years | 4,177 (64.4) | 20,044 (62.9) | |
| 65+ years | 2,307 (35.6) | 11,837 (37.1) | 0.02 |
| Male, N (%) | 3,274 (50.5) | 15,596 (48.9) | |
| Female, N (%) | 3,210 (49.5) | 16,284 (51.1) | 0.06 |
| Race, N (%) | |||
| White or Caucasian | 4,133 (63.7) | 18,798 (59.0) | |
| African American | 518 (8.0) | 3,020 (9.5) | |
| Asian | 703 (10.8) | 4,052 (12.7) | |
| Other | 1,130 (17.4) | 6,011 (18.9) | <0.0005 |
| Ethnicity, N (%) | |||
| Hispanic or Latino | 691 (10.7) | 3,713 (11.7) | |
| Not Hispanic or Latino | 5,597 (86.3) | 27,209 (85.4) | |
| Unknown/declined | 196 (3.0) | 959 (3.0) | 0.07 |
| Admission type, N (%) | |||
| Elective | 3,494 (53.9) | 13,881 (43.5) | |
| Emergency | 2,047 (31.6) | 12,145 (38.1) | |
| Urgent | 889 (13.7) | 5,459 (17.1) | |
| Other | 54 (0.8) | 396 (1.2) | <0.0005 |
| Payor class, N (%) | |||
| Medicare | 2,648 (40.8) | 13,808 (43.3) | |
| Medi‐Cal | 1,060 (16.4) | 5,913 (18.6) | |
| Commercial | 2,633 (40.6) | 11,242 (35.3) | |
| Other | 143 (2.2) | 918 (2.9) | <0.0005 |
We conducted a sensitivity analysis using propensity scores. The propensity score was based on demographic and clinical variables (as listed in Table 1) that exhibited P < 0.2 in bivariate analysis between the variable and being discharged before noon. We then used the propensity score as a covariate in a generalized linear model of the LOS with a gamma distribution and with generalized estimating equations as described above.
Finally, we performed prespecified secondary subset analyses of patients admitted emergently and nonemergently.
Statistical modeling and analysis was completed using Stata version 13 (StataCorp, College Station, TX).
RESULTS
Patient Demographics and Discharge Before Noon
Our study population comprised 27,983 patients for a total of 38,365 hospitalizations with a median LOS of 3.7 days. We observed 6484 discharges before noon (16.9%) and 31,881 discharges after noon (83.1%). The characteristics of the hospitalizations are shown in Table 1.
Patients who were discharged before noon tended to be younger, white, and discharged with a disposition to home without home health. The median CMI was slightly higher in discharges before noon (1.81, P = 0.006), and elective admissions were more likely than emergent to be discharged before noon (53.9% vs 31.6%, P < 0.0005).
Multivariable Analysis
A discharge before noon was associated with a 4.3% increase in LOS (adjusted odds ratio [OR]: 1.043, 95% confidence interval [CI]: 1.003‐1.086), adjusting for CMI, the service type, discharge on the weekend, discharge disposition, age, sex, ethnicity, race, urgency of admission, payor class, and a full interaction with the date of discharge (in 6‐month intervals). In preplanned subset analyses, the association between longer LOS and DCBN was more pronounced in patients admitted emergently (adjusted OR: 1.14, 95% CI: 1.033‐1.249) and less pronounced for patients not admitted emergently (adjusted OR: 1.03, 95% CI: 0.988‐1.074), although the latter did not meet statistical significance. In patients admitted emergently, this corresponds to approximately a 12‐hour increase in LOS. The interaction term of discharge date and DCBN was significant in the model. In further subset analyses, the association between longer LOS and DCBN was more pronounced in medicine patients (adjusted OR: 1.116, 95% CI: 1.014‐1.228) than in surgical patients (adjusted OR: 1.030, 95% CI: 0.989‐1.074), although the relationship in surgical patients did not meet statistical significance.
We also undertook sensitivity analyses utilizing propensity scores as a covariate in our base multivariable models. Results from these analyses did not differ from the base models and are not presented here. Results also did not differ when comparing discharges before and after the initiation of an attending only service.
DISCUSSION AND CONCLUSION
In our retrospective study of patients discharged from an academic medical center, discharge before noon was associated with a longer LOS, with the effect more pronounced in patients admitted emergently in the hospital. Our results suggest that efforts to discharge patients earlier in the day may have varying degrees of success depending on patient characteristics. Conceivably, elective admissions recover according to predictable plans, allowing for discharges earlier in the day. In contrast, patients discharged from emergent hospitalizations may have ongoing evolution of their care plan, making plans for discharging before noon more challenging.
Our results differ from a previous study,[3] which suggested that increasing the proportion of before‐noon discharges was associated with a fall in observed‐to‐expected LOS. However, observational studies of DCBN are challenging, because the association between early discharge and LOS is potentially bidirectional. One interpretation, for example, is that patients were kept longer in order to be discharged by noon the following day, which for the subgroups of patients admitted emergently corresponded to a roughly 12‐hour increase in LOS. However, it is also plausible that patients who stayed longer also had more time to plan for an early discharge. In either scenario, the ability of managers to utilize LOS as a key metric of throughput efforts may be flawed, and suggests that alternatives (eg, number of patients waiting for beds off unit) may be a more reasonable measure of throughput. Our results have several limitations. As in any observational study, our results are vulnerable to biases from unmeasured covariates that confound the analysis. We caution that a causal relationship between a discharge before noon and LOS cannot be determined from the nature of the study. Our results are also limited in that we were unable to adjust for day‐to‐day hospital capacity and other variables that affect LOS including caregiver and transportation availability, bed capacity at receiving care facilities, and patient consent to discharge. Finally, as a single‐site study, our findings may not be applicable to nonacademic settings.
In conclusion, our observational study discerned an association between discharging patients before noon and longer LOS. We believe our findings suggest a rationale for alternate approaches to measuring an early discharge program's effectiveness, namely, that the evaluation of the success of an early discharge initiative should consider multiple evaluation metrics including the effect on emergency department wait times, intensive care unit or postanesthesia transitions, and on patient reported experiences of care transitions.
Disclosures
Andrew Auerbach, MD, is supported by a K24 grant from the National Heart, Lung, and Blood Institute: K24HL098372. The authors report no conflicts of interest.
Slow hospital throughputthe process whereby a patient is admitted, placed in a room, and eventually dischargedcan worsen outcomes if admitted patients are boarded in emergency rooms or postanesthesia units.[1] One potential method to improve throughput is to discharge patients earlier in the day,[2] freeing up available beds and conceivably reducing hospital length of stay (LOS).
To quantify throughput, hospitals are beginning to measure the proportion of patients discharged before noon (DCBN). One study, looking at discharges on a single medical floor in an urban academic medical center, suggested that increasing the percentage of patients discharged by noon decreased observed‐to‐expected LOS in hospitalized medicine patients,[3] and a follow‐up study demonstrated that it was associated with admissions from the emergency department occurring earlier in the day.[4] However, these studies did not adjust for changes in case mix index (CMI) and other patient‐level characteristics that may also have affected these outcomes. Concerns persist that more efforts to discharge patients by noon could inadvertently increase LOS if staff chose to keep patients overnight for an early discharge the following day.
We undertook a retrospective analysis of data from patients discharged from a large academic medical center where an institution‐wide emphasis was placed on discharging more patients by noon. Using these data, we examined the association between discharges before noon and LOS in medical and surgical inpatients.
METHODS
Site and Subjects
Our study was based at the University of California, San Francisco (UCSF) Medical Center, a 400‐bed academic hospital located in San Francisco, California. We examined adult medical and surgical discharges from July 2012 through April 2015. Patients who stayed less than 24 hours or more than 20 days were excluded. Discharges from the hospital medicine service and the following surgical services were included in the analysis: cardiac surgery, colorectal surgery, cardiothoracic surgery, general surgery, gynecologic oncology, gynecology, neurosurgery, orthopedics, otolaryngology, head and neck surgery, plastic surgery, thoracic surgery, urology, and vascular surgery. No exclusions were made based on patient status (eg, observation vs inpatient). UCSF's institutional review board approved our study.
During the time of our study, discharges before noon time became an institutional priority. To this end, rates of DCBN were tracked using retrospective data, and various units undertook efforts such as informal afternoon meetings to prompt planning for the next morning's discharges. These efforts did not differentially affect medical or surgical units or emergent or nonemergent admissions, and no financial incentives or other changes in workflow were in place to increase DCBN rates.
Data Sources
We used the cost accounting system at UCSF (Enterprise Performance System Inc. [EPSI], Chicago, IL) to collect demographic information about each patient, including age, sex, primary race, and primary ethnicity. This system was also used to collect characteristics of each hospitalization including LOS (calculated from admission date time and discharge date time), hospital service at discharge, the discharge attending, discharge disposition of the patient, and the CMI, a marker of the severity of illness of the patient during that hospitalization. EPSI was also used to collect data on the admission type of all patients, either emergent, urgent, or routine, and the insurance status of the patient during that hospitalization.
Data on time of discharge were entered by the discharging nurse or unit assistant to reflect the time the patient left the hospital. Using these data, we defined a before‐noon discharge as one taking place between 8:00 am and 12:00 pm.
Statistical Analysis
Wilcoxon rank sum test and 2 statistics were used to compare baseline characteristics of hospitalizations of patients discharged before and after noon.
We used generalized linear models to assess the association of a discharge before noon on the LOS with gamma models. We accounted for clustering of discharge attendings using generalized estimating equations with exchangeable working correlation and robust standard errors. After the initial unadjusted analyses, covariates were included in the adjusted analysis if they were associated with an LOS at P < 0.05 or reasons of face validity. These variables are shown in Table 1. Because an effort to increase the discharges before noon was started in the 2014 academic year, we added an interaction term between the date of discharge and whether a discharge occurred before noon. The interaction term was included by dividing the study period into time periods corresponding to sequential 6‐month intervals. A new variable was defined by a categorical variable that indicated in which of these time periods a discharge occurred.
| Discharged Before Noon | Discharged After Noon | P Value | |
|---|---|---|---|
| |||
| Median LOS (IQR) | 3.4 (2.25.9) | 3.7 (2.36.3) | <0.0005 |
| Median CMI (IQR) | 1.8 (1.12.4) | 1.7 (1.12.5) | 0.006 |
| Service type, N (%) | |||
| Hospital medicine | 1,919 (29.6) | 11,290 (35.4) | |
| Surgical services | 4,565 (70.4) | 20,591 (64.6) | <0.0005 |
| Discharged before noon, N (%) | 6,484 (16.9) | 31,881 (83.1) | |
| Discharged on weekend, N (%) | |||
| Yes | 1,543 (23.8) | 7,411 (23.3) | |
| No | 4,941 (76.2) | 24,470 (76.8) | 0.34 |
| Discharge disposition, N (%) | |||
| Home with home health | 748 (11.5) | 5,774 (18.1) | |
| Home without home health | 3,997 (61.6) | 17,862 (56.0) | |
| SNF | 837 (12.9) | 3,082 (9.7) | |
| Other | 902 (13.9) | 5,163 (16.2) | <0.0005 |
| 6‐month interval, N (%) | |||
| JulyDecember 2012 | 993 (15.3) | 5,596 (17.6) | |
| JanuaryJune 2013 | 980 (15.1) | 5,721 (17.9) | |
| JulyDecember 2013 | 1,088 (16.8) | 5,690 (17.9) | |
| JanuaryJune 2014 | 1,288 (19.9) | 5,441 (17.1) | |
| JulyDecember 2014 | 1,275 (19.7) | 5,656 (17.7) | |
| JanuaryApril 2015 | 860 (13.3) | 3,777 (11.9) | <0.0005 |
| Age category, N (%) | |||
| 1864 years | 4,177 (64.4) | 20,044 (62.9) | |
| 65+ years | 2,307 (35.6) | 11,837 (37.1) | 0.02 |
| Male, N (%) | 3,274 (50.5) | 15,596 (48.9) | |
| Female, N (%) | 3,210 (49.5) | 16,284 (51.1) | 0.06 |
| Race, N (%) | |||
| White or Caucasian | 4,133 (63.7) | 18,798 (59.0) | |
| African American | 518 (8.0) | 3,020 (9.5) | |
| Asian | 703 (10.8) | 4,052 (12.7) | |
| Other | 1,130 (17.4) | 6,011 (18.9) | <0.0005 |
| Ethnicity, N (%) | |||
| Hispanic or Latino | 691 (10.7) | 3,713 (11.7) | |
| Not Hispanic or Latino | 5,597 (86.3) | 27,209 (85.4) | |
| Unknown/declined | 196 (3.0) | 959 (3.0) | 0.07 |
| Admission type, N (%) | |||
| Elective | 3,494 (53.9) | 13,881 (43.5) | |
| Emergency | 2,047 (31.6) | 12,145 (38.1) | |
| Urgent | 889 (13.7) | 5,459 (17.1) | |
| Other | 54 (0.8) | 396 (1.2) | <0.0005 |
| Payor class, N (%) | |||
| Medicare | 2,648 (40.8) | 13,808 (43.3) | |
| Medi‐Cal | 1,060 (16.4) | 5,913 (18.6) | |
| Commercial | 2,633 (40.6) | 11,242 (35.3) | |
| Other | 143 (2.2) | 918 (2.9) | <0.0005 |
We conducted a sensitivity analysis using propensity scores. The propensity score was based on demographic and clinical variables (as listed in Table 1) that exhibited P < 0.2 in bivariate analysis between the variable and being discharged before noon. We then used the propensity score as a covariate in a generalized linear model of the LOS with a gamma distribution and with generalized estimating equations as described above.
Finally, we performed prespecified secondary subset analyses of patients admitted emergently and nonemergently.
Statistical modeling and analysis was completed using Stata version 13 (StataCorp, College Station, TX).
RESULTS
Patient Demographics and Discharge Before Noon
Our study population comprised 27,983 patients for a total of 38,365 hospitalizations with a median LOS of 3.7 days. We observed 6484 discharges before noon (16.9%) and 31,881 discharges after noon (83.1%). The characteristics of the hospitalizations are shown in Table 1.
Patients who were discharged before noon tended to be younger, white, and discharged with a disposition to home without home health. The median CMI was slightly higher in discharges before noon (1.81, P = 0.006), and elective admissions were more likely than emergent to be discharged before noon (53.9% vs 31.6%, P < 0.0005).
Multivariable Analysis
A discharge before noon was associated with a 4.3% increase in LOS (adjusted odds ratio [OR]: 1.043, 95% confidence interval [CI]: 1.003‐1.086), adjusting for CMI, the service type, discharge on the weekend, discharge disposition, age, sex, ethnicity, race, urgency of admission, payor class, and a full interaction with the date of discharge (in 6‐month intervals). In preplanned subset analyses, the association between longer LOS and DCBN was more pronounced in patients admitted emergently (adjusted OR: 1.14, 95% CI: 1.033‐1.249) and less pronounced for patients not admitted emergently (adjusted OR: 1.03, 95% CI: 0.988‐1.074), although the latter did not meet statistical significance. In patients admitted emergently, this corresponds to approximately a 12‐hour increase in LOS. The interaction term of discharge date and DCBN was significant in the model. In further subset analyses, the association between longer LOS and DCBN was more pronounced in medicine patients (adjusted OR: 1.116, 95% CI: 1.014‐1.228) than in surgical patients (adjusted OR: 1.030, 95% CI: 0.989‐1.074), although the relationship in surgical patients did not meet statistical significance.
We also undertook sensitivity analyses utilizing propensity scores as a covariate in our base multivariable models. Results from these analyses did not differ from the base models and are not presented here. Results also did not differ when comparing discharges before and after the initiation of an attending only service.
DISCUSSION AND CONCLUSION
In our retrospective study of patients discharged from an academic medical center, discharge before noon was associated with a longer LOS, with the effect more pronounced in patients admitted emergently in the hospital. Our results suggest that efforts to discharge patients earlier in the day may have varying degrees of success depending on patient characteristics. Conceivably, elective admissions recover according to predictable plans, allowing for discharges earlier in the day. In contrast, patients discharged from emergent hospitalizations may have ongoing evolution of their care plan, making plans for discharging before noon more challenging.
Our results differ from a previous study,[3] which suggested that increasing the proportion of before‐noon discharges was associated with a fall in observed‐to‐expected LOS. However, observational studies of DCBN are challenging, because the association between early discharge and LOS is potentially bidirectional. One interpretation, for example, is that patients were kept longer in order to be discharged by noon the following day, which for the subgroups of patients admitted emergently corresponded to a roughly 12‐hour increase in LOS. However, it is also plausible that patients who stayed longer also had more time to plan for an early discharge. In either scenario, the ability of managers to utilize LOS as a key metric of throughput efforts may be flawed, and suggests that alternatives (eg, number of patients waiting for beds off unit) may be a more reasonable measure of throughput. Our results have several limitations. As in any observational study, our results are vulnerable to biases from unmeasured covariates that confound the analysis. We caution that a causal relationship between a discharge before noon and LOS cannot be determined from the nature of the study. Our results are also limited in that we were unable to adjust for day‐to‐day hospital capacity and other variables that affect LOS including caregiver and transportation availability, bed capacity at receiving care facilities, and patient consent to discharge. Finally, as a single‐site study, our findings may not be applicable to nonacademic settings.
In conclusion, our observational study discerned an association between discharging patients before noon and longer LOS. We believe our findings suggest a rationale for alternate approaches to measuring an early discharge program's effectiveness, namely, that the evaluation of the success of an early discharge initiative should consider multiple evaluation metrics including the effect on emergency department wait times, intensive care unit or postanesthesia transitions, and on patient reported experiences of care transitions.
Disclosures
Andrew Auerbach, MD, is supported by a K24 grant from the National Heart, Lung, and Blood Institute: K24HL098372. The authors report no conflicts of interest.
- , , , et al. The effect of emergency department crowding on clinically oriented outcomes. Acad Emerg Med. 2009;16(1):1–10.
- Centers for Medicare 2013.
- , , , et al. Discharge before noon: an achievable hospital goal. J Hosp Med. 2014;9(4):210–214.
- , , , et al. Discharge before noon: effect on throughput and sustainability. J Hosp Med. 2015;10(10):664–669.
- , , , et al. The effect of emergency department crowding on clinically oriented outcomes. Acad Emerg Med. 2009;16(1):1–10.
- Centers for Medicare 2013.
- , , , et al. Discharge before noon: an achievable hospital goal. J Hosp Med. 2014;9(4):210–214.
- , , , et al. Discharge before noon: effect on throughput and sustainability. J Hosp Med. 2015;10(10):664–669.
© 2015 Society of Hospital Medicine
Assessing Vascular Nursing Experience
Peripherally inserted central catheters (PICCs) are among the most prevalent of venous access devices in hospitalized patients.[1, 2] Although growing use of these devices reflects clinical advantages, such as a reduced risk of complications during insertion and durable venous access, use of PICCs is also likely related to the growth of vascular access nursing.[3, 4] A relatively new specialty, vascular access nurses obtain, maintain, and manage venous access in hospitalized patients.[4, 5] Depending on their scope of practice, these professionals are responsible not only for insertion of devices, such as peripheral intravenous catheters and PICCs, but also nontunneled central venous catheters and arterial catheters in some settings.[6]
Although a growing number of US hospitals have introduced vascular nursing teams,[7] little is known about the experience, practice, knowledge, and beliefs of vascular access nurses. This knowledge gap is relevant for hospitalists and hospital medicine as (1) vascular access nurses increasingly represent a key partner in the care of hospitalized patients; (2) the knowledge and practice of these individuals directly affects patient safety and clinical outcomes; and (3) understanding experience, practice, and beliefs of these specialists can help inform decision making and quality‐improvement efforts related to PICCs. As hospitalists increasingly order the placement of and care for patients with PICCs, they are also well suited to improve PICC practice.
Therefore, we conducted a survey of vascular access nurses employed by hospitals that participate in the Michigan Hospital Medicine Safety (HMS) Consortium, a Blue Cross Blue Shield of Michiganfunded collaborative quality initiative.[6] We aimed to understand experience, practice, knowledge, and beliefs related to PICC care and use.
METHODS
Study Setting and Participants
To quantify vascular nursing experience, practice, knowledge, and beliefs, we conducted a Web‐based survey of vascular nurses across 47 Michigan hospitals that participate in HMS. A statewide quality‐improvement initiative, HMS aims to prevent adverse events in hospitalized medical patients through the creation of a data registry and sharing of best practices. The setting and design of this multicenter initiative have been previously described.[8, 9] Although participation is voluntary, each hospital receives payment for participating in the consortium and for data collection. Because HMS has an ongoing initiative aimed at identifying and preventing PICC‐related complications, this study was particularly relevant for participating hospitals and nurses.
Each HMS site has a designated quality‐improvement lead, physician champion, and data abstractor. To coordinate distribution and dissemination of the survey, we contacted the quality‐improvement leads at each site and enquired whether their hospital employed vascular access nurses who placed PICCs. Because we were only interested in responses from vascular access nurses, HMS hospitals that did not have these providers or stated PICCs were placed by other specialists (eg, interventional radiology) were excluded. At eligible sites, we obtained the total number of vascular nurses employed so as to determine the number of eligible respondents. In this manner, a purposeful sample of vascular nurses at participating HMS hospitals was constituted.
Participation in the survey was solicited through hospital quality leads that either distributed an electronic survey link to vascular nurses at their facilities or sent us individual email addresses to contact them directly. A cover letter explaining the rationale and the purpose of the survey along with the survey link was then sent to respondents through either of these routes. The survey was administered at all HMS sites contemporaneously and kept open for a period of 5 weeks. During the 5‐week period, 2 e‐mail reminders were sent to encourage participation. As a token of appreciation, a $10 Amazon gift card was offered to those who took the survey.
Development and Validation of the Survey
We developed the survey instrument (which we call PICC1 as we hope to administer longitudinally to track changes over time) by first conducting a literature search to identify relevant evidence‐based guidelines and studies regarding vascular access nursing practices and experiences.[10, 11, 12, 13] In addition, we consulted and involved national and international leaders in vascular access nursing to ensure validity and representativeness of the questions posed. We were specifically interested in nursing background, hospital practices, types of PICCs used, use of various technologies, relationships with healthcare providers, and management of complications. To understand participant characteristics and quantify potential variation in responses, we collected basic participant data including demographics, years in practice, number of PICCs placed, leadership roles, and vascular access certification status. Based on clinical reasoning and existing studies,[14, 15] we hypothesized that responses regarding certain practices (ultrasound use, electrocardiography [ECG] guidance system use), management of complications, or perceptions regarding leadership might vary based on years of experience, number of PICCs placed, or certification status. We therefore examined these associations as prespecified subgroup analyses.
The initial survey instrument was pilot tested with vascular nurses outside of the sampling frame. Based on feedback from the pilot testers, the instrument was refined and edited to improve clarity of the questions. In addition, specific skip patterns and logic were programmed into the final survey to reduce respondent burden and allow participants to seamlessly bypass questions that were contingent on a prior response (eg, use of ECG to place PICCs would lead to a series of questions about ECG‐assisted placement only for those respondents who used the technology). This final version of the survey was tested by members of the study team (V.C., L.K., S.L.K.) and then posted to SurveyMonkey for dissemination.
Statistical Analysis
Descriptive statistics (percentage, n/N) were used to tabulate results. In accordance with our a priori hypothesis that variation to responses might be associated with respondent characteristics, responses to questions regarding insertion practice (eg, use of ultrasound, measurement of catheter:vein ratio, trimming of catheters) and approach to complications (eg, catheter occlusion, deep vein thrombosis [DVT] notification, and PICC removal in the setting of fever) were compared by respondent years in practice (dichotomized to <5 vs >5 years), volume of PICCs placed (<999 vs 1000), and certification status (yes/no). Bivariate comparisons were made using 2 or Fisher exact tests based on the number of responses in a cell as appropriate; 2‐sided with a P value <0.05 was considered statistically significant. All statistical analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC).
Ethical and Regulatory Oversight
Because our study sought to describe existing practice without collecting any individual or facility level identifiable information, the project received a Not Regulated status by the University of Michigan Medical School Institutional Review Board (HUM00088351).
RESULTS
Of 172 vascular nurses who received invitations, 140 completed the survey for a response rate of 81%. Respondents reported working in not‐for‐profit hospitals (36%), academic medical centers (29%), and for‐profit hospitals (21%). Although multiple providers (eg, interventional radiology staff and providers, physicians) placed PICCs, 95% of those surveyed reported that they placed the majority of the PICCs at their institutions. Although most respondents placed PICCs in adult patients (86%), a few also placed PICCs in pediatric populations (17%). Vascular nursing programs were largely housed in their own department, but some reported to general nursing or subspecialties such as interventional radiology, cardiology, and critical care. Most respondents indicated their facilities had written policies regarding standard insertion and care practices (87% and 95%, respectively), but only 30% had policies regarding the necessity or appropriateness of PICCs.
Experience among respondents was variable: approximately a third had placed PICCs for <5 years (28.6%), whereas 58% reported placing PICCs for 5 years Correspondingly, 26% reported having placed 100 to 500 PICCs, whereas 34% had placed 1000 or more PICCs. Only 23% of those surveyed held a dedicated vascular access certification, such as board certified in vascular access or certified registered nurse infusion, whereas 16% indicated that they served as the vascular access lead nurse for their facility. Following placement, 94% of respondents reported that their facilities tracked the number of PICCs inserted, but only 40% indicated that dwell times of devices were also recorded. Only 30% of nurses reported that their hospitals had a written policy to evaluate PICC necessity or appropriateness following placement (Table 1).
| No.* | % | |
|---|---|---|
| ||
| Participant characteristics | ||
| For how many years have you been inserting PICCs? | ||
| <5 years | 40 | 28.6% |
| 5 years | 81 | 57.9% |
| Missing | ||
| In which of the following populations do you insert PICCs? | ||
| Adult patients | 121 | 86.4% |
| Pediatric patients | 24 | 17.1% |
| Neonatal patients | 1 | 0.7% |
| In which of the following locations do you place PICCs? (Select all that apply.) | ||
| Adult medical ward | 115 | 82.1% |
| General adult surgical ward | 110 | 78.6% |
| General pediatric medical ward | 34 | 24.3% |
| General pediatric surgical ward | 24 | 17.1% |
| Adult intensive care unit | 114 | 81.4% |
| Pediatric intensive care unit | 19 | 13.6% |
| Neonatal intensive care unit | 3 | 2.1% |
| Other intensive care unit | 59 | 42.1% |
| Outpatient clinic or emergency department | 17 | 12.1% |
| Other | 10 | 7.1% |
| Approximately how many PICCs may you have placed in your career? | ||
| 099 | 15 | 10.7% |
| 100499 | 36 | 25.7% |
| 500999 | 23 | 16.4% |
| 1,000 | 47 | 33.6% |
| Are you the vascular access lead nurse for your facility or organization? | ||
| Yes | 22 | 15.7% |
| No | 98 | 70.0% |
| Do you currently hold a dedicated vascular access certification (BC‐VA, CRNI, etc.)? | ||
| Yes | 32 | 22.9% |
| No | 89 | 63.6% |
| Facility characteristics | ||
| Which of the following best describes your primary work location? | ||
| Academic medical center | 41 | 29.3% |
| For‐profit community‐based hospital or medical center | 30 | 21.4% |
| Not‐for‐profit community‐based hospital or medical center | 50 | 35.7% |
| Who inserts the most PICCs in your facility? | ||
| Vascular access nurses | 133 | 95.0% |
| Interventional radiology or other providers | 7 | 5.0% |
| In which department is vascular access nursing located? | ||
| Vascular nursing | 76 | 54.3% |
| General nursing | 38 | 27.1% |
| Interventional radiology | 15 | 10.7% |
| Other | 11 | 7.9% |
| Using your best guess, how many PICCs do you think your facility inserts each month? | ||
| <25 | 5 | 3.6% |
| 2549 | 13 | 9.3% |
| 50100 | 39 | 27.9% |
| >100 | 78 | 55.7% |
| Unknown | 2 | 1.4% |
| How many vascular access nurses are employed by your facility? | ||
| <4 | 14 | 10.0% |
| 46 | 33 | 23.6% |
| 79 | 15 | 10.7% |
| 1015 | 25 | 17.9% |
| >15 | 53 | 37.9% |
| Does your facility track the number of PICCs placed? | ||
| Yes | 132 | 94.3% |
| No | 5 | 3.6% |
| Unknown | 3 | 2.1% |
| Does your facility track the duration or dwell time of PICCs? | ||
| Yes | 56 | 40.0% |
| No | 60 | 42.9% |
| Unknown | 24 | 17.1% |
| Does your facility have a written policy regarding standard PICC insertion practices? | ||
| Yes | 122 | 87.1% |
| No | 8 | 5.7% |
| Unknown | 7 | 5.0% |
| Does your facility have a written policy regarding standard PICC care and maintenance? | ||
| Yes | 133 | 95.0% |
| No | 3 | 2.1% |
| Unknown | 1 | 0.7% |
| Does your facility have a written process to review the necessity or appropriateness of a PICC? | ||
| Yes | 42 | 30.0% |
| No | 63 | 45.0% |
| Unknown | 20 | 14.3% |
The most commonly reported indications for PICC placement included intravenous antibiotics at discharge, difficult venous access, and placement for chemotherapy in patients with cancer. Forty‐six percent of nurses indicated they had placed a PICC in a patient receiving some form of dialysis in the past several months; however, 91% of these respondents reported receiving approval from nephrology prior to placement in these patients. Although almost all nurses (91%) used ultrasound to find a suitable vein for PICC placement, a smaller percentage used ultrasound to estimate the catheter‐to‐vein ratio to prevent thrombosis (79%), and only a few (14%) documented this figure in the medical record. Three‐quarters of those surveyed (76%) indicated they used ECG‐based systems to position PICC tips at the cavoatrial junction to prevent thrombosis. Of those who used this technology, 36% still obtained chest x‐rays to verify the position of the PICC tip. According to 84% of respondents, flushing of PICCs was performed mainly by bedside nurses, whereas scheduled weekly dressing changes were most often performed by vascular access nurses (Table 2).
| Question | No. | % |
|---|---|---|
| ||
| Do you use ultrasound to find a suitable vein prior to PICC insertion? | ||
| Yes | 128 | 91.4% |
| No | 0 | 0.0% |
| Do you use ultrasound to estimate the catheter‐to‐vein ratio prior to PICC insertion? | ||
| Yes | 110 | 78.6% |
| No | 18 | 12.9% |
| When using ultrasound, do you document the catheter‐to‐vein ratio in the PICC insertion note? | ||
| Yes | 20 | 14.3% |
| No | 89 | 63.6% |
| Do you use ECG guidance‐assisted systems to place PICCs? | ||
| Yes | 106 | 75.7% |
| No | 21 | 15.0% |
| If using ECG guidance, do you still routinely obtain a chest x‐ray to verify PICC tip position after placing the PICC using ECG guidance? | ||
| Yes | 38 | 27.1% |
| No | 68 | 48.6% |
| Who is primarily responsible for administering and adhering to a flushing protocol after PICC insertion at your facility? | ||
| Bedside nurses | 118 | 83.6% |
| Patients | 1 | 0.7% |
| Vascular access nurses | 8 | 5.7% |
| Which of the following agents are most often used to flush PICCs? | ||
| Both heparin and normal saline flushes | 61 | 43.6% |
| Normal saline only | 63 | 45.0% |
| Heparin only | 3 | 2.1% |
| Who is responsible for scheduled weekly dressing changes for PICCs? | ||
| Vascular access nurses | 110 | 78.6% |
| Bedside nurses | 14 | 10.0% |
| Other (eg, IR staff, ICU staff) | 3 | 2.1% |
| In the past few months, have you placed a PICC in a patient who was receiving a form of dialysis (eg, peritoneal or hemodialysis)? | ||
| Yes | 65 | 46.4% |
| No | 64 | 45.7% |
| If you have placed PICCs in patients on dialysis, do you discuss PICC placement or receive approval from nephrology prior to inserting the PICC? | ||
| Yes | 59 | 90.8% |
| No | 6 | 9.2% |
With respect to complications, catheter occlusion, migration, and DVT were reported as the 3 most prevalent adverse events. Interestingly, respondents did not report central lineassociated bloodstream infection (CLABSI) as a common complication. Additionally, 51% of those surveyed indicated that physicians unnecessarily removed PICCs when CLABSI was suspected but not confirmed. When managing catheter occlusion, 50% of respondents began with normal saline flushes but used tissue‐plasminogen activator if saline failed to resolve occlusion. Management of catheter migration varied based on degree of device movement: when the PICC had migrated <5 cm, most respondents (77%) indicated they would first obtain a chest x‐ray to determine the position of the PICC tip, with few (4%) performing catheter exchange. However, if the PICC had migrated more than 5 cm, a significantly greater proportion of respondents (21%) indicated they would perform a catheter exchange. With regard to managing DVT, most vascular nurses reported they notified nurses and physicians to continue using the PICC but recommended tests to confirm the diagnosis.
To better understand the experiences of vascular nurses, we asked for their perceptions regarding appropriateness of PICC use and relationships with bedside nurses, physicians, and leadership. Over a third of respondents (36%) felt that <5% of all PICCs may be inappropriate in their facility, whereas 1 in 5 indicated that 10% to 24% of PICCs placed in their facilities may be inappropriate or could have been avoided. Almost all (98%) of the nurses stated they were not empowered to remove idle or clinically unnecessary PICCs without physician authorization. Although 51% of nurses described the support received from hospital leadership as excellent, very good, or good, 43% described leadership support as either fair or poor. Conversely, relationships with bedside nurses and physicians were rated as being very good or good by nearly two‐thirds of those surveyed (64% and 65%, respectively) (Table 3).
| Question | No. | % |
|---|---|---|
| ||
| Which of the following PICC‐related complications have you most frequently encountered in your practice? | ||
| Catheter occlusion | 81 | 57.9% |
| Catheter migration | 27 | 19.3% |
| PICC‐associated DVT | 6 | 4.3% |
| Catheter fracture or embolization | 3 | 2.1% |
| Exit site infection | 3 | 2.1% |
| Coiling or kinking after insertion | 2 | 1.4% |
| If you suspect a patient has catheter occlusion, which of the following best describes your approach to resolving this problem? | ||
| Begin with normal saline but use a tPA product if this fails to restore patency | 70 | 50.0% |
| Use a tPA product (eg, Cathflo, Activase, or Retavase) to restore patency | 44 | 31.4% |
| Begin with heparin‐based flushes but use a tPA product if this fails to restore | 7 | 5.0% |
| Use only normal saline flushes to restore patency | 3 | 2.1% |
| If you find a PICC that has migrated out or has been accidentally dislodged <5 cm in a patient without symptoms, and the device is still clinically needed, which of the following best describes your practice? | ||
| Obtain a chest x‐ray to verify tip position | 108 | 77.1% |
| Perform a complete catheter exchange over a guidewire if possible | 5 | 3.6% |
| Notify/discuss next steps with physician | 5 | 3.6% |
| Other | 6 | 4.3% |
| If you find a PICC that has migrated out or has been accidentally dislodged >5 cm in a patient without symptoms, and the device is still clinically needed, which of the following best describes your practice? | ||
| Obtain a chest x‐ray to verify tip position | 72 | 51.4% |
| Perform a catheter exchange over a guidewire if possible | 30 | 21.4% |
| Notify/discuss next steps with physician | 10 | 7.1% |
| Other | 12 | 8.6% |
| Which of the following best describes your first approach when you suspect a patient has PICC‐associated phlebitis? | ||
| Discuss best course of action with physician or nurse | 79 | 56.4% |
| Supportive measures (eg, warm compresses, analgesics, monitoring) | 25 | 17.9% |
| Remove the PICC | 15 | 10.7% |
| Other | 5 | 3.6% |
| Which of the following best describes your first approach when you suspect a patient has a PICC‐related DVT? | ||
| Notify caregivers to continue using PICC and consider tests such as ultrasound | 82 | 58.6% |
| Notify bedside nurse and physician not to continue use of the PICC and consider tests such as ultrasound | 42 | 30.0% |
| PICCs are often removed when physicians suspect, but have not yet confirmed, CLABSI. Considering your experiences, what percentage of PICCs may have been removed in this manner at your facility? | ||
| <5% | 11 | 7.9% |
| 59% | 16 | 11.4% |
| 1024% | 24 | 17.1% |
| 25% | 71 | 50.7% |
| Based on your experience, what percentage of PICCs do you think are inappropriate or could have been avoided at your facility? | ||
| <5% | 51 | 36.4% |
| 59% | 25 | 17.9% |
| 1024% | 28 | 20.0% |
| 2550% | 13 | 9.3% |
| >50% | 5 | 3.6% |
| Are vascular access nurses empowered to remove PICCs that are idle or clinically unnecessary without physician authorization? | ||
| Yes | 3 | 2.1% |
| No | 122 | 87.1% |
| How would you rank the overall support your vascular access service receives from hospital leadership? | ||
| Excellent | 5 | 3.6% |
| Very good | 32 | 22.9% |
| Good | 40 | 28.6% |
| Fair | 35 | 25.0% |
| Poor | 25 | 17.9% |
| How would you describe your relationship with physicians at your facility when it comes to communicating recommendations or management of PICCs? | ||
| Very good | 28 | 20.0% |
| Good | 63 | 45.0% |
| Fair | 35 | 25.0% |
| Poor | 7 | 5.0% |
| Very poor | 4 | 2.9% |
| How would you describe your relationship with bedside nurses at your facility when it comes to communicating recommendations or management of PICCs? | ||
| Very good | 32 | 22.9% |
| Good | 58 | 41.4% |
| Fair | 38 | 27.1% |
| Poor | 7 | 5.0% |
| Very poor | 2 | 1.4% |
Variation in Responses Based on Years in Practice or Certification
We initially hypothesized that responses regarding practice (ultrasound use, ECG guidance system use), management of complications, or perceptions regarding leadership might vary based on years of experience, number of PICCs placed, or certification status. However, no statistically significant associations with these factors and individual responses were identified.
DISCUSSION
In this survey of 140 vascular access nurses in hospitals across Michigan, new insights regarding the experience, practice, knowledge, and beliefs of this group of providers were obtained. We found that vascular access nurses varied with respect to years in practice, volume of PICCs placed, and certification status, reflecting heterogeneity in this provider group. Variation in insertion techniques, such as use of ultrasound to examine catheter‐to‐vein ratio (a key way to prevent thrombosis) or newer ECG technology to position the PICC, was also noted. Although indications for PICC insertion appeared consistent with published literature, the frequency with which these devices were placed in patients receiving dialysis (reportedly with nephrology approval) was surprising given national calls to avoid such use.[16] Opportunities to improve hospital practices, such as tracking PICC dwell times and PICC necessity, as well as the potential need to better educate physicians on when to remove PICCs for suspected CLABSI, were also identified. Collectively, these data are highly relevant to hospitalists and health systems as they help to identify areas for quality improvement and inform clinical practice regarding the use of PICCs in hospitalized patients. As hospitalists increasingly order PICCs and manage complications associated with these devices, they are well suited to use these data so as to improve patient safety and clinical outcomes.
Venous access is the most common medical procedure performed in hospitalized medical patients. Although a number of devices including peripheral intravenous catheters, central venous catheters, and PICCs are used for this purpose, the growing use of PICCs to secure venous access has been documented in several studies.[17] Such growth, in part, undoubtedly reflects increasing availability of vascular access nurses. Traditionally placed by interventional radiologists, the creation of dedicated vascular nursing teams has resulted in these subspecialists now serving in more of a backup or trouble‐shooting role rather than that of primary operator.[4, 14] This paradigm shift is well illustrated in a recent survey of infection preventionists, where over 60% of respondents reported that they had a vascular nursing team in their facility.[7] The growth of these nursing‐led vascular access teams has produced not only high rates of insertion success and low rates of complications, but also greater cost‐effectiveness when compared to interventional radiologybased insertion.[18]
Nonetheless, our survey also identified a number of important concerns regarding PICC practices and vascular nursing providers. First, we found variation in areas such as insertion practices and management of complications. Such variability highlights the importance of both growing and disseminating the evidence base for consistent practice in vascular nursing. Through their close clinical affiliation with vascular nurses and shared interests in obtaining safe and appropriate venous access for patients, hospitalists are ideally poised to lead this effort. Second, similarities between vascular nurse opinions regarding appropriateness of PICCs and those of hospitalists from a prior survey were noted.[19] Namely, a substantial proportion of both vascular nurses and hospitalists felt that some PICCs were inappropriate and could be avoided. Third, although relationships between vascular access nurses and leadership were reported as being variable, the survey responses suggested relatively good interprovider relationships with bedside nurses and physicians. Such relationships likely reflect the close clinical ties that emerge from being in the trenches of patient care and suggest that interventions to improve care in partnership with these providers are highly viable.
Our study has some limitations. First, despite a high response rate, our study used a survey design and reports findings from a convenience sample of vascular access nurses in a single state. Thus, nonrespondent and selection biases remain threats to our conclusions. Additionally, some respondents did not complete all responses, perhaps due to nonapplicability to practice or other unknown reasons. The pattern of missingness observed, however, suggested that such responses were missing at random. Second, we surveyed vascular nurses in hospitals that are actively engaged in improving PICC practices; our findings may therefore not be representative of vascular nursing professionals as a whole and may instead reflect those of a highly motivated group of individuals. Relatedly, the underlying reasons for adoption of specific practices or techniques cannot be discerned from our study. Third, although we did not find differences based on years in practice or certification status, our sample size was relatively small and likely underpowered for these comparisons. Finally, our study sample consists of vascular nurses who are clustered within hospitals in which they are employed. Therefore, overlap between reported practices and those required by the facility are possible.
Despite these limitations, our study has important strengths. First, this is among the most comprehensive of surveys examining vascular nursing experience, practice, knowledge, and beliefs. The growing presence of these providers across US hospitals, coupled with limited insight regarding their clinical practices, highlight the importance and utility of these data. Second, we noted important differences in experience, practices, and interprovider relationships between vascular providers in this field. Although we are unable to ascertain the drivers or significance of such variation, hospitals and health systems focused on improving patient safety should consider quantifying and exploring these factors. Third, findings from our survey within Michigan suggest the need for similar, larger studies across the country. Partnerships with nursing organizations or larger professional groups that represent vascular nursing specialists may be helpful in this regard.
In conclusion, we found important similarities and differences in vascular nursing experience, practice, knowledge, and beliefs in Michigan. These data are useful as they help provide context regarding the constitution of these teams, current practices, and opportunities for improving care. Hospitalists seeking to improve patient safety may use these data to better inform vascular access practice in hospitalized patients.
Acknowledgements
The authors thank Claire Rickard, PhD, RN, Britt Meyer, RN, Peter Carr, PhD, and David Dempsey, RN for their assistance in developing the survey instrument used in this study.
Disclosures: This project was funded through an Investigator Initiated Research Grant from the Blue Cross Blue Shield of Michigan (BCBSM) Foundation (grant number 2140.II). The funding source played no role in study design, data acquisition, analysis, or reporting of the data. Support for the Hospital Medicine Safety (HMS) Consortium is provided by BCBSM and the Blue Care Network as part of the BCBSM Value Partnerships program. Although BCBSM and HMS work collaboratively, the opinions, beliefs, and viewpoints expressed by the authors do not necessarily reflect the opinions, beliefs, and viewpoints of BCBSM or any of its employees. This work was also supported with resources from the Veterans Affairs Ann Arbor Healthcare System. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States government.
- , , , et al. Peripherally inserted central venous catheters in the acute care setting: a safe alternative to high‐risk short‐term central venous catheters. Am J Infect Control. 2010;38(2):149–153.
- , , , , . Risk of venous thromboembolism in hospitalized patients with peripherally inserted central catheters. J Hosp Med. 2009;4(7):417–422.
- , , , , , . Central venous catheter placement by advanced practice nurses demonstrates low procedural complication and infection rates‐‐a report from 13 years of service. Crit Care Med. 2014;42(3):536–543.
- . Developing an alternative workflow model for peripherally inserted central catheter placement. J Infus Nurs. 2012;34(1):34–42.
- , . Facility wide benefits of radiology vascular access teams. Radiol Manage. 2010;32(1):28–32; quiz 33–34.
- , . Moving the needle forward: the imperative for collaboration in vascular access. J Infus Nurs. 2015;38(2):100–102.
- , , , . Use of designated PICC teams by U.S. hospitals: a survey‐based study [published online November 10, 2015]. J Patient Saf. doi: 10.1097/PTS.0000000000000246
- , , , , . The association between PICC use and venous thromboembolism in upper and lower extremities. American J Med. 2015;128(9):986–993.e1.
- , , , et al. Hospital performance for pharmacologic venous thromboembolism prophylaxis and rate of venous thromboembolism: a cohort study. JAMA Intern Med. 2014;174(10):1577–1584.
- , , , et al. Risk of venous thromboembolism associated with peripherally inserted central catheters: a systematic review and meta‐analysis. Lancet. 2013;382(9889):311–325.
- Infusion Nurses Society. Infusion nursing standards of practice. J Infus Nurs. 2006;29(1 suppl):S1–S92.
- , , , et al. Guidelines for the prevention of intravascular catheter‐related infections. Am J Infect Control. 2011;39(4 suppl 1):S1–S34.
- , , , et al. International evidence‐based recommendations on ultrasound‐guided vascular access. Intensive Care Med. 2012;38(7):1105–1117.
- , , . A single institution experience of seven hundred consecutively placed peripherally inserted central venous catheters. J Vasc Access. 2014;15(6):498–502.
- , , . Central venous access devices site care practices: an international survey of 34 countries [published online September 3, 2015]. J Vasc Access. doi: 10.5301/jva.5000450
- American Society of Nephrology. World's Leading Kidney Society Joins Effort to Reduce Unnecessary Medical Tests and Procedures. Available at: https://www.asn‐online.org/policy/choosingwisely/PressReleaseChoosingWisely.pdf. Accessed September 4, 2015.
- , , , . A survey of the current use of peripherally inserted central venous catheter (PICC) in Swedish oncology departments. Acta Oncol. 2013;52(6):1241–1242.
- , . Nurse‐led PICC insertion: is it cost effective? Br J Nurs. 2013;22(19):S9–S15.
- , , , et al. Hospitalist experiences, practice, opinions, and knowledge regarding peripherally inserted central catheters: a Michigan survey. J Hosp Med. 2013;8(6):309–314.
Peripherally inserted central catheters (PICCs) are among the most prevalent of venous access devices in hospitalized patients.[1, 2] Although growing use of these devices reflects clinical advantages, such as a reduced risk of complications during insertion and durable venous access, use of PICCs is also likely related to the growth of vascular access nursing.[3, 4] A relatively new specialty, vascular access nurses obtain, maintain, and manage venous access in hospitalized patients.[4, 5] Depending on their scope of practice, these professionals are responsible not only for insertion of devices, such as peripheral intravenous catheters and PICCs, but also nontunneled central venous catheters and arterial catheters in some settings.[6]
Although a growing number of US hospitals have introduced vascular nursing teams,[7] little is known about the experience, practice, knowledge, and beliefs of vascular access nurses. This knowledge gap is relevant for hospitalists and hospital medicine as (1) vascular access nurses increasingly represent a key partner in the care of hospitalized patients; (2) the knowledge and practice of these individuals directly affects patient safety and clinical outcomes; and (3) understanding experience, practice, and beliefs of these specialists can help inform decision making and quality‐improvement efforts related to PICCs. As hospitalists increasingly order the placement of and care for patients with PICCs, they are also well suited to improve PICC practice.
Therefore, we conducted a survey of vascular access nurses employed by hospitals that participate in the Michigan Hospital Medicine Safety (HMS) Consortium, a Blue Cross Blue Shield of Michiganfunded collaborative quality initiative.[6] We aimed to understand experience, practice, knowledge, and beliefs related to PICC care and use.
METHODS
Study Setting and Participants
To quantify vascular nursing experience, practice, knowledge, and beliefs, we conducted a Web‐based survey of vascular nurses across 47 Michigan hospitals that participate in HMS. A statewide quality‐improvement initiative, HMS aims to prevent adverse events in hospitalized medical patients through the creation of a data registry and sharing of best practices. The setting and design of this multicenter initiative have been previously described.[8, 9] Although participation is voluntary, each hospital receives payment for participating in the consortium and for data collection. Because HMS has an ongoing initiative aimed at identifying and preventing PICC‐related complications, this study was particularly relevant for participating hospitals and nurses.
Each HMS site has a designated quality‐improvement lead, physician champion, and data abstractor. To coordinate distribution and dissemination of the survey, we contacted the quality‐improvement leads at each site and enquired whether their hospital employed vascular access nurses who placed PICCs. Because we were only interested in responses from vascular access nurses, HMS hospitals that did not have these providers or stated PICCs were placed by other specialists (eg, interventional radiology) were excluded. At eligible sites, we obtained the total number of vascular nurses employed so as to determine the number of eligible respondents. In this manner, a purposeful sample of vascular nurses at participating HMS hospitals was constituted.
Participation in the survey was solicited through hospital quality leads that either distributed an electronic survey link to vascular nurses at their facilities or sent us individual email addresses to contact them directly. A cover letter explaining the rationale and the purpose of the survey along with the survey link was then sent to respondents through either of these routes. The survey was administered at all HMS sites contemporaneously and kept open for a period of 5 weeks. During the 5‐week period, 2 e‐mail reminders were sent to encourage participation. As a token of appreciation, a $10 Amazon gift card was offered to those who took the survey.
Development and Validation of the Survey
We developed the survey instrument (which we call PICC1 as we hope to administer longitudinally to track changes over time) by first conducting a literature search to identify relevant evidence‐based guidelines and studies regarding vascular access nursing practices and experiences.[10, 11, 12, 13] In addition, we consulted and involved national and international leaders in vascular access nursing to ensure validity and representativeness of the questions posed. We were specifically interested in nursing background, hospital practices, types of PICCs used, use of various technologies, relationships with healthcare providers, and management of complications. To understand participant characteristics and quantify potential variation in responses, we collected basic participant data including demographics, years in practice, number of PICCs placed, leadership roles, and vascular access certification status. Based on clinical reasoning and existing studies,[14, 15] we hypothesized that responses regarding certain practices (ultrasound use, electrocardiography [ECG] guidance system use), management of complications, or perceptions regarding leadership might vary based on years of experience, number of PICCs placed, or certification status. We therefore examined these associations as prespecified subgroup analyses.
The initial survey instrument was pilot tested with vascular nurses outside of the sampling frame. Based on feedback from the pilot testers, the instrument was refined and edited to improve clarity of the questions. In addition, specific skip patterns and logic were programmed into the final survey to reduce respondent burden and allow participants to seamlessly bypass questions that were contingent on a prior response (eg, use of ECG to place PICCs would lead to a series of questions about ECG‐assisted placement only for those respondents who used the technology). This final version of the survey was tested by members of the study team (V.C., L.K., S.L.K.) and then posted to SurveyMonkey for dissemination.
Statistical Analysis
Descriptive statistics (percentage, n/N) were used to tabulate results. In accordance with our a priori hypothesis that variation to responses might be associated with respondent characteristics, responses to questions regarding insertion practice (eg, use of ultrasound, measurement of catheter:vein ratio, trimming of catheters) and approach to complications (eg, catheter occlusion, deep vein thrombosis [DVT] notification, and PICC removal in the setting of fever) were compared by respondent years in practice (dichotomized to <5 vs >5 years), volume of PICCs placed (<999 vs 1000), and certification status (yes/no). Bivariate comparisons were made using 2 or Fisher exact tests based on the number of responses in a cell as appropriate; 2‐sided with a P value <0.05 was considered statistically significant. All statistical analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC).
Ethical and Regulatory Oversight
Because our study sought to describe existing practice without collecting any individual or facility level identifiable information, the project received a Not Regulated status by the University of Michigan Medical School Institutional Review Board (HUM00088351).
RESULTS
Of 172 vascular nurses who received invitations, 140 completed the survey for a response rate of 81%. Respondents reported working in not‐for‐profit hospitals (36%), academic medical centers (29%), and for‐profit hospitals (21%). Although multiple providers (eg, interventional radiology staff and providers, physicians) placed PICCs, 95% of those surveyed reported that they placed the majority of the PICCs at their institutions. Although most respondents placed PICCs in adult patients (86%), a few also placed PICCs in pediatric populations (17%). Vascular nursing programs were largely housed in their own department, but some reported to general nursing or subspecialties such as interventional radiology, cardiology, and critical care. Most respondents indicated their facilities had written policies regarding standard insertion and care practices (87% and 95%, respectively), but only 30% had policies regarding the necessity or appropriateness of PICCs.
Experience among respondents was variable: approximately a third had placed PICCs for <5 years (28.6%), whereas 58% reported placing PICCs for 5 years Correspondingly, 26% reported having placed 100 to 500 PICCs, whereas 34% had placed 1000 or more PICCs. Only 23% of those surveyed held a dedicated vascular access certification, such as board certified in vascular access or certified registered nurse infusion, whereas 16% indicated that they served as the vascular access lead nurse for their facility. Following placement, 94% of respondents reported that their facilities tracked the number of PICCs inserted, but only 40% indicated that dwell times of devices were also recorded. Only 30% of nurses reported that their hospitals had a written policy to evaluate PICC necessity or appropriateness following placement (Table 1).
| No.* | % | |
|---|---|---|
| ||
| Participant characteristics | ||
| For how many years have you been inserting PICCs? | ||
| <5 years | 40 | 28.6% |
| 5 years | 81 | 57.9% |
| Missing | ||
| In which of the following populations do you insert PICCs? | ||
| Adult patients | 121 | 86.4% |
| Pediatric patients | 24 | 17.1% |
| Neonatal patients | 1 | 0.7% |
| In which of the following locations do you place PICCs? (Select all that apply.) | ||
| Adult medical ward | 115 | 82.1% |
| General adult surgical ward | 110 | 78.6% |
| General pediatric medical ward | 34 | 24.3% |
| General pediatric surgical ward | 24 | 17.1% |
| Adult intensive care unit | 114 | 81.4% |
| Pediatric intensive care unit | 19 | 13.6% |
| Neonatal intensive care unit | 3 | 2.1% |
| Other intensive care unit | 59 | 42.1% |
| Outpatient clinic or emergency department | 17 | 12.1% |
| Other | 10 | 7.1% |
| Approximately how many PICCs may you have placed in your career? | ||
| 099 | 15 | 10.7% |
| 100499 | 36 | 25.7% |
| 500999 | 23 | 16.4% |
| 1,000 | 47 | 33.6% |
| Are you the vascular access lead nurse for your facility or organization? | ||
| Yes | 22 | 15.7% |
| No | 98 | 70.0% |
| Do you currently hold a dedicated vascular access certification (BC‐VA, CRNI, etc.)? | ||
| Yes | 32 | 22.9% |
| No | 89 | 63.6% |
| Facility characteristics | ||
| Which of the following best describes your primary work location? | ||
| Academic medical center | 41 | 29.3% |
| For‐profit community‐based hospital or medical center | 30 | 21.4% |
| Not‐for‐profit community‐based hospital or medical center | 50 | 35.7% |
| Who inserts the most PICCs in your facility? | ||
| Vascular access nurses | 133 | 95.0% |
| Interventional radiology or other providers | 7 | 5.0% |
| In which department is vascular access nursing located? | ||
| Vascular nursing | 76 | 54.3% |
| General nursing | 38 | 27.1% |
| Interventional radiology | 15 | 10.7% |
| Other | 11 | 7.9% |
| Using your best guess, how many PICCs do you think your facility inserts each month? | ||
| <25 | 5 | 3.6% |
| 2549 | 13 | 9.3% |
| 50100 | 39 | 27.9% |
| >100 | 78 | 55.7% |
| Unknown | 2 | 1.4% |
| How many vascular access nurses are employed by your facility? | ||
| <4 | 14 | 10.0% |
| 46 | 33 | 23.6% |
| 79 | 15 | 10.7% |
| 1015 | 25 | 17.9% |
| >15 | 53 | 37.9% |
| Does your facility track the number of PICCs placed? | ||
| Yes | 132 | 94.3% |
| No | 5 | 3.6% |
| Unknown | 3 | 2.1% |
| Does your facility track the duration or dwell time of PICCs? | ||
| Yes | 56 | 40.0% |
| No | 60 | 42.9% |
| Unknown | 24 | 17.1% |
| Does your facility have a written policy regarding standard PICC insertion practices? | ||
| Yes | 122 | 87.1% |
| No | 8 | 5.7% |
| Unknown | 7 | 5.0% |
| Does your facility have a written policy regarding standard PICC care and maintenance? | ||
| Yes | 133 | 95.0% |
| No | 3 | 2.1% |
| Unknown | 1 | 0.7% |
| Does your facility have a written process to review the necessity or appropriateness of a PICC? | ||
| Yes | 42 | 30.0% |
| No | 63 | 45.0% |
| Unknown | 20 | 14.3% |
The most commonly reported indications for PICC placement included intravenous antibiotics at discharge, difficult venous access, and placement for chemotherapy in patients with cancer. Forty‐six percent of nurses indicated they had placed a PICC in a patient receiving some form of dialysis in the past several months; however, 91% of these respondents reported receiving approval from nephrology prior to placement in these patients. Although almost all nurses (91%) used ultrasound to find a suitable vein for PICC placement, a smaller percentage used ultrasound to estimate the catheter‐to‐vein ratio to prevent thrombosis (79%), and only a few (14%) documented this figure in the medical record. Three‐quarters of those surveyed (76%) indicated they used ECG‐based systems to position PICC tips at the cavoatrial junction to prevent thrombosis. Of those who used this technology, 36% still obtained chest x‐rays to verify the position of the PICC tip. According to 84% of respondents, flushing of PICCs was performed mainly by bedside nurses, whereas scheduled weekly dressing changes were most often performed by vascular access nurses (Table 2).
| Question | No. | % |
|---|---|---|
| ||
| Do you use ultrasound to find a suitable vein prior to PICC insertion? | ||
| Yes | 128 | 91.4% |
| No | 0 | 0.0% |
| Do you use ultrasound to estimate the catheter‐to‐vein ratio prior to PICC insertion? | ||
| Yes | 110 | 78.6% |
| No | 18 | 12.9% |
| When using ultrasound, do you document the catheter‐to‐vein ratio in the PICC insertion note? | ||
| Yes | 20 | 14.3% |
| No | 89 | 63.6% |
| Do you use ECG guidance‐assisted systems to place PICCs? | ||
| Yes | 106 | 75.7% |
| No | 21 | 15.0% |
| If using ECG guidance, do you still routinely obtain a chest x‐ray to verify PICC tip position after placing the PICC using ECG guidance? | ||
| Yes | 38 | 27.1% |
| No | 68 | 48.6% |
| Who is primarily responsible for administering and adhering to a flushing protocol after PICC insertion at your facility? | ||
| Bedside nurses | 118 | 83.6% |
| Patients | 1 | 0.7% |
| Vascular access nurses | 8 | 5.7% |
| Which of the following agents are most often used to flush PICCs? | ||
| Both heparin and normal saline flushes | 61 | 43.6% |
| Normal saline only | 63 | 45.0% |
| Heparin only | 3 | 2.1% |
| Who is responsible for scheduled weekly dressing changes for PICCs? | ||
| Vascular access nurses | 110 | 78.6% |
| Bedside nurses | 14 | 10.0% |
| Other (eg, IR staff, ICU staff) | 3 | 2.1% |
| In the past few months, have you placed a PICC in a patient who was receiving a form of dialysis (eg, peritoneal or hemodialysis)? | ||
| Yes | 65 | 46.4% |
| No | 64 | 45.7% |
| If you have placed PICCs in patients on dialysis, do you discuss PICC placement or receive approval from nephrology prior to inserting the PICC? | ||
| Yes | 59 | 90.8% |
| No | 6 | 9.2% |
With respect to complications, catheter occlusion, migration, and DVT were reported as the 3 most prevalent adverse events. Interestingly, respondents did not report central lineassociated bloodstream infection (CLABSI) as a common complication. Additionally, 51% of those surveyed indicated that physicians unnecessarily removed PICCs when CLABSI was suspected but not confirmed. When managing catheter occlusion, 50% of respondents began with normal saline flushes but used tissue‐plasminogen activator if saline failed to resolve occlusion. Management of catheter migration varied based on degree of device movement: when the PICC had migrated <5 cm, most respondents (77%) indicated they would first obtain a chest x‐ray to determine the position of the PICC tip, with few (4%) performing catheter exchange. However, if the PICC had migrated more than 5 cm, a significantly greater proportion of respondents (21%) indicated they would perform a catheter exchange. With regard to managing DVT, most vascular nurses reported they notified nurses and physicians to continue using the PICC but recommended tests to confirm the diagnosis.
To better understand the experiences of vascular nurses, we asked for their perceptions regarding appropriateness of PICC use and relationships with bedside nurses, physicians, and leadership. Over a third of respondents (36%) felt that <5% of all PICCs may be inappropriate in their facility, whereas 1 in 5 indicated that 10% to 24% of PICCs placed in their facilities may be inappropriate or could have been avoided. Almost all (98%) of the nurses stated they were not empowered to remove idle or clinically unnecessary PICCs without physician authorization. Although 51% of nurses described the support received from hospital leadership as excellent, very good, or good, 43% described leadership support as either fair or poor. Conversely, relationships with bedside nurses and physicians were rated as being very good or good by nearly two‐thirds of those surveyed (64% and 65%, respectively) (Table 3).
| Question | No. | % |
|---|---|---|
| ||
| Which of the following PICC‐related complications have you most frequently encountered in your practice? | ||
| Catheter occlusion | 81 | 57.9% |
| Catheter migration | 27 | 19.3% |
| PICC‐associated DVT | 6 | 4.3% |
| Catheter fracture or embolization | 3 | 2.1% |
| Exit site infection | 3 | 2.1% |
| Coiling or kinking after insertion | 2 | 1.4% |
| If you suspect a patient has catheter occlusion, which of the following best describes your approach to resolving this problem? | ||
| Begin with normal saline but use a tPA product if this fails to restore patency | 70 | 50.0% |
| Use a tPA product (eg, Cathflo, Activase, or Retavase) to restore patency | 44 | 31.4% |
| Begin with heparin‐based flushes but use a tPA product if this fails to restore | 7 | 5.0% |
| Use only normal saline flushes to restore patency | 3 | 2.1% |
| If you find a PICC that has migrated out or has been accidentally dislodged <5 cm in a patient without symptoms, and the device is still clinically needed, which of the following best describes your practice? | ||
| Obtain a chest x‐ray to verify tip position | 108 | 77.1% |
| Perform a complete catheter exchange over a guidewire if possible | 5 | 3.6% |
| Notify/discuss next steps with physician | 5 | 3.6% |
| Other | 6 | 4.3% |
| If you find a PICC that has migrated out or has been accidentally dislodged >5 cm in a patient without symptoms, and the device is still clinically needed, which of the following best describes your practice? | ||
| Obtain a chest x‐ray to verify tip position | 72 | 51.4% |
| Perform a catheter exchange over a guidewire if possible | 30 | 21.4% |
| Notify/discuss next steps with physician | 10 | 7.1% |
| Other | 12 | 8.6% |
| Which of the following best describes your first approach when you suspect a patient has PICC‐associated phlebitis? | ||
| Discuss best course of action with physician or nurse | 79 | 56.4% |
| Supportive measures (eg, warm compresses, analgesics, monitoring) | 25 | 17.9% |
| Remove the PICC | 15 | 10.7% |
| Other | 5 | 3.6% |
| Which of the following best describes your first approach when you suspect a patient has a PICC‐related DVT? | ||
| Notify caregivers to continue using PICC and consider tests such as ultrasound | 82 | 58.6% |
| Notify bedside nurse and physician not to continue use of the PICC and consider tests such as ultrasound | 42 | 30.0% |
| PICCs are often removed when physicians suspect, but have not yet confirmed, CLABSI. Considering your experiences, what percentage of PICCs may have been removed in this manner at your facility? | ||
| <5% | 11 | 7.9% |
| 59% | 16 | 11.4% |
| 1024% | 24 | 17.1% |
| 25% | 71 | 50.7% |
| Based on your experience, what percentage of PICCs do you think are inappropriate or could have been avoided at your facility? | ||
| <5% | 51 | 36.4% |
| 59% | 25 | 17.9% |
| 1024% | 28 | 20.0% |
| 2550% | 13 | 9.3% |
| >50% | 5 | 3.6% |
| Are vascular access nurses empowered to remove PICCs that are idle or clinically unnecessary without physician authorization? | ||
| Yes | 3 | 2.1% |
| No | 122 | 87.1% |
| How would you rank the overall support your vascular access service receives from hospital leadership? | ||
| Excellent | 5 | 3.6% |
| Very good | 32 | 22.9% |
| Good | 40 | 28.6% |
| Fair | 35 | 25.0% |
| Poor | 25 | 17.9% |
| How would you describe your relationship with physicians at your facility when it comes to communicating recommendations or management of PICCs? | ||
| Very good | 28 | 20.0% |
| Good | 63 | 45.0% |
| Fair | 35 | 25.0% |
| Poor | 7 | 5.0% |
| Very poor | 4 | 2.9% |
| How would you describe your relationship with bedside nurses at your facility when it comes to communicating recommendations or management of PICCs? | ||
| Very good | 32 | 22.9% |
| Good | 58 | 41.4% |
| Fair | 38 | 27.1% |
| Poor | 7 | 5.0% |
| Very poor | 2 | 1.4% |
Variation in Responses Based on Years in Practice or Certification
We initially hypothesized that responses regarding practice (ultrasound use, ECG guidance system use), management of complications, or perceptions regarding leadership might vary based on years of experience, number of PICCs placed, or certification status. However, no statistically significant associations with these factors and individual responses were identified.
DISCUSSION
In this survey of 140 vascular access nurses in hospitals across Michigan, new insights regarding the experience, practice, knowledge, and beliefs of this group of providers were obtained. We found that vascular access nurses varied with respect to years in practice, volume of PICCs placed, and certification status, reflecting heterogeneity in this provider group. Variation in insertion techniques, such as use of ultrasound to examine catheter‐to‐vein ratio (a key way to prevent thrombosis) or newer ECG technology to position the PICC, was also noted. Although indications for PICC insertion appeared consistent with published literature, the frequency with which these devices were placed in patients receiving dialysis (reportedly with nephrology approval) was surprising given national calls to avoid such use.[16] Opportunities to improve hospital practices, such as tracking PICC dwell times and PICC necessity, as well as the potential need to better educate physicians on when to remove PICCs for suspected CLABSI, were also identified. Collectively, these data are highly relevant to hospitalists and health systems as they help to identify areas for quality improvement and inform clinical practice regarding the use of PICCs in hospitalized patients. As hospitalists increasingly order PICCs and manage complications associated with these devices, they are well suited to use these data so as to improve patient safety and clinical outcomes.
Venous access is the most common medical procedure performed in hospitalized medical patients. Although a number of devices including peripheral intravenous catheters, central venous catheters, and PICCs are used for this purpose, the growing use of PICCs to secure venous access has been documented in several studies.[17] Such growth, in part, undoubtedly reflects increasing availability of vascular access nurses. Traditionally placed by interventional radiologists, the creation of dedicated vascular nursing teams has resulted in these subspecialists now serving in more of a backup or trouble‐shooting role rather than that of primary operator.[4, 14] This paradigm shift is well illustrated in a recent survey of infection preventionists, where over 60% of respondents reported that they had a vascular nursing team in their facility.[7] The growth of these nursing‐led vascular access teams has produced not only high rates of insertion success and low rates of complications, but also greater cost‐effectiveness when compared to interventional radiologybased insertion.[18]
Nonetheless, our survey also identified a number of important concerns regarding PICC practices and vascular nursing providers. First, we found variation in areas such as insertion practices and management of complications. Such variability highlights the importance of both growing and disseminating the evidence base for consistent practice in vascular nursing. Through their close clinical affiliation with vascular nurses and shared interests in obtaining safe and appropriate venous access for patients, hospitalists are ideally poised to lead this effort. Second, similarities between vascular nurse opinions regarding appropriateness of PICCs and those of hospitalists from a prior survey were noted.[19] Namely, a substantial proportion of both vascular nurses and hospitalists felt that some PICCs were inappropriate and could be avoided. Third, although relationships between vascular access nurses and leadership were reported as being variable, the survey responses suggested relatively good interprovider relationships with bedside nurses and physicians. Such relationships likely reflect the close clinical ties that emerge from being in the trenches of patient care and suggest that interventions to improve care in partnership with these providers are highly viable.
Our study has some limitations. First, despite a high response rate, our study used a survey design and reports findings from a convenience sample of vascular access nurses in a single state. Thus, nonrespondent and selection biases remain threats to our conclusions. Additionally, some respondents did not complete all responses, perhaps due to nonapplicability to practice or other unknown reasons. The pattern of missingness observed, however, suggested that such responses were missing at random. Second, we surveyed vascular nurses in hospitals that are actively engaged in improving PICC practices; our findings may therefore not be representative of vascular nursing professionals as a whole and may instead reflect those of a highly motivated group of individuals. Relatedly, the underlying reasons for adoption of specific practices or techniques cannot be discerned from our study. Third, although we did not find differences based on years in practice or certification status, our sample size was relatively small and likely underpowered for these comparisons. Finally, our study sample consists of vascular nurses who are clustered within hospitals in which they are employed. Therefore, overlap between reported practices and those required by the facility are possible.
Despite these limitations, our study has important strengths. First, this is among the most comprehensive of surveys examining vascular nursing experience, practice, knowledge, and beliefs. The growing presence of these providers across US hospitals, coupled with limited insight regarding their clinical practices, highlight the importance and utility of these data. Second, we noted important differences in experience, practices, and interprovider relationships between vascular providers in this field. Although we are unable to ascertain the drivers or significance of such variation, hospitals and health systems focused on improving patient safety should consider quantifying and exploring these factors. Third, findings from our survey within Michigan suggest the need for similar, larger studies across the country. Partnerships with nursing organizations or larger professional groups that represent vascular nursing specialists may be helpful in this regard.
In conclusion, we found important similarities and differences in vascular nursing experience, practice, knowledge, and beliefs in Michigan. These data are useful as they help provide context regarding the constitution of these teams, current practices, and opportunities for improving care. Hospitalists seeking to improve patient safety may use these data to better inform vascular access practice in hospitalized patients.
Acknowledgements
The authors thank Claire Rickard, PhD, RN, Britt Meyer, RN, Peter Carr, PhD, and David Dempsey, RN for their assistance in developing the survey instrument used in this study.
Disclosures: This project was funded through an Investigator Initiated Research Grant from the Blue Cross Blue Shield of Michigan (BCBSM) Foundation (grant number 2140.II). The funding source played no role in study design, data acquisition, analysis, or reporting of the data. Support for the Hospital Medicine Safety (HMS) Consortium is provided by BCBSM and the Blue Care Network as part of the BCBSM Value Partnerships program. Although BCBSM and HMS work collaboratively, the opinions, beliefs, and viewpoints expressed by the authors do not necessarily reflect the opinions, beliefs, and viewpoints of BCBSM or any of its employees. This work was also supported with resources from the Veterans Affairs Ann Arbor Healthcare System. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States government.
Peripherally inserted central catheters (PICCs) are among the most prevalent of venous access devices in hospitalized patients.[1, 2] Although growing use of these devices reflects clinical advantages, such as a reduced risk of complications during insertion and durable venous access, use of PICCs is also likely related to the growth of vascular access nursing.[3, 4] A relatively new specialty, vascular access nurses obtain, maintain, and manage venous access in hospitalized patients.[4, 5] Depending on their scope of practice, these professionals are responsible not only for insertion of devices, such as peripheral intravenous catheters and PICCs, but also nontunneled central venous catheters and arterial catheters in some settings.[6]
Although a growing number of US hospitals have introduced vascular nursing teams,[7] little is known about the experience, practice, knowledge, and beliefs of vascular access nurses. This knowledge gap is relevant for hospitalists and hospital medicine as (1) vascular access nurses increasingly represent a key partner in the care of hospitalized patients; (2) the knowledge and practice of these individuals directly affects patient safety and clinical outcomes; and (3) understanding experience, practice, and beliefs of these specialists can help inform decision making and quality‐improvement efforts related to PICCs. As hospitalists increasingly order the placement of and care for patients with PICCs, they are also well suited to improve PICC practice.
Therefore, we conducted a survey of vascular access nurses employed by hospitals that participate in the Michigan Hospital Medicine Safety (HMS) Consortium, a Blue Cross Blue Shield of Michiganfunded collaborative quality initiative.[6] We aimed to understand experience, practice, knowledge, and beliefs related to PICC care and use.
METHODS
Study Setting and Participants
To quantify vascular nursing experience, practice, knowledge, and beliefs, we conducted a Web‐based survey of vascular nurses across 47 Michigan hospitals that participate in HMS. A statewide quality‐improvement initiative, HMS aims to prevent adverse events in hospitalized medical patients through the creation of a data registry and sharing of best practices. The setting and design of this multicenter initiative have been previously described.[8, 9] Although participation is voluntary, each hospital receives payment for participating in the consortium and for data collection. Because HMS has an ongoing initiative aimed at identifying and preventing PICC‐related complications, this study was particularly relevant for participating hospitals and nurses.
Each HMS site has a designated quality‐improvement lead, physician champion, and data abstractor. To coordinate distribution and dissemination of the survey, we contacted the quality‐improvement leads at each site and enquired whether their hospital employed vascular access nurses who placed PICCs. Because we were only interested in responses from vascular access nurses, HMS hospitals that did not have these providers or stated PICCs were placed by other specialists (eg, interventional radiology) were excluded. At eligible sites, we obtained the total number of vascular nurses employed so as to determine the number of eligible respondents. In this manner, a purposeful sample of vascular nurses at participating HMS hospitals was constituted.
Participation in the survey was solicited through hospital quality leads that either distributed an electronic survey link to vascular nurses at their facilities or sent us individual email addresses to contact them directly. A cover letter explaining the rationale and the purpose of the survey along with the survey link was then sent to respondents through either of these routes. The survey was administered at all HMS sites contemporaneously and kept open for a period of 5 weeks. During the 5‐week period, 2 e‐mail reminders were sent to encourage participation. As a token of appreciation, a $10 Amazon gift card was offered to those who took the survey.
Development and Validation of the Survey
We developed the survey instrument (which we call PICC1 as we hope to administer longitudinally to track changes over time) by first conducting a literature search to identify relevant evidence‐based guidelines and studies regarding vascular access nursing practices and experiences.[10, 11, 12, 13] In addition, we consulted and involved national and international leaders in vascular access nursing to ensure validity and representativeness of the questions posed. We were specifically interested in nursing background, hospital practices, types of PICCs used, use of various technologies, relationships with healthcare providers, and management of complications. To understand participant characteristics and quantify potential variation in responses, we collected basic participant data including demographics, years in practice, number of PICCs placed, leadership roles, and vascular access certification status. Based on clinical reasoning and existing studies,[14, 15] we hypothesized that responses regarding certain practices (ultrasound use, electrocardiography [ECG] guidance system use), management of complications, or perceptions regarding leadership might vary based on years of experience, number of PICCs placed, or certification status. We therefore examined these associations as prespecified subgroup analyses.
The initial survey instrument was pilot tested with vascular nurses outside of the sampling frame. Based on feedback from the pilot testers, the instrument was refined and edited to improve clarity of the questions. In addition, specific skip patterns and logic were programmed into the final survey to reduce respondent burden and allow participants to seamlessly bypass questions that were contingent on a prior response (eg, use of ECG to place PICCs would lead to a series of questions about ECG‐assisted placement only for those respondents who used the technology). This final version of the survey was tested by members of the study team (V.C., L.K., S.L.K.) and then posted to SurveyMonkey for dissemination.
Statistical Analysis
Descriptive statistics (percentage, n/N) were used to tabulate results. In accordance with our a priori hypothesis that variation to responses might be associated with respondent characteristics, responses to questions regarding insertion practice (eg, use of ultrasound, measurement of catheter:vein ratio, trimming of catheters) and approach to complications (eg, catheter occlusion, deep vein thrombosis [DVT] notification, and PICC removal in the setting of fever) were compared by respondent years in practice (dichotomized to <5 vs >5 years), volume of PICCs placed (<999 vs 1000), and certification status (yes/no). Bivariate comparisons were made using 2 or Fisher exact tests based on the number of responses in a cell as appropriate; 2‐sided with a P value <0.05 was considered statistically significant. All statistical analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC).
Ethical and Regulatory Oversight
Because our study sought to describe existing practice without collecting any individual or facility level identifiable information, the project received a Not Regulated status by the University of Michigan Medical School Institutional Review Board (HUM00088351).
RESULTS
Of 172 vascular nurses who received invitations, 140 completed the survey for a response rate of 81%. Respondents reported working in not‐for‐profit hospitals (36%), academic medical centers (29%), and for‐profit hospitals (21%). Although multiple providers (eg, interventional radiology staff and providers, physicians) placed PICCs, 95% of those surveyed reported that they placed the majority of the PICCs at their institutions. Although most respondents placed PICCs in adult patients (86%), a few also placed PICCs in pediatric populations (17%). Vascular nursing programs were largely housed in their own department, but some reported to general nursing or subspecialties such as interventional radiology, cardiology, and critical care. Most respondents indicated their facilities had written policies regarding standard insertion and care practices (87% and 95%, respectively), but only 30% had policies regarding the necessity or appropriateness of PICCs.
Experience among respondents was variable: approximately a third had placed PICCs for <5 years (28.6%), whereas 58% reported placing PICCs for 5 years Correspondingly, 26% reported having placed 100 to 500 PICCs, whereas 34% had placed 1000 or more PICCs. Only 23% of those surveyed held a dedicated vascular access certification, such as board certified in vascular access or certified registered nurse infusion, whereas 16% indicated that they served as the vascular access lead nurse for their facility. Following placement, 94% of respondents reported that their facilities tracked the number of PICCs inserted, but only 40% indicated that dwell times of devices were also recorded. Only 30% of nurses reported that their hospitals had a written policy to evaluate PICC necessity or appropriateness following placement (Table 1).
| No.* | % | |
|---|---|---|
| ||
| Participant characteristics | ||
| For how many years have you been inserting PICCs? | ||
| <5 years | 40 | 28.6% |
| 5 years | 81 | 57.9% |
| Missing | ||
| In which of the following populations do you insert PICCs? | ||
| Adult patients | 121 | 86.4% |
| Pediatric patients | 24 | 17.1% |
| Neonatal patients | 1 | 0.7% |
| In which of the following locations do you place PICCs? (Select all that apply.) | ||
| Adult medical ward | 115 | 82.1% |
| General adult surgical ward | 110 | 78.6% |
| General pediatric medical ward | 34 | 24.3% |
| General pediatric surgical ward | 24 | 17.1% |
| Adult intensive care unit | 114 | 81.4% |
| Pediatric intensive care unit | 19 | 13.6% |
| Neonatal intensive care unit | 3 | 2.1% |
| Other intensive care unit | 59 | 42.1% |
| Outpatient clinic or emergency department | 17 | 12.1% |
| Other | 10 | 7.1% |
| Approximately how many PICCs may you have placed in your career? | ||
| 099 | 15 | 10.7% |
| 100499 | 36 | 25.7% |
| 500999 | 23 | 16.4% |
| 1,000 | 47 | 33.6% |
| Are you the vascular access lead nurse for your facility or organization? | ||
| Yes | 22 | 15.7% |
| No | 98 | 70.0% |
| Do you currently hold a dedicated vascular access certification (BC‐VA, CRNI, etc.)? | ||
| Yes | 32 | 22.9% |
| No | 89 | 63.6% |
| Facility characteristics | ||
| Which of the following best describes your primary work location? | ||
| Academic medical center | 41 | 29.3% |
| For‐profit community‐based hospital or medical center | 30 | 21.4% |
| Not‐for‐profit community‐based hospital or medical center | 50 | 35.7% |
| Who inserts the most PICCs in your facility? | ||
| Vascular access nurses | 133 | 95.0% |
| Interventional radiology or other providers | 7 | 5.0% |
| In which department is vascular access nursing located? | ||
| Vascular nursing | 76 | 54.3% |
| General nursing | 38 | 27.1% |
| Interventional radiology | 15 | 10.7% |
| Other | 11 | 7.9% |
| Using your best guess, how many PICCs do you think your facility inserts each month? | ||
| <25 | 5 | 3.6% |
| 2549 | 13 | 9.3% |
| 50100 | 39 | 27.9% |
| >100 | 78 | 55.7% |
| Unknown | 2 | 1.4% |
| How many vascular access nurses are employed by your facility? | ||
| <4 | 14 | 10.0% |
| 46 | 33 | 23.6% |
| 79 | 15 | 10.7% |
| 1015 | 25 | 17.9% |
| >15 | 53 | 37.9% |
| Does your facility track the number of PICCs placed? | ||
| Yes | 132 | 94.3% |
| No | 5 | 3.6% |
| Unknown | 3 | 2.1% |
| Does your facility track the duration or dwell time of PICCs? | ||
| Yes | 56 | 40.0% |
| No | 60 | 42.9% |
| Unknown | 24 | 17.1% |
| Does your facility have a written policy regarding standard PICC insertion practices? | ||
| Yes | 122 | 87.1% |
| No | 8 | 5.7% |
| Unknown | 7 | 5.0% |
| Does your facility have a written policy regarding standard PICC care and maintenance? | ||
| Yes | 133 | 95.0% |
| No | 3 | 2.1% |
| Unknown | 1 | 0.7% |
| Does your facility have a written process to review the necessity or appropriateness of a PICC? | ||
| Yes | 42 | 30.0% |
| No | 63 | 45.0% |
| Unknown | 20 | 14.3% |
The most commonly reported indications for PICC placement included intravenous antibiotics at discharge, difficult venous access, and placement for chemotherapy in patients with cancer. Forty‐six percent of nurses indicated they had placed a PICC in a patient receiving some form of dialysis in the past several months; however, 91% of these respondents reported receiving approval from nephrology prior to placement in these patients. Although almost all nurses (91%) used ultrasound to find a suitable vein for PICC placement, a smaller percentage used ultrasound to estimate the catheter‐to‐vein ratio to prevent thrombosis (79%), and only a few (14%) documented this figure in the medical record. Three‐quarters of those surveyed (76%) indicated they used ECG‐based systems to position PICC tips at the cavoatrial junction to prevent thrombosis. Of those who used this technology, 36% still obtained chest x‐rays to verify the position of the PICC tip. According to 84% of respondents, flushing of PICCs was performed mainly by bedside nurses, whereas scheduled weekly dressing changes were most often performed by vascular access nurses (Table 2).
| Question | No. | % |
|---|---|---|
| ||
| Do you use ultrasound to find a suitable vein prior to PICC insertion? | ||
| Yes | 128 | 91.4% |
| No | 0 | 0.0% |
| Do you use ultrasound to estimate the catheter‐to‐vein ratio prior to PICC insertion? | ||
| Yes | 110 | 78.6% |
| No | 18 | 12.9% |
| When using ultrasound, do you document the catheter‐to‐vein ratio in the PICC insertion note? | ||
| Yes | 20 | 14.3% |
| No | 89 | 63.6% |
| Do you use ECG guidance‐assisted systems to place PICCs? | ||
| Yes | 106 | 75.7% |
| No | 21 | 15.0% |
| If using ECG guidance, do you still routinely obtain a chest x‐ray to verify PICC tip position after placing the PICC using ECG guidance? | ||
| Yes | 38 | 27.1% |
| No | 68 | 48.6% |
| Who is primarily responsible for administering and adhering to a flushing protocol after PICC insertion at your facility? | ||
| Bedside nurses | 118 | 83.6% |
| Patients | 1 | 0.7% |
| Vascular access nurses | 8 | 5.7% |
| Which of the following agents are most often used to flush PICCs? | ||
| Both heparin and normal saline flushes | 61 | 43.6% |
| Normal saline only | 63 | 45.0% |
| Heparin only | 3 | 2.1% |
| Who is responsible for scheduled weekly dressing changes for PICCs? | ||
| Vascular access nurses | 110 | 78.6% |
| Bedside nurses | 14 | 10.0% |
| Other (eg, IR staff, ICU staff) | 3 | 2.1% |
| In the past few months, have you placed a PICC in a patient who was receiving a form of dialysis (eg, peritoneal or hemodialysis)? | ||
| Yes | 65 | 46.4% |
| No | 64 | 45.7% |
| If you have placed PICCs in patients on dialysis, do you discuss PICC placement or receive approval from nephrology prior to inserting the PICC? | ||
| Yes | 59 | 90.8% |
| No | 6 | 9.2% |
With respect to complications, catheter occlusion, migration, and DVT were reported as the 3 most prevalent adverse events. Interestingly, respondents did not report central lineassociated bloodstream infection (CLABSI) as a common complication. Additionally, 51% of those surveyed indicated that physicians unnecessarily removed PICCs when CLABSI was suspected but not confirmed. When managing catheter occlusion, 50% of respondents began with normal saline flushes but used tissue‐plasminogen activator if saline failed to resolve occlusion. Management of catheter migration varied based on degree of device movement: when the PICC had migrated <5 cm, most respondents (77%) indicated they would first obtain a chest x‐ray to determine the position of the PICC tip, with few (4%) performing catheter exchange. However, if the PICC had migrated more than 5 cm, a significantly greater proportion of respondents (21%) indicated they would perform a catheter exchange. With regard to managing DVT, most vascular nurses reported they notified nurses and physicians to continue using the PICC but recommended tests to confirm the diagnosis.
To better understand the experiences of vascular nurses, we asked for their perceptions regarding appropriateness of PICC use and relationships with bedside nurses, physicians, and leadership. Over a third of respondents (36%) felt that <5% of all PICCs may be inappropriate in their facility, whereas 1 in 5 indicated that 10% to 24% of PICCs placed in their facilities may be inappropriate or could have been avoided. Almost all (98%) of the nurses stated they were not empowered to remove idle or clinically unnecessary PICCs without physician authorization. Although 51% of nurses described the support received from hospital leadership as excellent, very good, or good, 43% described leadership support as either fair or poor. Conversely, relationships with bedside nurses and physicians were rated as being very good or good by nearly two‐thirds of those surveyed (64% and 65%, respectively) (Table 3).
| Question | No. | % |
|---|---|---|
| ||
| Which of the following PICC‐related complications have you most frequently encountered in your practice? | ||
| Catheter occlusion | 81 | 57.9% |
| Catheter migration | 27 | 19.3% |
| PICC‐associated DVT | 6 | 4.3% |
| Catheter fracture or embolization | 3 | 2.1% |
| Exit site infection | 3 | 2.1% |
| Coiling or kinking after insertion | 2 | 1.4% |
| If you suspect a patient has catheter occlusion, which of the following best describes your approach to resolving this problem? | ||
| Begin with normal saline but use a tPA product if this fails to restore patency | 70 | 50.0% |
| Use a tPA product (eg, Cathflo, Activase, or Retavase) to restore patency | 44 | 31.4% |
| Begin with heparin‐based flushes but use a tPA product if this fails to restore | 7 | 5.0% |
| Use only normal saline flushes to restore patency | 3 | 2.1% |
| If you find a PICC that has migrated out or has been accidentally dislodged <5 cm in a patient without symptoms, and the device is still clinically needed, which of the following best describes your practice? | ||
| Obtain a chest x‐ray to verify tip position | 108 | 77.1% |
| Perform a complete catheter exchange over a guidewire if possible | 5 | 3.6% |
| Notify/discuss next steps with physician | 5 | 3.6% |
| Other | 6 | 4.3% |
| If you find a PICC that has migrated out or has been accidentally dislodged >5 cm in a patient without symptoms, and the device is still clinically needed, which of the following best describes your practice? | ||
| Obtain a chest x‐ray to verify tip position | 72 | 51.4% |
| Perform a catheter exchange over a guidewire if possible | 30 | 21.4% |
| Notify/discuss next steps with physician | 10 | 7.1% |
| Other | 12 | 8.6% |
| Which of the following best describes your first approach when you suspect a patient has PICC‐associated phlebitis? | ||
| Discuss best course of action with physician or nurse | 79 | 56.4% |
| Supportive measures (eg, warm compresses, analgesics, monitoring) | 25 | 17.9% |
| Remove the PICC | 15 | 10.7% |
| Other | 5 | 3.6% |
| Which of the following best describes your first approach when you suspect a patient has a PICC‐related DVT? | ||
| Notify caregivers to continue using PICC and consider tests such as ultrasound | 82 | 58.6% |
| Notify bedside nurse and physician not to continue use of the PICC and consider tests such as ultrasound | 42 | 30.0% |
| PICCs are often removed when physicians suspect, but have not yet confirmed, CLABSI. Considering your experiences, what percentage of PICCs may have been removed in this manner at your facility? | ||
| <5% | 11 | 7.9% |
| 59% | 16 | 11.4% |
| 1024% | 24 | 17.1% |
| 25% | 71 | 50.7% |
| Based on your experience, what percentage of PICCs do you think are inappropriate or could have been avoided at your facility? | ||
| <5% | 51 | 36.4% |
| 59% | 25 | 17.9% |
| 1024% | 28 | 20.0% |
| 2550% | 13 | 9.3% |
| >50% | 5 | 3.6% |
| Are vascular access nurses empowered to remove PICCs that are idle or clinically unnecessary without physician authorization? | ||
| Yes | 3 | 2.1% |
| No | 122 | 87.1% |
| How would you rank the overall support your vascular access service receives from hospital leadership? | ||
| Excellent | 5 | 3.6% |
| Very good | 32 | 22.9% |
| Good | 40 | 28.6% |
| Fair | 35 | 25.0% |
| Poor | 25 | 17.9% |
| How would you describe your relationship with physicians at your facility when it comes to communicating recommendations or management of PICCs? | ||
| Very good | 28 | 20.0% |
| Good | 63 | 45.0% |
| Fair | 35 | 25.0% |
| Poor | 7 | 5.0% |
| Very poor | 4 | 2.9% |
| How would you describe your relationship with bedside nurses at your facility when it comes to communicating recommendations or management of PICCs? | ||
| Very good | 32 | 22.9% |
| Good | 58 | 41.4% |
| Fair | 38 | 27.1% |
| Poor | 7 | 5.0% |
| Very poor | 2 | 1.4% |
Variation in Responses Based on Years in Practice or Certification
We initially hypothesized that responses regarding practice (ultrasound use, ECG guidance system use), management of complications, or perceptions regarding leadership might vary based on years of experience, number of PICCs placed, or certification status. However, no statistically significant associations with these factors and individual responses were identified.
DISCUSSION
In this survey of 140 vascular access nurses in hospitals across Michigan, new insights regarding the experience, practice, knowledge, and beliefs of this group of providers were obtained. We found that vascular access nurses varied with respect to years in practice, volume of PICCs placed, and certification status, reflecting heterogeneity in this provider group. Variation in insertion techniques, such as use of ultrasound to examine catheter‐to‐vein ratio (a key way to prevent thrombosis) or newer ECG technology to position the PICC, was also noted. Although indications for PICC insertion appeared consistent with published literature, the frequency with which these devices were placed in patients receiving dialysis (reportedly with nephrology approval) was surprising given national calls to avoid such use.[16] Opportunities to improve hospital practices, such as tracking PICC dwell times and PICC necessity, as well as the potential need to better educate physicians on when to remove PICCs for suspected CLABSI, were also identified. Collectively, these data are highly relevant to hospitalists and health systems as they help to identify areas for quality improvement and inform clinical practice regarding the use of PICCs in hospitalized patients. As hospitalists increasingly order PICCs and manage complications associated with these devices, they are well suited to use these data so as to improve patient safety and clinical outcomes.
Venous access is the most common medical procedure performed in hospitalized medical patients. Although a number of devices including peripheral intravenous catheters, central venous catheters, and PICCs are used for this purpose, the growing use of PICCs to secure venous access has been documented in several studies.[17] Such growth, in part, undoubtedly reflects increasing availability of vascular access nurses. Traditionally placed by interventional radiologists, the creation of dedicated vascular nursing teams has resulted in these subspecialists now serving in more of a backup or trouble‐shooting role rather than that of primary operator.[4, 14] This paradigm shift is well illustrated in a recent survey of infection preventionists, where over 60% of respondents reported that they had a vascular nursing team in their facility.[7] The growth of these nursing‐led vascular access teams has produced not only high rates of insertion success and low rates of complications, but also greater cost‐effectiveness when compared to interventional radiologybased insertion.[18]
Nonetheless, our survey also identified a number of important concerns regarding PICC practices and vascular nursing providers. First, we found variation in areas such as insertion practices and management of complications. Such variability highlights the importance of both growing and disseminating the evidence base for consistent practice in vascular nursing. Through their close clinical affiliation with vascular nurses and shared interests in obtaining safe and appropriate venous access for patients, hospitalists are ideally poised to lead this effort. Second, similarities between vascular nurse opinions regarding appropriateness of PICCs and those of hospitalists from a prior survey were noted.[19] Namely, a substantial proportion of both vascular nurses and hospitalists felt that some PICCs were inappropriate and could be avoided. Third, although relationships between vascular access nurses and leadership were reported as being variable, the survey responses suggested relatively good interprovider relationships with bedside nurses and physicians. Such relationships likely reflect the close clinical ties that emerge from being in the trenches of patient care and suggest that interventions to improve care in partnership with these providers are highly viable.
Our study has some limitations. First, despite a high response rate, our study used a survey design and reports findings from a convenience sample of vascular access nurses in a single state. Thus, nonrespondent and selection biases remain threats to our conclusions. Additionally, some respondents did not complete all responses, perhaps due to nonapplicability to practice or other unknown reasons. The pattern of missingness observed, however, suggested that such responses were missing at random. Second, we surveyed vascular nurses in hospitals that are actively engaged in improving PICC practices; our findings may therefore not be representative of vascular nursing professionals as a whole and may instead reflect those of a highly motivated group of individuals. Relatedly, the underlying reasons for adoption of specific practices or techniques cannot be discerned from our study. Third, although we did not find differences based on years in practice or certification status, our sample size was relatively small and likely underpowered for these comparisons. Finally, our study sample consists of vascular nurses who are clustered within hospitals in which they are employed. Therefore, overlap between reported practices and those required by the facility are possible.
Despite these limitations, our study has important strengths. First, this is among the most comprehensive of surveys examining vascular nursing experience, practice, knowledge, and beliefs. The growing presence of these providers across US hospitals, coupled with limited insight regarding their clinical practices, highlight the importance and utility of these data. Second, we noted important differences in experience, practices, and interprovider relationships between vascular providers in this field. Although we are unable to ascertain the drivers or significance of such variation, hospitals and health systems focused on improving patient safety should consider quantifying and exploring these factors. Third, findings from our survey within Michigan suggest the need for similar, larger studies across the country. Partnerships with nursing organizations or larger professional groups that represent vascular nursing specialists may be helpful in this regard.
In conclusion, we found important similarities and differences in vascular nursing experience, practice, knowledge, and beliefs in Michigan. These data are useful as they help provide context regarding the constitution of these teams, current practices, and opportunities for improving care. Hospitalists seeking to improve patient safety may use these data to better inform vascular access practice in hospitalized patients.
Acknowledgements
The authors thank Claire Rickard, PhD, RN, Britt Meyer, RN, Peter Carr, PhD, and David Dempsey, RN for their assistance in developing the survey instrument used in this study.
Disclosures: This project was funded through an Investigator Initiated Research Grant from the Blue Cross Blue Shield of Michigan (BCBSM) Foundation (grant number 2140.II). The funding source played no role in study design, data acquisition, analysis, or reporting of the data. Support for the Hospital Medicine Safety (HMS) Consortium is provided by BCBSM and the Blue Care Network as part of the BCBSM Value Partnerships program. Although BCBSM and HMS work collaboratively, the opinions, beliefs, and viewpoints expressed by the authors do not necessarily reflect the opinions, beliefs, and viewpoints of BCBSM or any of its employees. This work was also supported with resources from the Veterans Affairs Ann Arbor Healthcare System. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States government.
- , , , et al. Peripherally inserted central venous catheters in the acute care setting: a safe alternative to high‐risk short‐term central venous catheters. Am J Infect Control. 2010;38(2):149–153.
- , , , , . Risk of venous thromboembolism in hospitalized patients with peripherally inserted central catheters. J Hosp Med. 2009;4(7):417–422.
- , , , , , . Central venous catheter placement by advanced practice nurses demonstrates low procedural complication and infection rates‐‐a report from 13 years of service. Crit Care Med. 2014;42(3):536–543.
- . Developing an alternative workflow model for peripherally inserted central catheter placement. J Infus Nurs. 2012;34(1):34–42.
- , . Facility wide benefits of radiology vascular access teams. Radiol Manage. 2010;32(1):28–32; quiz 33–34.
- , . Moving the needle forward: the imperative for collaboration in vascular access. J Infus Nurs. 2015;38(2):100–102.
- , , , . Use of designated PICC teams by U.S. hospitals: a survey‐based study [published online November 10, 2015]. J Patient Saf. doi: 10.1097/PTS.0000000000000246
- , , , , . The association between PICC use and venous thromboembolism in upper and lower extremities. American J Med. 2015;128(9):986–993.e1.
- , , , et al. Hospital performance for pharmacologic venous thromboembolism prophylaxis and rate of venous thromboembolism: a cohort study. JAMA Intern Med. 2014;174(10):1577–1584.
- , , , et al. Risk of venous thromboembolism associated with peripherally inserted central catheters: a systematic review and meta‐analysis. Lancet. 2013;382(9889):311–325.
- Infusion Nurses Society. Infusion nursing standards of practice. J Infus Nurs. 2006;29(1 suppl):S1–S92.
- , , , et al. Guidelines for the prevention of intravascular catheter‐related infections. Am J Infect Control. 2011;39(4 suppl 1):S1–S34.
- , , , et al. International evidence‐based recommendations on ultrasound‐guided vascular access. Intensive Care Med. 2012;38(7):1105–1117.
- , , . A single institution experience of seven hundred consecutively placed peripherally inserted central venous catheters. J Vasc Access. 2014;15(6):498–502.
- , , . Central venous access devices site care practices: an international survey of 34 countries [published online September 3, 2015]. J Vasc Access. doi: 10.5301/jva.5000450
- American Society of Nephrology. World's Leading Kidney Society Joins Effort to Reduce Unnecessary Medical Tests and Procedures. Available at: https://www.asn‐online.org/policy/choosingwisely/PressReleaseChoosingWisely.pdf. Accessed September 4, 2015.
- , , , . A survey of the current use of peripherally inserted central venous catheter (PICC) in Swedish oncology departments. Acta Oncol. 2013;52(6):1241–1242.
- , . Nurse‐led PICC insertion: is it cost effective? Br J Nurs. 2013;22(19):S9–S15.
- , , , et al. Hospitalist experiences, practice, opinions, and knowledge regarding peripherally inserted central catheters: a Michigan survey. J Hosp Med. 2013;8(6):309–314.
- , , , et al. Peripherally inserted central venous catheters in the acute care setting: a safe alternative to high‐risk short‐term central venous catheters. Am J Infect Control. 2010;38(2):149–153.
- , , , , . Risk of venous thromboembolism in hospitalized patients with peripherally inserted central catheters. J Hosp Med. 2009;4(7):417–422.
- , , , , , . Central venous catheter placement by advanced practice nurses demonstrates low procedural complication and infection rates‐‐a report from 13 years of service. Crit Care Med. 2014;42(3):536–543.
- . Developing an alternative workflow model for peripherally inserted central catheter placement. J Infus Nurs. 2012;34(1):34–42.
- , . Facility wide benefits of radiology vascular access teams. Radiol Manage. 2010;32(1):28–32; quiz 33–34.
- , . Moving the needle forward: the imperative for collaboration in vascular access. J Infus Nurs. 2015;38(2):100–102.
- , , , . Use of designated PICC teams by U.S. hospitals: a survey‐based study [published online November 10, 2015]. J Patient Saf. doi: 10.1097/PTS.0000000000000246
- , , , , . The association between PICC use and venous thromboembolism in upper and lower extremities. American J Med. 2015;128(9):986–993.e1.
- , , , et al. Hospital performance for pharmacologic venous thromboembolism prophylaxis and rate of venous thromboembolism: a cohort study. JAMA Intern Med. 2014;174(10):1577–1584.
- , , , et al. Risk of venous thromboembolism associated with peripherally inserted central catheters: a systematic review and meta‐analysis. Lancet. 2013;382(9889):311–325.
- Infusion Nurses Society. Infusion nursing standards of practice. J Infus Nurs. 2006;29(1 suppl):S1–S92.
- , , , et al. Guidelines for the prevention of intravascular catheter‐related infections. Am J Infect Control. 2011;39(4 suppl 1):S1–S34.
- , , , et al. International evidence‐based recommendations on ultrasound‐guided vascular access. Intensive Care Med. 2012;38(7):1105–1117.
- , , . A single institution experience of seven hundred consecutively placed peripherally inserted central venous catheters. J Vasc Access. 2014;15(6):498–502.
- , , . Central venous access devices site care practices: an international survey of 34 countries [published online September 3, 2015]. J Vasc Access. doi: 10.5301/jva.5000450
- American Society of Nephrology. World's Leading Kidney Society Joins Effort to Reduce Unnecessary Medical Tests and Procedures. Available at: https://www.asn‐online.org/policy/choosingwisely/PressReleaseChoosingWisely.pdf. Accessed September 4, 2015.
- , , , . A survey of the current use of peripherally inserted central venous catheter (PICC) in Swedish oncology departments. Acta Oncol. 2013;52(6):1241–1242.
- , . Nurse‐led PICC insertion: is it cost effective? Br J Nurs. 2013;22(19):S9–S15.
- , , , et al. Hospitalist experiences, practice, opinions, and knowledge regarding peripherally inserted central catheters: a Michigan survey. J Hosp Med. 2013;8(6):309–314.
© 2015 Society of Hospital Medicine
Length of Different‐Hospital Readmissions
Readmissions within a relatively short time after discharge are receiving considerable attention as an area of quality improvement,[1, 2] with increasing emphasis on the relatively large share of readmissions to different hospitals, accounting for 20% to 30% of all readmissions.[3, 4, 5, 6] Returning to a different hospital may affect patient and healthcare outcomes due to breaches in continuity. When information from the previous recent hospitalization is not transferred efficiently and accurately to the next admitting hospital, omissions and duplications can occur, resulting in delayed care and potentially worse outcomes (compared to same hospital readmissions [SHRs]), such as longer length of readmission stay (LORS) and increased costs.[7]
Electronic health records (EHRs) and health information exchange (HIE) systems are increasingly used for storage and retrieval of patient information from various sources, such as laboratories and previous physician visits and hospitalizations, enabling informational continuity by providing vital historical medical information for decision‐making. Whereas EHRs collect, store, and present information that is locally created within a specific clinic or hospital, HIEs connect EHR systems between multiple institutions, allowing providers to share clinical data and achieve interorganizational continuity. Such integrative systems are increasingly being implemented across healthcare systems worldwide.[8, 9, 10] Yet, technical difficulties, costs, competitive concerns, data privacy, and workflow implementation challenges have been described as hindering HIE participation.[11, 12, 13, 14] Moreover, major concerns exist regarding the poor usability of EHRs, their limited ability to support multidisciplinary care, and major difficulties in achieving interoperability with HIEs, which undermine efforts to deliver integrated patient‐centered care.[15] Nonetheless, previous research has demonstrated that HIEs can positively affect healthcare resource use and outcomes, including reduced rates of repeated diagnostic imaging in the emergency evaluation of back pain,[16] reduction in admissions via the emergency department (ED),[17] and reduced rates of readmissions within 7 days.[18] However, it is not known whether HIEs can contribute to positive outcomes when patients are readmitted to a different hospital than the hospital from which they were recently (within the previous 30 days) discharged, potentially bridging the transitional‐care information divide.
In Israel, an innovative HIE system, OFEK (literally horizon), was implemented in 2005 at the largest not‐for‐profit insurer and provider of services, Clalit Health Services (Clalit). Clalit operates as an integrated healthcare delivery system, serving more than 50% of the Israeli population, as part of the country's national health insurance system. OFEK links information on all Clalit enrollees from all hospitals, primary care, and specialty care clinics, laboratories, and diagnostic services into a single, virtual, patient file, enabling providers to obtain complete, real‐time information needed for healthcare decision making at the point of care. Like similar HIE systems, OFEK includes information on previous medical encounters and hospitalizations, previous diagnoses, chronically prescribed medications, previous lab and imaging tests, known allergies, and some demographic information.[19] At the time of this study, OFEK was available in all Clalit hospitals as well as in 2 non‐Clalit (government‐owned and operated) large tertiary‐care centers, resulting in 40% coverage of all hospitalizations through the OFEK HIE system. As part of a large organization‐wide readmission reduction program recently implemented by Clalit for all its members admitted to any hospital in Israel, aimed at early detection and intervention,[20] OFEK was viewed as an important mechanism to help maintain continuity and improve transitions.
To inform current knowledge on different‐hospital readmissions (DHRs) and HIEs, we examined whether having HIE systems can contribute to information continuity and prevent delays in care and the need for more expensive, lengthy readmission stays when patients are readmitted to a different hospital. More specifically, we tested whether there is a difference in the LORS between SHRs and DHRs, and whether DHRs the LORS differ by the availability of an HIE (whether index and readmitting hospital are or are not connected through HIE systems).
METHODS
Study Design and Setting
We conducted a retrospective cohort study based on data of hospitalized Clalit members. Clalit has a centralized data warehouse with a comprehensive EHR containing data on all patients' medical encounters, administrative data, and clinical data compiled from laboratories, imaging centers, and hospitals. At the time of the study, OFEK was operating in all 8 Clalit hospitals and in 2 large government‐owned and operated hospitals in the central and northern parts of the country. Information is linked in the Clalit system and OFEK‐affiliated hospitals through an individual identity number assigned by the Israeli Interior Ministry to every Israeli resident for general identification purposes.
Population
The study examined all internal medicine and intensive‐care unit (ICU) readmissions of adult Clalit members (aged 18 years and older) previously (within the prior 30 days) discharged from internal medicine departments during January 1, 2010 until December 31, 2010 (ie, index hospitalizations). Only readmissions of index hospitalizations with more than a 24‐hour stay were included. A total of 146,266 index hospitalizations met the inclusion criteria. Index admissions that resulted in a transfer to another hospital, a long‐term care facility, or rehabilitation center were not included (N = 11,831). The final study sample included 27,057 readmissions (20.1% of the 134,435 index admissions), which could have resulted in any type of discharge (to patient's home, a long‐term care or rehabilitation facility, or due to death). The study was approved by Clalit's institutional review board.
Outcome Variable
We defined the LORS as the number of days from admission to discharge during readmission.
Main Independent Variable
We assessed information continuity as a categorical variable in which 0 = no information continuity (DHRs with either no HIE at either hospital or an HIE in only 1 of the hospitals), 1 = information continuity through an HIE (DHRs with both hospitals having an HIE), and 2 = full information continuity (readmission to the same hospital).
Covariates
We examined the following known correlates of length of stay (LOS): age, gender, residency in a nursing home, socioeconomic status (SES) based on an indicator of social security entitlement received by low‐income members,[21] and the occurrence of common chronic conditions registered in Clalit's EHR registries[22]: congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), chronic renal failure (CRF), malignancy, diabetes, hypertension, ischemic heart disease, atrial fibrillation, asthma, and disability (indication of a functional limitation). To provide comorbidity adjustment we used the Charlson Comorbidity Index.[23] Additionally, we assessed LOS of the index hospitalization. We included an indicator for the size of the index hospital: small, fewer than 100 beds; medium, 100 to 200 beds; and large, more than 200 beds. Finally, to account for a well‐known correlate of length of hospital stay,[24] we included an indicator for an ICU stay during the readmission.
Statistical Analysis
We first examined the study populations' characteristics and calculated the average LORS for each SHR and DHR category. Due to the skewed distribution of LORS, we also calculated the median and interquartile range (IQR) of LORS and evaluated the difference between categories using the Kruskall‐Wallis test.[25] Sample‐size calculations showed that we would need a sample of >3000 admissions to have 80% power to detect a difference of 0.8 hospitalization days given the 1:3 ratio between the DHR groups. To examine the association between LORS and information continuity, we employed a univariate marginal Cox model.[26] Variables that were significantly (P < 0.05) associated with LORS in the univariate model were entered into a multivariate marginal Cox model, clustering by patient and using a robust sandwich covariance matrix estimate. Additionally, we performed a sensitivity analysis using hierarchichal modeling to account for potential variations due to hospital level factors. A low hazard ratio (<1) represented an association of the covariate with decreased likelihood of discharge, that is, longer LORS. All analyses were conducted with SPSS version 20 (IBM, Armonk, NY) and SAS version 9.3 (SAS Institute Inc., Cary, NC).
RESULTS
The study included a total of 27,057 readmissions, of which 23,927 (88.4%) were SHRs and 3130 (11.6%) were DHRs. Of all DHRs, in 792 (2.9%) of the cases, both hospitals had HIEs (partial information continuity), and in 2338 (8.6%), either 1 or both did not have an HIE system (thus meaning there was no information continuity). Characteristics of the study population are shown in Table 1. Most (75%) of the readmissions were of patients over the age of 65 years, though only 7% were nursing home residents. More than half the study's population consisted of patients with low SES. The most common chronic conditions were hypertension (77%), ischemic heart disease (52%), and diabetes (48%). Other chronic conditions were arrhythmia (38%), CHF (35%), disability (31%), COPD (28%), malignancy (28%), and asthma (16%). In more than 55% of the index hospitalizations, the LOS was 4 days or less, and most index admissions (64%) were in large hospitals. Table 1 also displays the study population by the type of readmission: SHR, DHR with HIE, and DHR without HIE. As compared to patients readmitted to the same hospital, patients with DHRs were younger (P < 0.001), less likely to be nursing home residents (P < 0.001), and had longer LOS during the index admission (P < 0.001). Additionally, patients with SHRs were more likely to have their index admission at a large hospital (P < 0.001), had a higher comorbidity score (P < 0.043), and were less likely be treated in the ICU during their readmission (P < 0.001) compared to their DHR counterparts. Patients with DHRs without an HIE were similar in most characteristics to those with an HIE, except for having an ICU stay during their readmission (6.4% compared with 9.2%, respectively).
| Characteristics | All Readmissions, n = 27,057 | SHR, n = 23,927 | DHR With HIE, n = 792 | DHR Without HIE, n = 2,338 | P Value |
|---|---|---|---|---|---|
| |||||
| All personal characteristics | |||||
| Age, n (%) | <0.001 | ||||
| 1844 years | 1,328 (4.9) | 1,095 (4.6) | 58 (7.3) | 175 (7.5) | |
| 4564 years | 5,370 (19.8) | 4,597 (19.2) | 197 (24.9) | 576 (24.6) | |
| 6584 years | 14,059 (52.0) | 12,500 (52.2) | 402 (50.8) | 1,157 (49.5) | |
| 85+ years | 6,300 (23.3) | 5,735 (24.0) | 135 (17.0) | 430 (18.4) | |
| Female sex, n (%) | 13,742 (50.8) | 12,040 (50.3) | 418 (52.8) | 1,284 (54.9) | <0.001 |
| Low socioeconomic status, n (%) | 15,473 (57.2) | 13,670 (57.1) | 453 (57.2) | 1,350 (57.7) | |
| Residency in a nursing home, n (%) | 1,857 (6.9) | 1,743 (7.3) | 27 (3.4) | 87 (3.7) | <0.001 |
| Common chronic conditions, n (%) | |||||
| Hypertension | 20,797 (76.9) | 18,484 (77.3) | 588 (74.2) | 1,725 (73.8) | <0.001 |
| Ischemic heart disease | 14,150 (52.3) | 12,577 (52.6) | 397 (50.1) | 1,176 (50.3) | 0.052 |
| Diabetes | 13,052 (48.2) | 11,589 (48.4) | 345 (43.6) | 1,118 (47.8) | 0.024 |
| Arrhythmia | 10,306 (38.1) | 9,197 (38.4) | 292 (36.9) | 817 (34.9) | 0.003 |
| Chronic renal failure | 9,486 (35.1) | 8,454 (35.3) | 262 (33.1) | 770 (32.9) | 0.034 |
| Congestive heart failure | 9,216 (34.1) | 8,232 (34.4) | 270 (34.1) | 714 (30.5) | 0.001 |
| Disability | 8,362 (30.9) | 7,600 (31.8) | 165 (20.8) | 597 (25.5) | <0.001 |
| Chronic obstructive pulmonary disease | 7,671 (28.4) | 6,888 (28.8) | 201 (25.4) | 582 (24.9) | <0.001 |
| Malignancy | 7,642 (28.2) | 6,763 (28.3) | 220 (27.8) | 659 (28.2) | 0.954 |
| Asthma | 4,491 (16.6) | 4,040 (16.9) | 109 (13.8) | 342 (14.6) | 0.002 |
| Charlson score, mean [SD] | 4.54 [3.15] | 4.58 [3.14] | 4.14 [3.08] | 4.25 [3.24] | 0.043 |
| Index hospitalization characteristics (LOS during index hospitalization), n (%) | <0.001 | ||||
| 24 days | 14,961 (55.3) | 13,310 (55.6) | 428 (54.0) | 1,223 (52.3) | |
| 57 days | 6,366 (23.5) | 5,654 (23.6) | 174 (22.0) | 538 (23.0) | |
| 8 days and more | 5,730 (21.2) | 4,963 (20.7) | 190 (24.0) | 577 (24.7) | |
| Hospital size in index hospitalization (no. of hospitals in each category), n (%) | <0.001 | ||||
| Small, <100 beds (8) | 1,498 (5.5) | 1,166 (4.9) | 23 (2.9) | 309 (13.2) | |
| Medium, 100200 beds (9) | 8,129 (30.0) | 7,113 (29.7) | 316 (39.9) | 700 (29.9) | |
| Large, >200 beds (10) | 17,430 (64.4) | 15,648 (65.4) | 453 (57.2) | 1,329 (56.8) | |
| Intensive care unit during readmission, n (%) | 869 (3.2) | 647 (2.7) | 73 (9.2) | 149 (6.4) | <0.001 |
The mean LORS in SHRs was shorter by 1 day than the mean LORS for DHRs: 6.3 (95% confidence interval [CI]: 6.2‐6.4) versus 7.3 (95% CI: 7.0‐7.6), respectively. Mean LORS in DHRs with or without HIE was 7.6 (95% CI: 7.0‐8.3) and 7.2 (95% CI: 6.8‐7.6), respectively. Although median LORS was similar (4 days), the IQR differed, resulting in significant differences between the SHR and DHR groups (Table 2).
| Information Continuity | No. of Readmissions | Mean LORS (95% CI) | Median (Q1Q3) | Kruskal‐Wallis P Value |
|---|---|---|---|---|
| ||||
| SHRs | 23,927 (88.4) | 6.3 (6.26.4) | 4 (27) | |
| DHRs | 3,130 (11.6) | 7.3 (7.07.6) | 4 (28) | |
| DHRs with HIE | 792 (2.9) | 7.6 (7.08.3) | 4 (29) | |
| DHRs without HIE | 2,338 (8.7) | 7.2 (6.87.6) | 4 (28) | |
| Total | 27,057 | 6.4 (6.36.5) | 4 (27) | <0.001 |
In the multivariate model, partial continuity (DHRs with an HIE) was associated with decreased likelihood of discharge on any given day compared with full continuity (SHR) (hazard ratio [HR] = 0.85, 95% CI: 0.79‐0.91). Similar results were obtained for no continuity (DHRs without an HIE) (HR = 0.90, 95% CI: 0.86‐0.94). The difference between DHRs with and without an HIE was not significant (overlapping confidence intervals). Other factors associated with a lower HR for discharge during each day of the readmission were older age, residency in a nursing home, CHF, CRF, disability, malignancy, and long LOS (8+ days) during the index hospitalization. Patients with asthma or ischemic heart disease had a higher HR for discharge during each readmission day (Table 3). We performed a sensitivity analysis using hierarchical modeling (patients nested within hospitals), which resulted in similar findings in terms of directionality and magnitude of the relationships and significance levels.
| Characteristics | Univariate Model | Multivariate Model | ||
|---|---|---|---|---|
| Hazard Ratio (95% CI) | P Value | Hazard Ratio (95% CI) | P Value | |
| ||||
| Information continuity | ||||
| SHR | Reference | Reference | ||
| DHR with HIE | 0.87 (0.810.93) | <0.001 | 0.86 (0.800.93) | <0.001 |
| DHR without HIE | 0.91 (0.870.94) | <0.001 | 0.90 (0.870.94) | <0.001 |
| Age | ||||
| 844 years | 1.22 (1.181.26) | <0.001 | 1.14 (1.071.22) | <0.001 |
| 4564 years | 1.16 (1.141.18) | <0.001 | 1.11 (1.061.1) | <0.001 |
| 6584 years | 1.01 (0.991.02) | 0.53 | 0.99 (0.961.02) | 0.60 |
| 85+ years | Reference | Reference | ||
| Sex | ||||
| Male | 0.97 (0.950.99) | 0.008 | 0.98 (0.961.01) | 0.19 |
| Female | Reference | Reference | ||
| Socioeconomic status | ||||
| Low | 0.98 (0.970.99) | 0.11 | ||
| Other | Reference | |||
| Residency in a nursing home | ||||
| Nursing home | 0.90 (0.880.92) | <0.001 | 0.90 (0.860.95) | <0.001 |
| All others | Reference | Reference | ||
| Common chronic conditions (reference: without condition) | ||||
| Hypertension | 0.94 (0.930.96) | <0.001 | 1.01 (0.971.04) | 0.69 |
| Ischemic heart disease | 1.00 (0.991.01) | 0.93 | 1.06 (1.031.09) | <0.001 |
| Diabetes | 0.97 (0.950.98) | 0.004 | 0.99 (0.971.02) | 0.64 |
| Arrhythmia | 0.96 (0.950.97) | 0.002 | 1.01 (0.981.04) | 0.39 |
| Chronic renal failure | 0.92 (0.910.93) | <0.001 | 0.96 (0.930.99) | 0.01 |
| Congestive heart failure | 0.93 (0.920.94) | <0.001 | 0.96 (0.930.99) | 0.01 |
| Disability | 0.86 (0.850.87) | <0.001 | 0.90 (0.870.92) | <0.001 |
| Chronic obstructive pulmonary disease | 0.99 (0.981.01) | 0.66 | ||
| Malignancy | 0.97 (0.960.98) | 0.03 | 0.98 (0.961.01) | 0.28 |
| Asthma | 1.04 (1.021.06) | 0.03 | 1.04 (1.001.07) | 0.03 |
| Charlson score | 0.99 (0.980.99) | <0.001 | 0.99 (0.991.00) | 0.04 |
| LOS during index hospitalization | ||||
| Days 24 | 1.52 (1.491.54) | <0.001 | 1.49 (1.451.54) | <0.001 |
| Days 57 | 1.21 (1.191.23) | <0.001 | 1.20 (1.161.24) | <0.001 |
| 8 days and more | Reference | Reference | ||
| Hospital size in index hospitalization | ||||
| Small, <100 beds (8) | 0.94 (0.920.97) | 0.02 | 1.00 (0.951.05) | 0.93 |
| Medium, 100200 beds (9) | 1.00 (0.991.02) | 0.78 | 1.01 (0.991.04) | 0.38 |
| Large, >200 beds (10) | Reference | Reference | ||
| Intensive care unit in readmission | ||||
| Yes | 0.75 (0.700.80) | <0.001 | 0.74 (0.690.79) | <0.001 |
| No | Reference | Reference | ||
DISCUSSION
This study shows that readmission to a different hospital results in longer duration of the readmission stay compared with readmission to the same index hospital. Our results also show that having HIE systems in both the index and readmitting hospitals does not protect against these negative outcomes, as there was no difference in the length of the readmission stay based on the availability of HIE systems. Factors that were found to be associated with longer readmission stays are well known indicators of the complexity of the patient's medical condition, such as the presence of disability, comorbidity, and ICU treatment during the readmission.[24, 27]
The shorter LORS in SHRs may be due to the familiarity of physicians and other healthcare providers with the patient and his or her condition, especially as the policy in SHRs in Israel is to readmit to the same unit from which the patient was recently discharged. This same hospital familiarity is especially important as hospital care in Israel follows the hospitalist model, in which responsibility for patient care is transferred from the patient's primary care physician to the hospital's physician, resulting in increased need for integration through HIE systems, especially when patients are readmitted to a different hospital.[28, 29]
Our findings, congruent with previous research on DHRs and poor outcomes,[7] could also be explained by the inefficiency associated with transitions. For example, patients frequently leave the hospital with pending lab tests, often with abnormal results that would change the course of care.[30] Because these pending tests are often omitted from the hospital discharge summaries,[31] if patients are hospitalized in a different hospital, the same tests may be ordered again, or a course of treatment that does not acknowledge the test results could be taken. Such time‐consuming duplication can be prevented in SHRs, where the index‐hospital records may be already more complete.
Our null findings regarding the contribution of HIE systems may be explained by the low levels of HIE actual use. Although we did not directly assess use, previous research reports that actual use of HIE is limited.[12] An Israeli study on the effects of the use of the OFEK system on ED physicians' admission decisions found that the patient's medical history was viewed in only 31.2% of all 281,750 ED referrals.[19] In another Israeli‐based ED study, even lower usage levels were found, with the OFEK system having been accessed in only 16% of all 3,219,910 ED referrals.[32] Low levels of HIE use have also been reported in the United States. An additional study, which tested the implementation of HIE in hospitals and clinics, showed that in only 2.3% of encounters did providers access the HIE record.[33] Another study conducted in 12 ED sites and 2 ambulatory clinics reported rates of 6.8% HIE use.[34] Moreover, the null effect of integrated health information reported here is congruent with findings from a US study on implementation of an electronic discharge instructions form with embedded computerized medication reconciliation, which was not found to be associated with postdischarge outcomes.[35]
A wide range of factors may influence decisions on HIE use: patient‐level factors,[36] perceived medical complexity of the patient,[33, 34] and the number of prior hospitalizations.[33, 34, 36] Healthcare systemlevel factors may include: time constraints, which may be positively[32] or negatively[33] associated with HIE use, and organizational policies or incentives.[33] Use may also be associated with features of the HIE technology itself, such as difficulty to access, difficulty to use once accessed, and the quality of information it contains.[37] Additionally, there is some evidence of the link between tight functional integration and higher proportions of usage.[38] Although comprehensive studies on factors affecting the use of the OFEK system in Israeli internal medicine units are still needed, the lack of its integration within each hospital's EHR system may serve as a major explanatory factor for the low usage levels.
The findings from this study should be interpreted in light of its limitations. First, compared with previously reported DHR rates (20%30%),[3, 5] the rate observed in our population was relatively low (about 12%). Previous research was restricted to heart failure patients[3] or assessed DHR in surgical, as well as internal medicine, patients.[5] Our lower rates may have been affected by the type of population (hospitalized internal medicine patients) and/or by characteristics of the Clalit healthcare system, which serves as an integrated provider network as well as insurer. Generalization from 1 health care system to others should be made with caution. Nonetheless, our results may underestimate the potential effect in other healthcare systems with less structural integration. Additionally, as noted above, information on the actual use of an HIE in the course of medical decision making during readmission was absent. Future studies should examine the potential benefit of an HIE with measures that capture providers' use of HIEs. Also, the LORS may be influenced by other factors not investigated here, and further future studies should examine additional outcomes such as costs, patient well‐being, and satisfaction. Finally, causality could not be determined, and future research in this realm should aim to search for the pathways connecting readmission to a different hospital, with and without HIEs, to readmission LOS and additional outcomes.
To conclude, our findings show that patients readmitted to a different hospital are at risk for prolonged LORS, regardless of the availability of HIE systems. Implementing HIE systems is the focus of substantial efforts by policymakers and is considered a key part of the meaningful use of electronic health information. HIE features in the provisions of the Health Information Technology for Economic and Clinical Health Act[39] because it can furnish providers with complete, timely information at the point of care. Moreover, although there has been substantial growth in the number of healthcare organizations that have operational an HIE, its ability to lead to improved outcomes has yet to be realized.[8, 10] The Israeli experience reported here suggests that provisions are needed that will ensure actual use of HIEs, which might in turn minimize the difference between DHRs and SHRs.
Acknowledgements
The authors acknowledge Chandra Cohen‐Stavi, MPA, and Orly Tonkikh, MA, for their contribution to this study.
Disclosures
The study was supported in part by a grant from the Israel National Institute for Health Policy Research (NIHP) (10/127). The authors report no conflicts of interest.
- , , , , , . Assessing preventability in the quest to reduce hospital readmissions. J Hosp Med. 2014;9:598–603.
- , , , et al. Preventing 30‐day hospital readmissions: a systematic review and meta‐analysis of randomized trials. JAMA Intern Med. 2014;174:1095–1107.
- , , , et al. Is same‐hospital readmission rate a good surrogate for all‐hospital readmission rate? Med Care. 2010;48:477–481.
- , , , . Hospital readmission rates: the impacts of age, payer, and mental health diagnoses. J Ambul Care Manage. 2013;36(2):147–155.
- , , , . Limitations of using same‐hospital readmission metrics. Int J Qual Health Care. 2013;25(6):633–639.
- Hospital inpatient and outpatient services. In: Report to the Congress: promoting greater efficiency in Medicare. Washington, DC: Medicare Payment Advisory Commission., March 2012;45–66.
- , , , , . For‐profit hospital status and rehospitalizations at different hospitals: an analysis of Medicare data. Ann Intern Med. 2010;153:718–727.
- , , , . Health information technology: an updated systematic review with a focus on meaningful use. Ann Intern Med. 2014;160:48–54.
- , , , . The benefits of health information technology: a review of the recent literature shows predominantly positive results. Health Aff (Millwood). 2011;30(3):464–471.
- , , . Health information exchange among US hospitals. Am J Manag Care. 2011;17:761–768.
- , , . A survey of health information exchange organizations in the United States: implications for meaningful use. Ann Intern Med. 2011;54:666–671.
- , , , , . Physicians' potential use and preferences related to health information exchange. Int J Med Inform. 2011;80:171–180.
- , , , et al. Provider stakeholders' perceived benefit from a nascent health information exchange: a qualitative analysis. J Med Syst. 2012;36:601–613.
- . More than just a question of technology: factors related to hospitals' adoption and implementation of health information exchange. Int J Med Inform. 2010;79:797–806.
- , , . Leveraging health information technology to achieve the “triple aim” of healthcare reform. J Am Med Inform Assoc. 2015;22(4):849–856.
- , , , , , . Health information exchange reduces repeated diagnostic imaging for back pain. Ann Emerg Med. 2013;62:16–24.
- , , , , . Association between use of a health information exchange system and hospital admissions. Appl Clin Inform. 2014;5:219.
- , , . The impact of EHR and HIE on reducing avoidable admissions: controlling main differential diagnoses. BMC Med Inform Decis Mak. 2013;13:49.
- , , , et al. The impact of an integrated hospital‐community medical information system on quality and service utilization in hospital departments. Int J Med Inform. 2010;79(9):649–657.
- , , , , , Predicting 30‐day readmissions with preadmission electronic health record data. Med Care. 2015;53:283–289.
- , , , , . Assessing socioeconomic health care utilization inequity in Israel: impact of alternative approaches to morbidity adjustment. BMC Public Health. 2011;11(1):609.
- , . Prevalence of selected chronic diseases in Israel. Isr Med Assoc J. 2001;3:404–408.
- , , , . A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis. 1987;40:373–383.
- , , , , . Systematic review of risk adjustment models of hospital length of stay (LOS). Med Care. 2015;53:355–365.
- , . Use of ranks in one‐criterion variance analysis. J Am Stat Assoc. 1952;47:583–621.
- , , . Regression analysis of multivariate incomplete failure time data by modeling marginal distributions. J Am Stat Assoc. 1989;84:1065–1073.
- , , , , , . Disability impacts length of stay in general internal medicine patients. J Gen Intern Med. 2014;29:885–890.
- . The hospitalist movement—time to move on. N Engl J Med. 2007;357:2627–2629.
- , , . Association of hospitalist presence and hospital‐level outcome measures among Medicare patients. J Hosp Med. 2014;9:1–6.
- , , , et al. Patient safety concerns arising from test results that return after hospital discharge. Ann Intern Med. 2005;143:121–128.
- , , , et al. Adequacy of hospital discharge summaries in documenting tests with pending results and outpatient follow‐up providers. J Gen Intern Med. 2009;24:1002–1006.
- , , . Using electronic medical record systems for admission decisions in emergency departments: examining the crowdedness effect. J Med Syst. 2012;36:3795–3803.
- , , , , , . Factors motivating and affecting health information exchange usage. J Am Med Inform Assoc. 2011;18(2):143–149.
- , , , et al. Health information exchange usage in emergency departments and clinics: the who, what, and why. J Am Med Inform Assoc. 2011;18:690–697.
- , , , , . Effect of standardized electronic discharge instructions on post‐discharge hospital utilization. J Gen Intern Med. 2011;26:718–723.
- , , . Health information exchange technology on the front lines of healthcare: workflow factors and patterns of use. J Am Med Inform Assoc. 2012;19:392–400.
- , . The DeLone and McLean model of information systems success: a ten‐year update. J Manag Inf Syst. 2003;19:9–30.
- , , , et al. Architectural strategies and issues with health information exchange. AMIA Annu Symp Proc. 2006:814–818.
- . Launching HITECH. N Engl J Med. 2010;362(5):382–385.
Readmissions within a relatively short time after discharge are receiving considerable attention as an area of quality improvement,[1, 2] with increasing emphasis on the relatively large share of readmissions to different hospitals, accounting for 20% to 30% of all readmissions.[3, 4, 5, 6] Returning to a different hospital may affect patient and healthcare outcomes due to breaches in continuity. When information from the previous recent hospitalization is not transferred efficiently and accurately to the next admitting hospital, omissions and duplications can occur, resulting in delayed care and potentially worse outcomes (compared to same hospital readmissions [SHRs]), such as longer length of readmission stay (LORS) and increased costs.[7]
Electronic health records (EHRs) and health information exchange (HIE) systems are increasingly used for storage and retrieval of patient information from various sources, such as laboratories and previous physician visits and hospitalizations, enabling informational continuity by providing vital historical medical information for decision‐making. Whereas EHRs collect, store, and present information that is locally created within a specific clinic or hospital, HIEs connect EHR systems between multiple institutions, allowing providers to share clinical data and achieve interorganizational continuity. Such integrative systems are increasingly being implemented across healthcare systems worldwide.[8, 9, 10] Yet, technical difficulties, costs, competitive concerns, data privacy, and workflow implementation challenges have been described as hindering HIE participation.[11, 12, 13, 14] Moreover, major concerns exist regarding the poor usability of EHRs, their limited ability to support multidisciplinary care, and major difficulties in achieving interoperability with HIEs, which undermine efforts to deliver integrated patient‐centered care.[15] Nonetheless, previous research has demonstrated that HIEs can positively affect healthcare resource use and outcomes, including reduced rates of repeated diagnostic imaging in the emergency evaluation of back pain,[16] reduction in admissions via the emergency department (ED),[17] and reduced rates of readmissions within 7 days.[18] However, it is not known whether HIEs can contribute to positive outcomes when patients are readmitted to a different hospital than the hospital from which they were recently (within the previous 30 days) discharged, potentially bridging the transitional‐care information divide.
In Israel, an innovative HIE system, OFEK (literally horizon), was implemented in 2005 at the largest not‐for‐profit insurer and provider of services, Clalit Health Services (Clalit). Clalit operates as an integrated healthcare delivery system, serving more than 50% of the Israeli population, as part of the country's national health insurance system. OFEK links information on all Clalit enrollees from all hospitals, primary care, and specialty care clinics, laboratories, and diagnostic services into a single, virtual, patient file, enabling providers to obtain complete, real‐time information needed for healthcare decision making at the point of care. Like similar HIE systems, OFEK includes information on previous medical encounters and hospitalizations, previous diagnoses, chronically prescribed medications, previous lab and imaging tests, known allergies, and some demographic information.[19] At the time of this study, OFEK was available in all Clalit hospitals as well as in 2 non‐Clalit (government‐owned and operated) large tertiary‐care centers, resulting in 40% coverage of all hospitalizations through the OFEK HIE system. As part of a large organization‐wide readmission reduction program recently implemented by Clalit for all its members admitted to any hospital in Israel, aimed at early detection and intervention,[20] OFEK was viewed as an important mechanism to help maintain continuity and improve transitions.
To inform current knowledge on different‐hospital readmissions (DHRs) and HIEs, we examined whether having HIE systems can contribute to information continuity and prevent delays in care and the need for more expensive, lengthy readmission stays when patients are readmitted to a different hospital. More specifically, we tested whether there is a difference in the LORS between SHRs and DHRs, and whether DHRs the LORS differ by the availability of an HIE (whether index and readmitting hospital are or are not connected through HIE systems).
METHODS
Study Design and Setting
We conducted a retrospective cohort study based on data of hospitalized Clalit members. Clalit has a centralized data warehouse with a comprehensive EHR containing data on all patients' medical encounters, administrative data, and clinical data compiled from laboratories, imaging centers, and hospitals. At the time of the study, OFEK was operating in all 8 Clalit hospitals and in 2 large government‐owned and operated hospitals in the central and northern parts of the country. Information is linked in the Clalit system and OFEK‐affiliated hospitals through an individual identity number assigned by the Israeli Interior Ministry to every Israeli resident for general identification purposes.
Population
The study examined all internal medicine and intensive‐care unit (ICU) readmissions of adult Clalit members (aged 18 years and older) previously (within the prior 30 days) discharged from internal medicine departments during January 1, 2010 until December 31, 2010 (ie, index hospitalizations). Only readmissions of index hospitalizations with more than a 24‐hour stay were included. A total of 146,266 index hospitalizations met the inclusion criteria. Index admissions that resulted in a transfer to another hospital, a long‐term care facility, or rehabilitation center were not included (N = 11,831). The final study sample included 27,057 readmissions (20.1% of the 134,435 index admissions), which could have resulted in any type of discharge (to patient's home, a long‐term care or rehabilitation facility, or due to death). The study was approved by Clalit's institutional review board.
Outcome Variable
We defined the LORS as the number of days from admission to discharge during readmission.
Main Independent Variable
We assessed information continuity as a categorical variable in which 0 = no information continuity (DHRs with either no HIE at either hospital or an HIE in only 1 of the hospitals), 1 = information continuity through an HIE (DHRs with both hospitals having an HIE), and 2 = full information continuity (readmission to the same hospital).
Covariates
We examined the following known correlates of length of stay (LOS): age, gender, residency in a nursing home, socioeconomic status (SES) based on an indicator of social security entitlement received by low‐income members,[21] and the occurrence of common chronic conditions registered in Clalit's EHR registries[22]: congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), chronic renal failure (CRF), malignancy, diabetes, hypertension, ischemic heart disease, atrial fibrillation, asthma, and disability (indication of a functional limitation). To provide comorbidity adjustment we used the Charlson Comorbidity Index.[23] Additionally, we assessed LOS of the index hospitalization. We included an indicator for the size of the index hospital: small, fewer than 100 beds; medium, 100 to 200 beds; and large, more than 200 beds. Finally, to account for a well‐known correlate of length of hospital stay,[24] we included an indicator for an ICU stay during the readmission.
Statistical Analysis
We first examined the study populations' characteristics and calculated the average LORS for each SHR and DHR category. Due to the skewed distribution of LORS, we also calculated the median and interquartile range (IQR) of LORS and evaluated the difference between categories using the Kruskall‐Wallis test.[25] Sample‐size calculations showed that we would need a sample of >3000 admissions to have 80% power to detect a difference of 0.8 hospitalization days given the 1:3 ratio between the DHR groups. To examine the association between LORS and information continuity, we employed a univariate marginal Cox model.[26] Variables that were significantly (P < 0.05) associated with LORS in the univariate model were entered into a multivariate marginal Cox model, clustering by patient and using a robust sandwich covariance matrix estimate. Additionally, we performed a sensitivity analysis using hierarchichal modeling to account for potential variations due to hospital level factors. A low hazard ratio (<1) represented an association of the covariate with decreased likelihood of discharge, that is, longer LORS. All analyses were conducted with SPSS version 20 (IBM, Armonk, NY) and SAS version 9.3 (SAS Institute Inc., Cary, NC).
RESULTS
The study included a total of 27,057 readmissions, of which 23,927 (88.4%) were SHRs and 3130 (11.6%) were DHRs. Of all DHRs, in 792 (2.9%) of the cases, both hospitals had HIEs (partial information continuity), and in 2338 (8.6%), either 1 or both did not have an HIE system (thus meaning there was no information continuity). Characteristics of the study population are shown in Table 1. Most (75%) of the readmissions were of patients over the age of 65 years, though only 7% were nursing home residents. More than half the study's population consisted of patients with low SES. The most common chronic conditions were hypertension (77%), ischemic heart disease (52%), and diabetes (48%). Other chronic conditions were arrhythmia (38%), CHF (35%), disability (31%), COPD (28%), malignancy (28%), and asthma (16%). In more than 55% of the index hospitalizations, the LOS was 4 days or less, and most index admissions (64%) were in large hospitals. Table 1 also displays the study population by the type of readmission: SHR, DHR with HIE, and DHR without HIE. As compared to patients readmitted to the same hospital, patients with DHRs were younger (P < 0.001), less likely to be nursing home residents (P < 0.001), and had longer LOS during the index admission (P < 0.001). Additionally, patients with SHRs were more likely to have their index admission at a large hospital (P < 0.001), had a higher comorbidity score (P < 0.043), and were less likely be treated in the ICU during their readmission (P < 0.001) compared to their DHR counterparts. Patients with DHRs without an HIE were similar in most characteristics to those with an HIE, except for having an ICU stay during their readmission (6.4% compared with 9.2%, respectively).
| Characteristics | All Readmissions, n = 27,057 | SHR, n = 23,927 | DHR With HIE, n = 792 | DHR Without HIE, n = 2,338 | P Value |
|---|---|---|---|---|---|
| |||||
| All personal characteristics | |||||
| Age, n (%) | <0.001 | ||||
| 1844 years | 1,328 (4.9) | 1,095 (4.6) | 58 (7.3) | 175 (7.5) | |
| 4564 years | 5,370 (19.8) | 4,597 (19.2) | 197 (24.9) | 576 (24.6) | |
| 6584 years | 14,059 (52.0) | 12,500 (52.2) | 402 (50.8) | 1,157 (49.5) | |
| 85+ years | 6,300 (23.3) | 5,735 (24.0) | 135 (17.0) | 430 (18.4) | |
| Female sex, n (%) | 13,742 (50.8) | 12,040 (50.3) | 418 (52.8) | 1,284 (54.9) | <0.001 |
| Low socioeconomic status, n (%) | 15,473 (57.2) | 13,670 (57.1) | 453 (57.2) | 1,350 (57.7) | |
| Residency in a nursing home, n (%) | 1,857 (6.9) | 1,743 (7.3) | 27 (3.4) | 87 (3.7) | <0.001 |
| Common chronic conditions, n (%) | |||||
| Hypertension | 20,797 (76.9) | 18,484 (77.3) | 588 (74.2) | 1,725 (73.8) | <0.001 |
| Ischemic heart disease | 14,150 (52.3) | 12,577 (52.6) | 397 (50.1) | 1,176 (50.3) | 0.052 |
| Diabetes | 13,052 (48.2) | 11,589 (48.4) | 345 (43.6) | 1,118 (47.8) | 0.024 |
| Arrhythmia | 10,306 (38.1) | 9,197 (38.4) | 292 (36.9) | 817 (34.9) | 0.003 |
| Chronic renal failure | 9,486 (35.1) | 8,454 (35.3) | 262 (33.1) | 770 (32.9) | 0.034 |
| Congestive heart failure | 9,216 (34.1) | 8,232 (34.4) | 270 (34.1) | 714 (30.5) | 0.001 |
| Disability | 8,362 (30.9) | 7,600 (31.8) | 165 (20.8) | 597 (25.5) | <0.001 |
| Chronic obstructive pulmonary disease | 7,671 (28.4) | 6,888 (28.8) | 201 (25.4) | 582 (24.9) | <0.001 |
| Malignancy | 7,642 (28.2) | 6,763 (28.3) | 220 (27.8) | 659 (28.2) | 0.954 |
| Asthma | 4,491 (16.6) | 4,040 (16.9) | 109 (13.8) | 342 (14.6) | 0.002 |
| Charlson score, mean [SD] | 4.54 [3.15] | 4.58 [3.14] | 4.14 [3.08] | 4.25 [3.24] | 0.043 |
| Index hospitalization characteristics (LOS during index hospitalization), n (%) | <0.001 | ||||
| 24 days | 14,961 (55.3) | 13,310 (55.6) | 428 (54.0) | 1,223 (52.3) | |
| 57 days | 6,366 (23.5) | 5,654 (23.6) | 174 (22.0) | 538 (23.0) | |
| 8 days and more | 5,730 (21.2) | 4,963 (20.7) | 190 (24.0) | 577 (24.7) | |
| Hospital size in index hospitalization (no. of hospitals in each category), n (%) | <0.001 | ||||
| Small, <100 beds (8) | 1,498 (5.5) | 1,166 (4.9) | 23 (2.9) | 309 (13.2) | |
| Medium, 100200 beds (9) | 8,129 (30.0) | 7,113 (29.7) | 316 (39.9) | 700 (29.9) | |
| Large, >200 beds (10) | 17,430 (64.4) | 15,648 (65.4) | 453 (57.2) | 1,329 (56.8) | |
| Intensive care unit during readmission, n (%) | 869 (3.2) | 647 (2.7) | 73 (9.2) | 149 (6.4) | <0.001 |
The mean LORS in SHRs was shorter by 1 day than the mean LORS for DHRs: 6.3 (95% confidence interval [CI]: 6.2‐6.4) versus 7.3 (95% CI: 7.0‐7.6), respectively. Mean LORS in DHRs with or without HIE was 7.6 (95% CI: 7.0‐8.3) and 7.2 (95% CI: 6.8‐7.6), respectively. Although median LORS was similar (4 days), the IQR differed, resulting in significant differences between the SHR and DHR groups (Table 2).
| Information Continuity | No. of Readmissions | Mean LORS (95% CI) | Median (Q1Q3) | Kruskal‐Wallis P Value |
|---|---|---|---|---|
| ||||
| SHRs | 23,927 (88.4) | 6.3 (6.26.4) | 4 (27) | |
| DHRs | 3,130 (11.6) | 7.3 (7.07.6) | 4 (28) | |
| DHRs with HIE | 792 (2.9) | 7.6 (7.08.3) | 4 (29) | |
| DHRs without HIE | 2,338 (8.7) | 7.2 (6.87.6) | 4 (28) | |
| Total | 27,057 | 6.4 (6.36.5) | 4 (27) | <0.001 |
In the multivariate model, partial continuity (DHRs with an HIE) was associated with decreased likelihood of discharge on any given day compared with full continuity (SHR) (hazard ratio [HR] = 0.85, 95% CI: 0.79‐0.91). Similar results were obtained for no continuity (DHRs without an HIE) (HR = 0.90, 95% CI: 0.86‐0.94). The difference between DHRs with and without an HIE was not significant (overlapping confidence intervals). Other factors associated with a lower HR for discharge during each day of the readmission were older age, residency in a nursing home, CHF, CRF, disability, malignancy, and long LOS (8+ days) during the index hospitalization. Patients with asthma or ischemic heart disease had a higher HR for discharge during each readmission day (Table 3). We performed a sensitivity analysis using hierarchical modeling (patients nested within hospitals), which resulted in similar findings in terms of directionality and magnitude of the relationships and significance levels.
| Characteristics | Univariate Model | Multivariate Model | ||
|---|---|---|---|---|
| Hazard Ratio (95% CI) | P Value | Hazard Ratio (95% CI) | P Value | |
| ||||
| Information continuity | ||||
| SHR | Reference | Reference | ||
| DHR with HIE | 0.87 (0.810.93) | <0.001 | 0.86 (0.800.93) | <0.001 |
| DHR without HIE | 0.91 (0.870.94) | <0.001 | 0.90 (0.870.94) | <0.001 |
| Age | ||||
| 844 years | 1.22 (1.181.26) | <0.001 | 1.14 (1.071.22) | <0.001 |
| 4564 years | 1.16 (1.141.18) | <0.001 | 1.11 (1.061.1) | <0.001 |
| 6584 years | 1.01 (0.991.02) | 0.53 | 0.99 (0.961.02) | 0.60 |
| 85+ years | Reference | Reference | ||
| Sex | ||||
| Male | 0.97 (0.950.99) | 0.008 | 0.98 (0.961.01) | 0.19 |
| Female | Reference | Reference | ||
| Socioeconomic status | ||||
| Low | 0.98 (0.970.99) | 0.11 | ||
| Other | Reference | |||
| Residency in a nursing home | ||||
| Nursing home | 0.90 (0.880.92) | <0.001 | 0.90 (0.860.95) | <0.001 |
| All others | Reference | Reference | ||
| Common chronic conditions (reference: without condition) | ||||
| Hypertension | 0.94 (0.930.96) | <0.001 | 1.01 (0.971.04) | 0.69 |
| Ischemic heart disease | 1.00 (0.991.01) | 0.93 | 1.06 (1.031.09) | <0.001 |
| Diabetes | 0.97 (0.950.98) | 0.004 | 0.99 (0.971.02) | 0.64 |
| Arrhythmia | 0.96 (0.950.97) | 0.002 | 1.01 (0.981.04) | 0.39 |
| Chronic renal failure | 0.92 (0.910.93) | <0.001 | 0.96 (0.930.99) | 0.01 |
| Congestive heart failure | 0.93 (0.920.94) | <0.001 | 0.96 (0.930.99) | 0.01 |
| Disability | 0.86 (0.850.87) | <0.001 | 0.90 (0.870.92) | <0.001 |
| Chronic obstructive pulmonary disease | 0.99 (0.981.01) | 0.66 | ||
| Malignancy | 0.97 (0.960.98) | 0.03 | 0.98 (0.961.01) | 0.28 |
| Asthma | 1.04 (1.021.06) | 0.03 | 1.04 (1.001.07) | 0.03 |
| Charlson score | 0.99 (0.980.99) | <0.001 | 0.99 (0.991.00) | 0.04 |
| LOS during index hospitalization | ||||
| Days 24 | 1.52 (1.491.54) | <0.001 | 1.49 (1.451.54) | <0.001 |
| Days 57 | 1.21 (1.191.23) | <0.001 | 1.20 (1.161.24) | <0.001 |
| 8 days and more | Reference | Reference | ||
| Hospital size in index hospitalization | ||||
| Small, <100 beds (8) | 0.94 (0.920.97) | 0.02 | 1.00 (0.951.05) | 0.93 |
| Medium, 100200 beds (9) | 1.00 (0.991.02) | 0.78 | 1.01 (0.991.04) | 0.38 |
| Large, >200 beds (10) | Reference | Reference | ||
| Intensive care unit in readmission | ||||
| Yes | 0.75 (0.700.80) | <0.001 | 0.74 (0.690.79) | <0.001 |
| No | Reference | Reference | ||
DISCUSSION
This study shows that readmission to a different hospital results in longer duration of the readmission stay compared with readmission to the same index hospital. Our results also show that having HIE systems in both the index and readmitting hospitals does not protect against these negative outcomes, as there was no difference in the length of the readmission stay based on the availability of HIE systems. Factors that were found to be associated with longer readmission stays are well known indicators of the complexity of the patient's medical condition, such as the presence of disability, comorbidity, and ICU treatment during the readmission.[24, 27]
The shorter LORS in SHRs may be due to the familiarity of physicians and other healthcare providers with the patient and his or her condition, especially as the policy in SHRs in Israel is to readmit to the same unit from which the patient was recently discharged. This same hospital familiarity is especially important as hospital care in Israel follows the hospitalist model, in which responsibility for patient care is transferred from the patient's primary care physician to the hospital's physician, resulting in increased need for integration through HIE systems, especially when patients are readmitted to a different hospital.[28, 29]
Our findings, congruent with previous research on DHRs and poor outcomes,[7] could also be explained by the inefficiency associated with transitions. For example, patients frequently leave the hospital with pending lab tests, often with abnormal results that would change the course of care.[30] Because these pending tests are often omitted from the hospital discharge summaries,[31] if patients are hospitalized in a different hospital, the same tests may be ordered again, or a course of treatment that does not acknowledge the test results could be taken. Such time‐consuming duplication can be prevented in SHRs, where the index‐hospital records may be already more complete.
Our null findings regarding the contribution of HIE systems may be explained by the low levels of HIE actual use. Although we did not directly assess use, previous research reports that actual use of HIE is limited.[12] An Israeli study on the effects of the use of the OFEK system on ED physicians' admission decisions found that the patient's medical history was viewed in only 31.2% of all 281,750 ED referrals.[19] In another Israeli‐based ED study, even lower usage levels were found, with the OFEK system having been accessed in only 16% of all 3,219,910 ED referrals.[32] Low levels of HIE use have also been reported in the United States. An additional study, which tested the implementation of HIE in hospitals and clinics, showed that in only 2.3% of encounters did providers access the HIE record.[33] Another study conducted in 12 ED sites and 2 ambulatory clinics reported rates of 6.8% HIE use.[34] Moreover, the null effect of integrated health information reported here is congruent with findings from a US study on implementation of an electronic discharge instructions form with embedded computerized medication reconciliation, which was not found to be associated with postdischarge outcomes.[35]
A wide range of factors may influence decisions on HIE use: patient‐level factors,[36] perceived medical complexity of the patient,[33, 34] and the number of prior hospitalizations.[33, 34, 36] Healthcare systemlevel factors may include: time constraints, which may be positively[32] or negatively[33] associated with HIE use, and organizational policies or incentives.[33] Use may also be associated with features of the HIE technology itself, such as difficulty to access, difficulty to use once accessed, and the quality of information it contains.[37] Additionally, there is some evidence of the link between tight functional integration and higher proportions of usage.[38] Although comprehensive studies on factors affecting the use of the OFEK system in Israeli internal medicine units are still needed, the lack of its integration within each hospital's EHR system may serve as a major explanatory factor for the low usage levels.
The findings from this study should be interpreted in light of its limitations. First, compared with previously reported DHR rates (20%30%),[3, 5] the rate observed in our population was relatively low (about 12%). Previous research was restricted to heart failure patients[3] or assessed DHR in surgical, as well as internal medicine, patients.[5] Our lower rates may have been affected by the type of population (hospitalized internal medicine patients) and/or by characteristics of the Clalit healthcare system, which serves as an integrated provider network as well as insurer. Generalization from 1 health care system to others should be made with caution. Nonetheless, our results may underestimate the potential effect in other healthcare systems with less structural integration. Additionally, as noted above, information on the actual use of an HIE in the course of medical decision making during readmission was absent. Future studies should examine the potential benefit of an HIE with measures that capture providers' use of HIEs. Also, the LORS may be influenced by other factors not investigated here, and further future studies should examine additional outcomes such as costs, patient well‐being, and satisfaction. Finally, causality could not be determined, and future research in this realm should aim to search for the pathways connecting readmission to a different hospital, with and without HIEs, to readmission LOS and additional outcomes.
To conclude, our findings show that patients readmitted to a different hospital are at risk for prolonged LORS, regardless of the availability of HIE systems. Implementing HIE systems is the focus of substantial efforts by policymakers and is considered a key part of the meaningful use of electronic health information. HIE features in the provisions of the Health Information Technology for Economic and Clinical Health Act[39] because it can furnish providers with complete, timely information at the point of care. Moreover, although there has been substantial growth in the number of healthcare organizations that have operational an HIE, its ability to lead to improved outcomes has yet to be realized.[8, 10] The Israeli experience reported here suggests that provisions are needed that will ensure actual use of HIEs, which might in turn minimize the difference between DHRs and SHRs.
Acknowledgements
The authors acknowledge Chandra Cohen‐Stavi, MPA, and Orly Tonkikh, MA, for their contribution to this study.
Disclosures
The study was supported in part by a grant from the Israel National Institute for Health Policy Research (NIHP) (10/127). The authors report no conflicts of interest.
Readmissions within a relatively short time after discharge are receiving considerable attention as an area of quality improvement,[1, 2] with increasing emphasis on the relatively large share of readmissions to different hospitals, accounting for 20% to 30% of all readmissions.[3, 4, 5, 6] Returning to a different hospital may affect patient and healthcare outcomes due to breaches in continuity. When information from the previous recent hospitalization is not transferred efficiently and accurately to the next admitting hospital, omissions and duplications can occur, resulting in delayed care and potentially worse outcomes (compared to same hospital readmissions [SHRs]), such as longer length of readmission stay (LORS) and increased costs.[7]
Electronic health records (EHRs) and health information exchange (HIE) systems are increasingly used for storage and retrieval of patient information from various sources, such as laboratories and previous physician visits and hospitalizations, enabling informational continuity by providing vital historical medical information for decision‐making. Whereas EHRs collect, store, and present information that is locally created within a specific clinic or hospital, HIEs connect EHR systems between multiple institutions, allowing providers to share clinical data and achieve interorganizational continuity. Such integrative systems are increasingly being implemented across healthcare systems worldwide.[8, 9, 10] Yet, technical difficulties, costs, competitive concerns, data privacy, and workflow implementation challenges have been described as hindering HIE participation.[11, 12, 13, 14] Moreover, major concerns exist regarding the poor usability of EHRs, their limited ability to support multidisciplinary care, and major difficulties in achieving interoperability with HIEs, which undermine efforts to deliver integrated patient‐centered care.[15] Nonetheless, previous research has demonstrated that HIEs can positively affect healthcare resource use and outcomes, including reduced rates of repeated diagnostic imaging in the emergency evaluation of back pain,[16] reduction in admissions via the emergency department (ED),[17] and reduced rates of readmissions within 7 days.[18] However, it is not known whether HIEs can contribute to positive outcomes when patients are readmitted to a different hospital than the hospital from which they were recently (within the previous 30 days) discharged, potentially bridging the transitional‐care information divide.
In Israel, an innovative HIE system, OFEK (literally horizon), was implemented in 2005 at the largest not‐for‐profit insurer and provider of services, Clalit Health Services (Clalit). Clalit operates as an integrated healthcare delivery system, serving more than 50% of the Israeli population, as part of the country's national health insurance system. OFEK links information on all Clalit enrollees from all hospitals, primary care, and specialty care clinics, laboratories, and diagnostic services into a single, virtual, patient file, enabling providers to obtain complete, real‐time information needed for healthcare decision making at the point of care. Like similar HIE systems, OFEK includes information on previous medical encounters and hospitalizations, previous diagnoses, chronically prescribed medications, previous lab and imaging tests, known allergies, and some demographic information.[19] At the time of this study, OFEK was available in all Clalit hospitals as well as in 2 non‐Clalit (government‐owned and operated) large tertiary‐care centers, resulting in 40% coverage of all hospitalizations through the OFEK HIE system. As part of a large organization‐wide readmission reduction program recently implemented by Clalit for all its members admitted to any hospital in Israel, aimed at early detection and intervention,[20] OFEK was viewed as an important mechanism to help maintain continuity and improve transitions.
To inform current knowledge on different‐hospital readmissions (DHRs) and HIEs, we examined whether having HIE systems can contribute to information continuity and prevent delays in care and the need for more expensive, lengthy readmission stays when patients are readmitted to a different hospital. More specifically, we tested whether there is a difference in the LORS between SHRs and DHRs, and whether DHRs the LORS differ by the availability of an HIE (whether index and readmitting hospital are or are not connected through HIE systems).
METHODS
Study Design and Setting
We conducted a retrospective cohort study based on data of hospitalized Clalit members. Clalit has a centralized data warehouse with a comprehensive EHR containing data on all patients' medical encounters, administrative data, and clinical data compiled from laboratories, imaging centers, and hospitals. At the time of the study, OFEK was operating in all 8 Clalit hospitals and in 2 large government‐owned and operated hospitals in the central and northern parts of the country. Information is linked in the Clalit system and OFEK‐affiliated hospitals through an individual identity number assigned by the Israeli Interior Ministry to every Israeli resident for general identification purposes.
Population
The study examined all internal medicine and intensive‐care unit (ICU) readmissions of adult Clalit members (aged 18 years and older) previously (within the prior 30 days) discharged from internal medicine departments during January 1, 2010 until December 31, 2010 (ie, index hospitalizations). Only readmissions of index hospitalizations with more than a 24‐hour stay were included. A total of 146,266 index hospitalizations met the inclusion criteria. Index admissions that resulted in a transfer to another hospital, a long‐term care facility, or rehabilitation center were not included (N = 11,831). The final study sample included 27,057 readmissions (20.1% of the 134,435 index admissions), which could have resulted in any type of discharge (to patient's home, a long‐term care or rehabilitation facility, or due to death). The study was approved by Clalit's institutional review board.
Outcome Variable
We defined the LORS as the number of days from admission to discharge during readmission.
Main Independent Variable
We assessed information continuity as a categorical variable in which 0 = no information continuity (DHRs with either no HIE at either hospital or an HIE in only 1 of the hospitals), 1 = information continuity through an HIE (DHRs with both hospitals having an HIE), and 2 = full information continuity (readmission to the same hospital).
Covariates
We examined the following known correlates of length of stay (LOS): age, gender, residency in a nursing home, socioeconomic status (SES) based on an indicator of social security entitlement received by low‐income members,[21] and the occurrence of common chronic conditions registered in Clalit's EHR registries[22]: congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), chronic renal failure (CRF), malignancy, diabetes, hypertension, ischemic heart disease, atrial fibrillation, asthma, and disability (indication of a functional limitation). To provide comorbidity adjustment we used the Charlson Comorbidity Index.[23] Additionally, we assessed LOS of the index hospitalization. We included an indicator for the size of the index hospital: small, fewer than 100 beds; medium, 100 to 200 beds; and large, more than 200 beds. Finally, to account for a well‐known correlate of length of hospital stay,[24] we included an indicator for an ICU stay during the readmission.
Statistical Analysis
We first examined the study populations' characteristics and calculated the average LORS for each SHR and DHR category. Due to the skewed distribution of LORS, we also calculated the median and interquartile range (IQR) of LORS and evaluated the difference between categories using the Kruskall‐Wallis test.[25] Sample‐size calculations showed that we would need a sample of >3000 admissions to have 80% power to detect a difference of 0.8 hospitalization days given the 1:3 ratio between the DHR groups. To examine the association between LORS and information continuity, we employed a univariate marginal Cox model.[26] Variables that were significantly (P < 0.05) associated with LORS in the univariate model were entered into a multivariate marginal Cox model, clustering by patient and using a robust sandwich covariance matrix estimate. Additionally, we performed a sensitivity analysis using hierarchichal modeling to account for potential variations due to hospital level factors. A low hazard ratio (<1) represented an association of the covariate with decreased likelihood of discharge, that is, longer LORS. All analyses were conducted with SPSS version 20 (IBM, Armonk, NY) and SAS version 9.3 (SAS Institute Inc., Cary, NC).
RESULTS
The study included a total of 27,057 readmissions, of which 23,927 (88.4%) were SHRs and 3130 (11.6%) were DHRs. Of all DHRs, in 792 (2.9%) of the cases, both hospitals had HIEs (partial information continuity), and in 2338 (8.6%), either 1 or both did not have an HIE system (thus meaning there was no information continuity). Characteristics of the study population are shown in Table 1. Most (75%) of the readmissions were of patients over the age of 65 years, though only 7% were nursing home residents. More than half the study's population consisted of patients with low SES. The most common chronic conditions were hypertension (77%), ischemic heart disease (52%), and diabetes (48%). Other chronic conditions were arrhythmia (38%), CHF (35%), disability (31%), COPD (28%), malignancy (28%), and asthma (16%). In more than 55% of the index hospitalizations, the LOS was 4 days or less, and most index admissions (64%) were in large hospitals. Table 1 also displays the study population by the type of readmission: SHR, DHR with HIE, and DHR without HIE. As compared to patients readmitted to the same hospital, patients with DHRs were younger (P < 0.001), less likely to be nursing home residents (P < 0.001), and had longer LOS during the index admission (P < 0.001). Additionally, patients with SHRs were more likely to have their index admission at a large hospital (P < 0.001), had a higher comorbidity score (P < 0.043), and were less likely be treated in the ICU during their readmission (P < 0.001) compared to their DHR counterparts. Patients with DHRs without an HIE were similar in most characteristics to those with an HIE, except for having an ICU stay during their readmission (6.4% compared with 9.2%, respectively).
| Characteristics | All Readmissions, n = 27,057 | SHR, n = 23,927 | DHR With HIE, n = 792 | DHR Without HIE, n = 2,338 | P Value |
|---|---|---|---|---|---|
| |||||
| All personal characteristics | |||||
| Age, n (%) | <0.001 | ||||
| 1844 years | 1,328 (4.9) | 1,095 (4.6) | 58 (7.3) | 175 (7.5) | |
| 4564 years | 5,370 (19.8) | 4,597 (19.2) | 197 (24.9) | 576 (24.6) | |
| 6584 years | 14,059 (52.0) | 12,500 (52.2) | 402 (50.8) | 1,157 (49.5) | |
| 85+ years | 6,300 (23.3) | 5,735 (24.0) | 135 (17.0) | 430 (18.4) | |
| Female sex, n (%) | 13,742 (50.8) | 12,040 (50.3) | 418 (52.8) | 1,284 (54.9) | <0.001 |
| Low socioeconomic status, n (%) | 15,473 (57.2) | 13,670 (57.1) | 453 (57.2) | 1,350 (57.7) | |
| Residency in a nursing home, n (%) | 1,857 (6.9) | 1,743 (7.3) | 27 (3.4) | 87 (3.7) | <0.001 |
| Common chronic conditions, n (%) | |||||
| Hypertension | 20,797 (76.9) | 18,484 (77.3) | 588 (74.2) | 1,725 (73.8) | <0.001 |
| Ischemic heart disease | 14,150 (52.3) | 12,577 (52.6) | 397 (50.1) | 1,176 (50.3) | 0.052 |
| Diabetes | 13,052 (48.2) | 11,589 (48.4) | 345 (43.6) | 1,118 (47.8) | 0.024 |
| Arrhythmia | 10,306 (38.1) | 9,197 (38.4) | 292 (36.9) | 817 (34.9) | 0.003 |
| Chronic renal failure | 9,486 (35.1) | 8,454 (35.3) | 262 (33.1) | 770 (32.9) | 0.034 |
| Congestive heart failure | 9,216 (34.1) | 8,232 (34.4) | 270 (34.1) | 714 (30.5) | 0.001 |
| Disability | 8,362 (30.9) | 7,600 (31.8) | 165 (20.8) | 597 (25.5) | <0.001 |
| Chronic obstructive pulmonary disease | 7,671 (28.4) | 6,888 (28.8) | 201 (25.4) | 582 (24.9) | <0.001 |
| Malignancy | 7,642 (28.2) | 6,763 (28.3) | 220 (27.8) | 659 (28.2) | 0.954 |
| Asthma | 4,491 (16.6) | 4,040 (16.9) | 109 (13.8) | 342 (14.6) | 0.002 |
| Charlson score, mean [SD] | 4.54 [3.15] | 4.58 [3.14] | 4.14 [3.08] | 4.25 [3.24] | 0.043 |
| Index hospitalization characteristics (LOS during index hospitalization), n (%) | <0.001 | ||||
| 24 days | 14,961 (55.3) | 13,310 (55.6) | 428 (54.0) | 1,223 (52.3) | |
| 57 days | 6,366 (23.5) | 5,654 (23.6) | 174 (22.0) | 538 (23.0) | |
| 8 days and more | 5,730 (21.2) | 4,963 (20.7) | 190 (24.0) | 577 (24.7) | |
| Hospital size in index hospitalization (no. of hospitals in each category), n (%) | <0.001 | ||||
| Small, <100 beds (8) | 1,498 (5.5) | 1,166 (4.9) | 23 (2.9) | 309 (13.2) | |
| Medium, 100200 beds (9) | 8,129 (30.0) | 7,113 (29.7) | 316 (39.9) | 700 (29.9) | |
| Large, >200 beds (10) | 17,430 (64.4) | 15,648 (65.4) | 453 (57.2) | 1,329 (56.8) | |
| Intensive care unit during readmission, n (%) | 869 (3.2) | 647 (2.7) | 73 (9.2) | 149 (6.4) | <0.001 |
The mean LORS in SHRs was shorter by 1 day than the mean LORS for DHRs: 6.3 (95% confidence interval [CI]: 6.2‐6.4) versus 7.3 (95% CI: 7.0‐7.6), respectively. Mean LORS in DHRs with or without HIE was 7.6 (95% CI: 7.0‐8.3) and 7.2 (95% CI: 6.8‐7.6), respectively. Although median LORS was similar (4 days), the IQR differed, resulting in significant differences between the SHR and DHR groups (Table 2).
| Information Continuity | No. of Readmissions | Mean LORS (95% CI) | Median (Q1Q3) | Kruskal‐Wallis P Value |
|---|---|---|---|---|
| ||||
| SHRs | 23,927 (88.4) | 6.3 (6.26.4) | 4 (27) | |
| DHRs | 3,130 (11.6) | 7.3 (7.07.6) | 4 (28) | |
| DHRs with HIE | 792 (2.9) | 7.6 (7.08.3) | 4 (29) | |
| DHRs without HIE | 2,338 (8.7) | 7.2 (6.87.6) | 4 (28) | |
| Total | 27,057 | 6.4 (6.36.5) | 4 (27) | <0.001 |
In the multivariate model, partial continuity (DHRs with an HIE) was associated with decreased likelihood of discharge on any given day compared with full continuity (SHR) (hazard ratio [HR] = 0.85, 95% CI: 0.79‐0.91). Similar results were obtained for no continuity (DHRs without an HIE) (HR = 0.90, 95% CI: 0.86‐0.94). The difference between DHRs with and without an HIE was not significant (overlapping confidence intervals). Other factors associated with a lower HR for discharge during each day of the readmission were older age, residency in a nursing home, CHF, CRF, disability, malignancy, and long LOS (8+ days) during the index hospitalization. Patients with asthma or ischemic heart disease had a higher HR for discharge during each readmission day (Table 3). We performed a sensitivity analysis using hierarchical modeling (patients nested within hospitals), which resulted in similar findings in terms of directionality and magnitude of the relationships and significance levels.
| Characteristics | Univariate Model | Multivariate Model | ||
|---|---|---|---|---|
| Hazard Ratio (95% CI) | P Value | Hazard Ratio (95% CI) | P Value | |
| ||||
| Information continuity | ||||
| SHR | Reference | Reference | ||
| DHR with HIE | 0.87 (0.810.93) | <0.001 | 0.86 (0.800.93) | <0.001 |
| DHR without HIE | 0.91 (0.870.94) | <0.001 | 0.90 (0.870.94) | <0.001 |
| Age | ||||
| 844 years | 1.22 (1.181.26) | <0.001 | 1.14 (1.071.22) | <0.001 |
| 4564 years | 1.16 (1.141.18) | <0.001 | 1.11 (1.061.1) | <0.001 |
| 6584 years | 1.01 (0.991.02) | 0.53 | 0.99 (0.961.02) | 0.60 |
| 85+ years | Reference | Reference | ||
| Sex | ||||
| Male | 0.97 (0.950.99) | 0.008 | 0.98 (0.961.01) | 0.19 |
| Female | Reference | Reference | ||
| Socioeconomic status | ||||
| Low | 0.98 (0.970.99) | 0.11 | ||
| Other | Reference | |||
| Residency in a nursing home | ||||
| Nursing home | 0.90 (0.880.92) | <0.001 | 0.90 (0.860.95) | <0.001 |
| All others | Reference | Reference | ||
| Common chronic conditions (reference: without condition) | ||||
| Hypertension | 0.94 (0.930.96) | <0.001 | 1.01 (0.971.04) | 0.69 |
| Ischemic heart disease | 1.00 (0.991.01) | 0.93 | 1.06 (1.031.09) | <0.001 |
| Diabetes | 0.97 (0.950.98) | 0.004 | 0.99 (0.971.02) | 0.64 |
| Arrhythmia | 0.96 (0.950.97) | 0.002 | 1.01 (0.981.04) | 0.39 |
| Chronic renal failure | 0.92 (0.910.93) | <0.001 | 0.96 (0.930.99) | 0.01 |
| Congestive heart failure | 0.93 (0.920.94) | <0.001 | 0.96 (0.930.99) | 0.01 |
| Disability | 0.86 (0.850.87) | <0.001 | 0.90 (0.870.92) | <0.001 |
| Chronic obstructive pulmonary disease | 0.99 (0.981.01) | 0.66 | ||
| Malignancy | 0.97 (0.960.98) | 0.03 | 0.98 (0.961.01) | 0.28 |
| Asthma | 1.04 (1.021.06) | 0.03 | 1.04 (1.001.07) | 0.03 |
| Charlson score | 0.99 (0.980.99) | <0.001 | 0.99 (0.991.00) | 0.04 |
| LOS during index hospitalization | ||||
| Days 24 | 1.52 (1.491.54) | <0.001 | 1.49 (1.451.54) | <0.001 |
| Days 57 | 1.21 (1.191.23) | <0.001 | 1.20 (1.161.24) | <0.001 |
| 8 days and more | Reference | Reference | ||
| Hospital size in index hospitalization | ||||
| Small, <100 beds (8) | 0.94 (0.920.97) | 0.02 | 1.00 (0.951.05) | 0.93 |
| Medium, 100200 beds (9) | 1.00 (0.991.02) | 0.78 | 1.01 (0.991.04) | 0.38 |
| Large, >200 beds (10) | Reference | Reference | ||
| Intensive care unit in readmission | ||||
| Yes | 0.75 (0.700.80) | <0.001 | 0.74 (0.690.79) | <0.001 |
| No | Reference | Reference | ||
DISCUSSION
This study shows that readmission to a different hospital results in longer duration of the readmission stay compared with readmission to the same index hospital. Our results also show that having HIE systems in both the index and readmitting hospitals does not protect against these negative outcomes, as there was no difference in the length of the readmission stay based on the availability of HIE systems. Factors that were found to be associated with longer readmission stays are well known indicators of the complexity of the patient's medical condition, such as the presence of disability, comorbidity, and ICU treatment during the readmission.[24, 27]
The shorter LORS in SHRs may be due to the familiarity of physicians and other healthcare providers with the patient and his or her condition, especially as the policy in SHRs in Israel is to readmit to the same unit from which the patient was recently discharged. This same hospital familiarity is especially important as hospital care in Israel follows the hospitalist model, in which responsibility for patient care is transferred from the patient's primary care physician to the hospital's physician, resulting in increased need for integration through HIE systems, especially when patients are readmitted to a different hospital.[28, 29]
Our findings, congruent with previous research on DHRs and poor outcomes,[7] could also be explained by the inefficiency associated with transitions. For example, patients frequently leave the hospital with pending lab tests, often with abnormal results that would change the course of care.[30] Because these pending tests are often omitted from the hospital discharge summaries,[31] if patients are hospitalized in a different hospital, the same tests may be ordered again, or a course of treatment that does not acknowledge the test results could be taken. Such time‐consuming duplication can be prevented in SHRs, where the index‐hospital records may be already more complete.
Our null findings regarding the contribution of HIE systems may be explained by the low levels of HIE actual use. Although we did not directly assess use, previous research reports that actual use of HIE is limited.[12] An Israeli study on the effects of the use of the OFEK system on ED physicians' admission decisions found that the patient's medical history was viewed in only 31.2% of all 281,750 ED referrals.[19] In another Israeli‐based ED study, even lower usage levels were found, with the OFEK system having been accessed in only 16% of all 3,219,910 ED referrals.[32] Low levels of HIE use have also been reported in the United States. An additional study, which tested the implementation of HIE in hospitals and clinics, showed that in only 2.3% of encounters did providers access the HIE record.[33] Another study conducted in 12 ED sites and 2 ambulatory clinics reported rates of 6.8% HIE use.[34] Moreover, the null effect of integrated health information reported here is congruent with findings from a US study on implementation of an electronic discharge instructions form with embedded computerized medication reconciliation, which was not found to be associated with postdischarge outcomes.[35]
A wide range of factors may influence decisions on HIE use: patient‐level factors,[36] perceived medical complexity of the patient,[33, 34] and the number of prior hospitalizations.[33, 34, 36] Healthcare systemlevel factors may include: time constraints, which may be positively[32] or negatively[33] associated with HIE use, and organizational policies or incentives.[33] Use may also be associated with features of the HIE technology itself, such as difficulty to access, difficulty to use once accessed, and the quality of information it contains.[37] Additionally, there is some evidence of the link between tight functional integration and higher proportions of usage.[38] Although comprehensive studies on factors affecting the use of the OFEK system in Israeli internal medicine units are still needed, the lack of its integration within each hospital's EHR system may serve as a major explanatory factor for the low usage levels.
The findings from this study should be interpreted in light of its limitations. First, compared with previously reported DHR rates (20%30%),[3, 5] the rate observed in our population was relatively low (about 12%). Previous research was restricted to heart failure patients[3] or assessed DHR in surgical, as well as internal medicine, patients.[5] Our lower rates may have been affected by the type of population (hospitalized internal medicine patients) and/or by characteristics of the Clalit healthcare system, which serves as an integrated provider network as well as insurer. Generalization from 1 health care system to others should be made with caution. Nonetheless, our results may underestimate the potential effect in other healthcare systems with less structural integration. Additionally, as noted above, information on the actual use of an HIE in the course of medical decision making during readmission was absent. Future studies should examine the potential benefit of an HIE with measures that capture providers' use of HIEs. Also, the LORS may be influenced by other factors not investigated here, and further future studies should examine additional outcomes such as costs, patient well‐being, and satisfaction. Finally, causality could not be determined, and future research in this realm should aim to search for the pathways connecting readmission to a different hospital, with and without HIEs, to readmission LOS and additional outcomes.
To conclude, our findings show that patients readmitted to a different hospital are at risk for prolonged LORS, regardless of the availability of HIE systems. Implementing HIE systems is the focus of substantial efforts by policymakers and is considered a key part of the meaningful use of electronic health information. HIE features in the provisions of the Health Information Technology for Economic and Clinical Health Act[39] because it can furnish providers with complete, timely information at the point of care. Moreover, although there has been substantial growth in the number of healthcare organizations that have operational an HIE, its ability to lead to improved outcomes has yet to be realized.[8, 10] The Israeli experience reported here suggests that provisions are needed that will ensure actual use of HIEs, which might in turn minimize the difference between DHRs and SHRs.
Acknowledgements
The authors acknowledge Chandra Cohen‐Stavi, MPA, and Orly Tonkikh, MA, for their contribution to this study.
Disclosures
The study was supported in part by a grant from the Israel National Institute for Health Policy Research (NIHP) (10/127). The authors report no conflicts of interest.
- , , , , , . Assessing preventability in the quest to reduce hospital readmissions. J Hosp Med. 2014;9:598–603.
- , , , et al. Preventing 30‐day hospital readmissions: a systematic review and meta‐analysis of randomized trials. JAMA Intern Med. 2014;174:1095–1107.
- , , , et al. Is same‐hospital readmission rate a good surrogate for all‐hospital readmission rate? Med Care. 2010;48:477–481.
- , , , . Hospital readmission rates: the impacts of age, payer, and mental health diagnoses. J Ambul Care Manage. 2013;36(2):147–155.
- , , , . Limitations of using same‐hospital readmission metrics. Int J Qual Health Care. 2013;25(6):633–639.
- Hospital inpatient and outpatient services. In: Report to the Congress: promoting greater efficiency in Medicare. Washington, DC: Medicare Payment Advisory Commission., March 2012;45–66.
- , , , , . For‐profit hospital status and rehospitalizations at different hospitals: an analysis of Medicare data. Ann Intern Med. 2010;153:718–727.
- , , , . Health information technology: an updated systematic review with a focus on meaningful use. Ann Intern Med. 2014;160:48–54.
- , , , . The benefits of health information technology: a review of the recent literature shows predominantly positive results. Health Aff (Millwood). 2011;30(3):464–471.
- , , . Health information exchange among US hospitals. Am J Manag Care. 2011;17:761–768.
- , , . A survey of health information exchange organizations in the United States: implications for meaningful use. Ann Intern Med. 2011;54:666–671.
- , , , , . Physicians' potential use and preferences related to health information exchange. Int J Med Inform. 2011;80:171–180.
- , , , et al. Provider stakeholders' perceived benefit from a nascent health information exchange: a qualitative analysis. J Med Syst. 2012;36:601–613.
- . More than just a question of technology: factors related to hospitals' adoption and implementation of health information exchange. Int J Med Inform. 2010;79:797–806.
- , , . Leveraging health information technology to achieve the “triple aim” of healthcare reform. J Am Med Inform Assoc. 2015;22(4):849–856.
- , , , , , . Health information exchange reduces repeated diagnostic imaging for back pain. Ann Emerg Med. 2013;62:16–24.
- , , , , . Association between use of a health information exchange system and hospital admissions. Appl Clin Inform. 2014;5:219.
- , , . The impact of EHR and HIE on reducing avoidable admissions: controlling main differential diagnoses. BMC Med Inform Decis Mak. 2013;13:49.
- , , , et al. The impact of an integrated hospital‐community medical information system on quality and service utilization in hospital departments. Int J Med Inform. 2010;79(9):649–657.
- , , , , , Predicting 30‐day readmissions with preadmission electronic health record data. Med Care. 2015;53:283–289.
- , , , , . Assessing socioeconomic health care utilization inequity in Israel: impact of alternative approaches to morbidity adjustment. BMC Public Health. 2011;11(1):609.
- , . Prevalence of selected chronic diseases in Israel. Isr Med Assoc J. 2001;3:404–408.
- , , , . A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis. 1987;40:373–383.
- , , , , . Systematic review of risk adjustment models of hospital length of stay (LOS). Med Care. 2015;53:355–365.
- , . Use of ranks in one‐criterion variance analysis. J Am Stat Assoc. 1952;47:583–621.
- , , . Regression analysis of multivariate incomplete failure time data by modeling marginal distributions. J Am Stat Assoc. 1989;84:1065–1073.
- , , , , , . Disability impacts length of stay in general internal medicine patients. J Gen Intern Med. 2014;29:885–890.
- . The hospitalist movement—time to move on. N Engl J Med. 2007;357:2627–2629.
- , , . Association of hospitalist presence and hospital‐level outcome measures among Medicare patients. J Hosp Med. 2014;9:1–6.
- , , , et al. Patient safety concerns arising from test results that return after hospital discharge. Ann Intern Med. 2005;143:121–128.
- , , , et al. Adequacy of hospital discharge summaries in documenting tests with pending results and outpatient follow‐up providers. J Gen Intern Med. 2009;24:1002–1006.
- , , . Using electronic medical record systems for admission decisions in emergency departments: examining the crowdedness effect. J Med Syst. 2012;36:3795–3803.
- , , , , , . Factors motivating and affecting health information exchange usage. J Am Med Inform Assoc. 2011;18(2):143–149.
- , , , et al. Health information exchange usage in emergency departments and clinics: the who, what, and why. J Am Med Inform Assoc. 2011;18:690–697.
- , , , , . Effect of standardized electronic discharge instructions on post‐discharge hospital utilization. J Gen Intern Med. 2011;26:718–723.
- , , . Health information exchange technology on the front lines of healthcare: workflow factors and patterns of use. J Am Med Inform Assoc. 2012;19:392–400.
- , . The DeLone and McLean model of information systems success: a ten‐year update. J Manag Inf Syst. 2003;19:9–30.
- , , , et al. Architectural strategies and issues with health information exchange. AMIA Annu Symp Proc. 2006:814–818.
- . Launching HITECH. N Engl J Med. 2010;362(5):382–385.
- , , , , , . Assessing preventability in the quest to reduce hospital readmissions. J Hosp Med. 2014;9:598–603.
- , , , et al. Preventing 30‐day hospital readmissions: a systematic review and meta‐analysis of randomized trials. JAMA Intern Med. 2014;174:1095–1107.
- , , , et al. Is same‐hospital readmission rate a good surrogate for all‐hospital readmission rate? Med Care. 2010;48:477–481.
- , , , . Hospital readmission rates: the impacts of age, payer, and mental health diagnoses. J Ambul Care Manage. 2013;36(2):147–155.
- , , , . Limitations of using same‐hospital readmission metrics. Int J Qual Health Care. 2013;25(6):633–639.
- Hospital inpatient and outpatient services. In: Report to the Congress: promoting greater efficiency in Medicare. Washington, DC: Medicare Payment Advisory Commission., March 2012;45–66.
- , , , , . For‐profit hospital status and rehospitalizations at different hospitals: an analysis of Medicare data. Ann Intern Med. 2010;153:718–727.
- , , , . Health information technology: an updated systematic review with a focus on meaningful use. Ann Intern Med. 2014;160:48–54.
- , , , . The benefits of health information technology: a review of the recent literature shows predominantly positive results. Health Aff (Millwood). 2011;30(3):464–471.
- , , . Health information exchange among US hospitals. Am J Manag Care. 2011;17:761–768.
- , , . A survey of health information exchange organizations in the United States: implications for meaningful use. Ann Intern Med. 2011;54:666–671.
- , , , , . Physicians' potential use and preferences related to health information exchange. Int J Med Inform. 2011;80:171–180.
- , , , et al. Provider stakeholders' perceived benefit from a nascent health information exchange: a qualitative analysis. J Med Syst. 2012;36:601–613.
- . More than just a question of technology: factors related to hospitals' adoption and implementation of health information exchange. Int J Med Inform. 2010;79:797–806.
- , , . Leveraging health information technology to achieve the “triple aim” of healthcare reform. J Am Med Inform Assoc. 2015;22(4):849–856.
- , , , , , . Health information exchange reduces repeated diagnostic imaging for back pain. Ann Emerg Med. 2013;62:16–24.
- , , , , . Association between use of a health information exchange system and hospital admissions. Appl Clin Inform. 2014;5:219.
- , , . The impact of EHR and HIE on reducing avoidable admissions: controlling main differential diagnoses. BMC Med Inform Decis Mak. 2013;13:49.
- , , , et al. The impact of an integrated hospital‐community medical information system on quality and service utilization in hospital departments. Int J Med Inform. 2010;79(9):649–657.
- , , , , , Predicting 30‐day readmissions with preadmission electronic health record data. Med Care. 2015;53:283–289.
- , , , , . Assessing socioeconomic health care utilization inequity in Israel: impact of alternative approaches to morbidity adjustment. BMC Public Health. 2011;11(1):609.
- , . Prevalence of selected chronic diseases in Israel. Isr Med Assoc J. 2001;3:404–408.
- , , , . A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis. 1987;40:373–383.
- , , , , . Systematic review of risk adjustment models of hospital length of stay (LOS). Med Care. 2015;53:355–365.
- , . Use of ranks in one‐criterion variance analysis. J Am Stat Assoc. 1952;47:583–621.
- , , . Regression analysis of multivariate incomplete failure time data by modeling marginal distributions. J Am Stat Assoc. 1989;84:1065–1073.
- , , , , , . Disability impacts length of stay in general internal medicine patients. J Gen Intern Med. 2014;29:885–890.
- . The hospitalist movement—time to move on. N Engl J Med. 2007;357:2627–2629.
- , , . Association of hospitalist presence and hospital‐level outcome measures among Medicare patients. J Hosp Med. 2014;9:1–6.
- , , , et al. Patient safety concerns arising from test results that return after hospital discharge. Ann Intern Med. 2005;143:121–128.
- , , , et al. Adequacy of hospital discharge summaries in documenting tests with pending results and outpatient follow‐up providers. J Gen Intern Med. 2009;24:1002–1006.
- , , . Using electronic medical record systems for admission decisions in emergency departments: examining the crowdedness effect. J Med Syst. 2012;36:3795–3803.
- , , , , , . Factors motivating and affecting health information exchange usage. J Am Med Inform Assoc. 2011;18(2):143–149.
- , , , et al. Health information exchange usage in emergency departments and clinics: the who, what, and why. J Am Med Inform Assoc. 2011;18:690–697.
- , , , , . Effect of standardized electronic discharge instructions on post‐discharge hospital utilization. J Gen Intern Med. 2011;26:718–723.
- , , . Health information exchange technology on the front lines of healthcare: workflow factors and patterns of use. J Am Med Inform Assoc. 2012;19:392–400.
- , . The DeLone and McLean model of information systems success: a ten‐year update. J Manag Inf Syst. 2003;19:9–30.
- , , , et al. Architectural strategies and issues with health information exchange. AMIA Annu Symp Proc. 2006:814–818.
- . Launching HITECH. N Engl J Med. 2010;362(5):382–385.
© 2015 Society of Hospital Medicine
Female Athletes: Unique Challenges Facing Women Warriors
Since Title IX passed in 1972, women have become exponentially more involved in competitive sports, from high school to professional levels. With more women engaging in serious athletics, the specific challenges they face have come to the forefront of sports medicine. These problems include the female athlete triad, concussions, exercise safety in pregnancy, anterior cruciate ligament (ACL) injuries, and continued sex discrimination and social injustice. Orthopedists treating female athletes should be aware of these problems, each of which is discussed in this review.
1. Female athlete triad
In 1992, the term female athlete triad was coined to describe 3 problems that often coexist in high-intensity female athletes.1 Since then, the definition has evolved, but the problem has remained essentially the same. The modern definition incorporates menstrual abnormalities, low energy availability with or without disordered eating, and decreased bone mineral density (BMD).2
With intense exercise and weight loss comes a variety of menstrual disturbances.3 In affected athletes, the hypothalamus is underactivated, and changes in gonadotropin-releasing hormone and luteinizing hormone lead to decreased estrogen production. Research suggests abnormal menses result from having inadequate energy and insufficient caloric intake to support extensive exercise.1 This phenomenon can occur in athletes in any sport but is most prevalent in lean-body sports, such as swimming, gymnastics, and ballet. The incidence of abnormal menses is as high as 79% in ballet dancers but only 5% in the general population.3 Menstrual abnormalities indicate hormonal abnormalities that can interfere with growth and maturation in young athletes.
Although full-blown eating disorders are uncommon among female athletes, disordered eating patterns are often found among women in competitive sports. Disordered eating can involve a spectrum of inadequate caloric intake and purging behavior, such as vomiting or laxative abuse, and has been reported in up to 25% of collegiate female athletes.4 Physicians must recognize these conditions and initiate counseling and treatment when appropriate. Women with disordered eating are at risk for developing electrolyte imbalances, malnutrition syndromes, and osteopenia.
Although careful evaluation and counseling are important, physicians must note that, in most cases, athletics participation may also protect against disordered eating and body image difficulties. A study of 146 college-age women found better body satisfaction among athletes than among nonathletes.5 Lean-sport athletes (eg, swimmers, gymnasts) were at higher risk for disordered eating and body image problems than other athletes were. Similarly, other studies have found that a majority of athletes have healthy eating habits.4
For poorly nourished and hormonally imbalanced female athletes, decreased BMD poses substantial risk. One study found a significant difference in BMD between athletes with amenorrhea and athletes with normal menses.6 In a cohort of female Navy recruits, those with amenorrhea were at 91% higher risk for stress fractures; calcium and vitamin D supplementation reduced risk by 20%.7 Osteopenia may be a special problem for prepubescent athletes. Girls who engage in intense exercise and have delayed menarche may have a low estrogen state, predisposing them to low BMD.3 Osteopenia and osteoporosis are difficult to reverse and can put these athletes at risk for stress fractures the rest of their lives. If unrecognized, stress fractures can end an athlete’s career.
Recommendations for dual-energy X-ray absorptiometry (DXA) include testing female athletes who have a diagnosed eating disorder, body mass index under 17.5, history of delayed menarche, oligomenorrhea, 2 prior stress fractures, or prior abnormal DXA scan. Complete testing recommendations appear in the 2014 consensus statement on the female athlete triad and return to sport.2,8
Orthopedists performing physical examinations for sports participation can screen for the female athlete triad through thoughtful questioning about menstrual history, nutrition habits, and stress fracture symptoms. Best treatment for a diagnosed case of the triad is multidisciplinary care with strong social support. When abnormal menses are an issue, referral to a gynecologist or endocrinologist and consideration of estrogen replacement should be discussed. Some cases require a psychiatrist’s assistance in treating disordered eating. Athletic trainers, coaches, and parents should be involved over the treatment course.1 Orthopedists must counsel women with osteopenia and osteoporosis about decreasing exercise to a safe level, improving nutritional intake, and supplementing with calcium (1200-1500 mg/d) and vitamin D (600-800 IU/d).3,7
2. Concussions
Increasing awareness of males’ sport-related concussions, particularly of concussions that occur during National Football League practice and games, has made physicians and researchers more aware of the rate of concussion in female athletes. That rate has increased, and, according to some reports, the risk for sport-related injury is higher for female athletes.9 A study of high school athletes found that the rate of concussion in girl’s soccer was second only to that in football.10
Concussions are categorized as mild traumatic brain injuries, and manifestations of the diagnosis are divided into physical, emotional, cognitive, and observed symptoms. The spectrum of symptoms is wide, ranging from difficulty concentrating and thinking clearly to headaches and dizziness.11 Compared with male athletes who sustain a concussion, female athletes report more of these concussive symptoms and have worse visual memory scores.12
Efforts to change sports at the player level have been resisted. Helmets have been proposed for field hockey and lacrosse but have not passed stringent concussion testing. In soccer, which has a high rate of concussion, a reform to eliminate heading the ball has been considered. Resistance to these suggestions stems from the thought that changes could alter the traditions of the games. Some individuals have indicated that helmets may give players a false sense of security and thereby cause them to play more aggressively.
Orthopedic surgeons must be aware of concussion symptoms. Multiple concussions may have a cumulative effect on functional ability and emotional well-being and may lead to chronic traumatic encephalopathy.13 Concern about the long-term effects of concussion has led to the implementation of universal “return to play” laws. These laws vary by state but have 3 steps in common: Educate coaches, players, and athletes; remove athletes from play; and obtain health care professionals’ permission to return to play.14 These guidelines set up an action plan for treating an athlete who has sustained a concussion.
Encouraging results of educating coaches have been noted. Coaches who were given Centers for Disease Control and Prevention–sponsored material on preventing, recognizing, and responding to concussions were able to effectively address concussions; 6 months later, 63% were better able to appreciate the severity of concussions.15 Continued education of athletic communities should help bring this injury to the attention of those treating female athletes.
3. Exercise safety in pregnancy
Women in sports can continue their athletic regimens during pregnancy. It is important to address challenges to the pregnant woman and to the fetus when assessing the risks of exercise.
The physiologic changes that occur during pregnancy may affect how a pregnant athlete responds to stress. Plasma volume, red blood cell volume, and cardiac function and output all increase during normal pregnancy.3,16 Abnormal heart rate during pregnancy can adversely affect the fetus. During and after exercise, fetal bradycardia can occur. Therefore, recommendations should include not exceeding pre-pregnancy activity levels.3 Careful monitoring of exercise intensity is recommended by the American College of Obstetrics and Gynecology; the guideline is to maintain less than 70% of maximal heart rate.17,18
The negative effects of exercise on the pregnant athlete are limited, but it is important to educate patients and to consider preventive strategies. One physiologic change that occurs during pregnancy is ligamentous laxity, which is caused by the hormone relaxin.16 Ligamentous laxity has the potential to put pregnant athletes at risk for soft-tissue and bony injury during impact sports. However, the positive effects of exercise during pregnancy include improved appetite, sleep, and emotional health.19 Aerobic exercise during pregnancy may reverse insulin resistance as demonstrated in animal studies; though this outcome has not been demonstrated in human studies,20 women should be reassured that moderate exercise has overall beneficial effects.
Some research suggests that exercise may expose the fetus to hyperthermia, blood sugar changes, physical injury, and premature labor.16 Typically, fetal heat is dissipated from the mother. After intense exercise, maternal body temperature rises and leads to some degree of fetal hyperthermia.16 Animal model studies have suggested that hyperthermia may result in a slightly higher rate of congenital abnormalities. Pregnant women should keep their exercise routines to less than 60 minutes, should exercise in a thermally regulated environment, and should keep themselves hydrated to avoid fetal hyperthermia.18
Reduced blood flow, accompanied by a deficit of oxygen to the uterus and the developing fetus, is another concern for pregnant athletes. During exercise, when more blood is flowing to the muscles, less is flowing to the uterus.16 Furthermore, during the third trimester, women should avoid supine exercise, as venous outflow is poor with the body in that position.21
Elite athletes who continue training during pregnancy should be carefully counseled about adjusting their training regimens. Because of increased cardiac output and blood volume, the heart rate will be lower than usual, demanding an adjustment in interpretation. Blood cell counts do not increase as much as plasma volume does—often leading to relative anemia. For elite athletes, this means iron supplementation is crucial.22 Thermal regulation may be more difficult, as training regimens may demand prolonged exercise. Physicians should recommend adequate hydration for these athletes.18
Although continued exercise is generally safe for a pregnant athlete and her fetus, caution is required when there is increased risk for premature delivery, or other special conditions exist. Multiple gestation, placenta previa, history of early labor or premature births, and incompetent cervix all contraindicate aerobic exercise during pregnancy.18 With these exceptions in mind, physicians can safely counsel pregnant women to do moderate exercise 30 minutes every day.17,18 Other recommendations are listed at the American College of Obstetricians and Gynecologists website.23
4. Anterior cruciate ligament injuries
ACL injuries affect a staggering number of athletes. In the United States, approximately 100,000 people sustain these injuries annually.24 As they occur up to 8 times more often in women than in men, ACL injuries are a top concern for physicians treating female athletes.
This disproportionate injury rate is influenced by differences between male and female anatomy. The width and shape of the femoral intercondylar notch have been studied as potential variables influencing the risk for ACL injury. Analysis of notch-view radiographs revealed a significant inverse relationship between notch width and ACL injury.25 A-shaped notches, notches with a significantly larger base and a narrowed roof, were more prevalent in women but did not correlate with increased risk for ACL injury. Studies have shown that female athletes with a noncontact ACL injury have a higher lateral tibial plateau posterior slope; this slope is associated with increased peak anteromedial ACL strain, which may contribute to injury.26 An analysis of magnetic resonance imaging scans in patients with and without ACL injury revealed that, for female patients, decreased femoral intercondylar notch width at the anterior outlet combined with increased lateral compartment posterior slope correlated best with risk for ACL injury.27
Although static anatomical factors contribute to ACL injuries in female athletes, dynamic neuromuscular influences are potential opportunities for intervention. Female athletes with high relative quadriceps strength and weak hamstring strength may be at increased risk for ACL injury.28 This “quadriceps dominance” becomes important in sports involving high-risk activities, such as running, cutting, pivoting, and jumping. In addition, compared with male athletes, female athletes demonstrate increased lateral trunk motion and knee valgus torque while landing during noncontact ACL tears, making core stability a factor in ACL injury.29
The collaborative efforts of physicians, physical therapists, athletic trainers, and coaches have yielded multifactorial neuromuscular training programs for the prevention of noncontact ACL injuries. Ideal ACL prevention protocols involve sessions that last for at least 10 minutes and take place 3 times a week. At these sessions, exercises are focused on strengthening, balance, and proprioceptive training.30 The programs last about 8 weeks, but sustained benefits require maintenance after the program has been completed and during the off-season. Program adherence must be encouraged and can be facilitated by varying workouts and raising risk awareness. The most effective programs have reduced the relative risk of noncontact ACL injuries by 75% to 100%.31 These promising results have led to increased focus on program implementation in an effort to prevent ACL injury.
5. Continued sex discrimination and social injustice
In 1972, Title IX was passed as part of the Education Amendments Act. Title IX states, “No person in the United States shall, on the basis of sex, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any educational program or activity receiving Federal financial assistance.” Passage of this law, which has implications outside of athletic participation, marked an important turning point in women’s ability to participate equally in college sports.32,33 The Civil Rights Restoration Act, passed in 1988, strengthened Title IX and made it applicable to all institutions receiving federal funding.34 Before the 1970s, women typically were restricted to club sports, and funding and participation opportunities were weighted heavily toward men. Over the past 40 years, women’s participation in high school, college, and professional sports has taken a huge leap forward.32 For example, the number of women participating in high school sports increased from 294,000 (7.4% of all athletes) in 1972 to 3.4 million (>41% of all athletes) in 2014.
Despite advances in women’s civil rights, examples of inequality in US schools remain, particularly in the distribution of funding, which still strongly favors men’s football.32 Men’s sports receive 90% of media coverage.33 In 2002, women represented 55% of college students but only 42% of varsity athletes.34 The schools that have complied the least with Title IX are schools in the Midwest and the South and those with football teams.34 Women are underrepresented as coaches, and funding continues to be disproportionately spent on men’s sports.
For women, the benefits of participating in sports are far-reaching and significant. These benefits include improvements in academic success, mental health, and responsible behavior.33 Women’s gaining acceptance and respect throughout the athletic world seems to have carried over elsewhere. Although many institutions remain noncompliant with Title IX, efforts continue to have a strongly positive effect on gender equality in the United States.
1. Nattiv A, Loucks AB, Manore MM, Sanborn CF, Sundgot-Borgen J, Warren MP; American College of Sports Medicine. American College of Sports Medicine position stand. The female athlete triad. Med Sci Sports Exerc. 2007;39(10):1867-1882.
2. De Souza MJ, Nattiv A, Joy E, et al; Expert Panel. 2014 Female Athlete Triad Coalition consensus statement on treatment and return to play of the female athlete triad: 1st international conference held in San Francisco, California, May 2012 and 2nd international conference held in Indianapolis, Indiana, May 2013. Br J Sports Med. 2014;48(4):289.
3. Warren MP, Shantha S. The female athlete. Baillieres Best Pract Res Clin Endocrinol Metab. 2000;14(1):37-53.
4. Greenleaf C, Petrie TA, Carter J, Reel JJ. Female collegiate athletes: prevalence of eating disorders and disordered eating behaviors. J Am Coll Health. 2009;57(5):489-495.
5. Reinking MF, Alexander LE. Prevalence of disordered-eating behaviors in undergraduate female collegiate athletes and nonathletes. J Athl Train. 2005;40(1):47-51.
6. Rencken ML, Chesnut CH 3rd, Drinkwater BL. Bone density at multiple skeletal sites in amenorrheic athletes. JAMA. 1996;276(3):238-240.
7. Lappe J, Cullen D, Haynatzki G, Recker R, Ahlf R, Thompson K. Calcium and vitamin D supplementation decreases incidence of stress fractures in female Navy recruits. J Bone Miner Res. 2008;23(5):741-749.
8. De Souza MJ. 2014 Female athlete triad consensus statement on guidelines for treatment and return to play. National Collegiate Athletic Association (NCAA) website. http://www.ncaa.org/health-and-safety/nutrition-and-performance/2014-female-athlete-triad-consensus-statement-guidelines. Accessed November 24, 2015.
9. Preiss-Farzanegan SJ, Chapman B, Wong TM, Wu J, Bazarian JJ. The relationship between gender and postconcussion symptoms after sport-related mild traumatic brain injury. PM R. 2009;1(3):245-253.
10. Marar M, McIlvain NM, Fields SK, Comstock RD. Epidemiology of concussions among United States high school athletes in 20 sports. Am J Sports Med. 2012;40(4):747-755.
11. Uhl RL, Rosenbaum AJ, Czajka C, Mulligan M, King C. Minor traumatic brain injury: a primer for the orthopaedic surgeon. J Am Acad Orthop Surg. 2013;21(10):624-631.
12. Covassin T, Elbin RJ, Harris W, Parker T, Kontos A. The role of age and sex in symptoms, neurocognitive performance, and postural stability in athletes after concussion. Am J Sports Med. 2012;40(6):1303-1312.
13. Covassin T, Moran R, Wilhelm K. Concussion symptoms and neurocognitive performance of high school and college athletes who incur multiple concussions. Am J Sports Med. 2013;41(12):2885-2889.
14. Sports concussion policies and laws: information for parents, coaches, and school & sports professionals. Centers for Disease Control and Prevention website. http://www.cdc.gov/headsup/policy/index.html. Updated February 16, 2015. Accessed November 24, 2015.
15. Covassin T, Elbin RJ, Sarmiento K. Educating coaches about concussion in sports: evaluation of the CDC’s “Heads Up: concussion in youth sports” initiative. J Sch Health. 2012;82(5):233-238.
16. Lumbers ER. Exercise in pregnancy: physiological basis of exercise prescription for the pregnant woman. J Sci Med Sport. 2002;5(1):20-31.
17. ACOG Committee Obstetric Practice. ACOG Committee opinion. Number 267, January 2002: exercise during pregnancy and the postpartum period. Obstet Gynecol. 2002;99(1):171-173.
18. Artal R, O’Toole M. Guidelines of the American College of Obstetricians and Gynecologists for exercise during pregnancy and the postpartum period. Br J Sports Med. 2003;37(1):6-12.
19. Kramer MS. Regular aerobic exercise during pregnancy. Cochrane Database Syst Rev. 2000;(2):CD000180. Update in: Cochrane Database Syst Rev. 2002;(2):CD000180.
20. Stafne SN, Salvesen KA, Romundstad PR, Stuge B, Morkved S. Does regular exercise during pregnancy influence lumbopelvic pain? A randomized controlled trial. Acta Obstet Gynecol Scand. 2012;91(5):552-559.
21. Nascimento SL, Surita FG, Cecatti JG. Physical exercise during pregnancy: a systematic review. Curr Opin Obstet Gynecol. 2012;24(6):387-394.
22. Hale RW, Milne L. The elite athlete and exercise in pregnancy. Semin Perinatol. 1996;20(4):277-284.
23. Exercise during pregnancy. American College of Obstetricians and Gynecologists website. http://www.acog.org/Patients/FAQs/Exercise-During-Pregnancy. Published August 2011. Accessed November 24, 2015.
24. Giugliano DN, Solomon JL. ACL tears in female athletes. Phys Med Rehabil Clin North Am. 2007;18(3):417-438, viii.
25. Ireland ML, Ballantyne BT, Little K, McClay IS. A radiographic analysis of the relationship between the size and shape of the intercondylar notch and anterior cruciate ligament injury. Knee Surg Sports Traumatol Arthrosc. 2001;9(4):200-205.
26. Lipps DB, Oh YK, Ashton-Miller JA, Wojtys EM. Morphologic characteristics help explain the gender difference in peak anterior cruciate ligament strain during a simulated pivot landing. Am J Sports Med. 2012;40(1):32-40.
27. Sturnick DR, Vacek PM, DeSarno MJ, et al. Combined anatomic factors predicting risk of anterior cruciate ligament injury for males and females. Am J Sports Med. 2015;43(4):839-847.
28. Myer GD, Ford KR, Barber Foss KD, Liu C, Nick TG, Hewett TE. The relationship of hamstrings and quadriceps strength to anterior cruciate ligament injury in female athletes. Clin J Sport Med. 2009;19(1):3-8.
29. Hewett TE, Torg JS, Boden BP. Video analysis of trunk and knee motion during non-contact anterior cruciate ligament injury in female athletes: lateral trunk and knee abduction motion are combined components of the injury mechanism. Br J Sports Med. 2009;43(6):417-422.
30. Sutton KM, Bullock JM. Anterior cruciate ligament rupture: differences between males and females. J Am Acad Orthop Surg. 2013;21(1):41-50.
31. Noyes FR, Barber-Westin SD. Neuromuscular retraining intervention programs: do they reduce noncontact anterior cruciate ligament injury rates in adolescent female athletes? Arthroscopy. 2014;30(2):245-255.
32. Ladd AL. The sports bra, the ACL, and Title IX—the game in play. Clin Orthop Relat Res. 2014;472(6):1681-1684.
33. Lopiano DA. Modern history of women in sports. Twenty-five years of Title IX. Clin Sports Med. 2000;19(2):163-173, vii.
34. Anderson DJ, Cheslock JJ, Ehrenberg RG. Gender equity in intercollegiate athletics: determinants of Title IX compliance. J High Educ. 2006;77(2):225-250.
Since Title IX passed in 1972, women have become exponentially more involved in competitive sports, from high school to professional levels. With more women engaging in serious athletics, the specific challenges they face have come to the forefront of sports medicine. These problems include the female athlete triad, concussions, exercise safety in pregnancy, anterior cruciate ligament (ACL) injuries, and continued sex discrimination and social injustice. Orthopedists treating female athletes should be aware of these problems, each of which is discussed in this review.
1. Female athlete triad
In 1992, the term female athlete triad was coined to describe 3 problems that often coexist in high-intensity female athletes.1 Since then, the definition has evolved, but the problem has remained essentially the same. The modern definition incorporates menstrual abnormalities, low energy availability with or without disordered eating, and decreased bone mineral density (BMD).2
With intense exercise and weight loss comes a variety of menstrual disturbances.3 In affected athletes, the hypothalamus is underactivated, and changes in gonadotropin-releasing hormone and luteinizing hormone lead to decreased estrogen production. Research suggests abnormal menses result from having inadequate energy and insufficient caloric intake to support extensive exercise.1 This phenomenon can occur in athletes in any sport but is most prevalent in lean-body sports, such as swimming, gymnastics, and ballet. The incidence of abnormal menses is as high as 79% in ballet dancers but only 5% in the general population.3 Menstrual abnormalities indicate hormonal abnormalities that can interfere with growth and maturation in young athletes.
Although full-blown eating disorders are uncommon among female athletes, disordered eating patterns are often found among women in competitive sports. Disordered eating can involve a spectrum of inadequate caloric intake and purging behavior, such as vomiting or laxative abuse, and has been reported in up to 25% of collegiate female athletes.4 Physicians must recognize these conditions and initiate counseling and treatment when appropriate. Women with disordered eating are at risk for developing electrolyte imbalances, malnutrition syndromes, and osteopenia.
Although careful evaluation and counseling are important, physicians must note that, in most cases, athletics participation may also protect against disordered eating and body image difficulties. A study of 146 college-age women found better body satisfaction among athletes than among nonathletes.5 Lean-sport athletes (eg, swimmers, gymnasts) were at higher risk for disordered eating and body image problems than other athletes were. Similarly, other studies have found that a majority of athletes have healthy eating habits.4
For poorly nourished and hormonally imbalanced female athletes, decreased BMD poses substantial risk. One study found a significant difference in BMD between athletes with amenorrhea and athletes with normal menses.6 In a cohort of female Navy recruits, those with amenorrhea were at 91% higher risk for stress fractures; calcium and vitamin D supplementation reduced risk by 20%.7 Osteopenia may be a special problem for prepubescent athletes. Girls who engage in intense exercise and have delayed menarche may have a low estrogen state, predisposing them to low BMD.3 Osteopenia and osteoporosis are difficult to reverse and can put these athletes at risk for stress fractures the rest of their lives. If unrecognized, stress fractures can end an athlete’s career.
Recommendations for dual-energy X-ray absorptiometry (DXA) include testing female athletes who have a diagnosed eating disorder, body mass index under 17.5, history of delayed menarche, oligomenorrhea, 2 prior stress fractures, or prior abnormal DXA scan. Complete testing recommendations appear in the 2014 consensus statement on the female athlete triad and return to sport.2,8
Orthopedists performing physical examinations for sports participation can screen for the female athlete triad through thoughtful questioning about menstrual history, nutrition habits, and stress fracture symptoms. Best treatment for a diagnosed case of the triad is multidisciplinary care with strong social support. When abnormal menses are an issue, referral to a gynecologist or endocrinologist and consideration of estrogen replacement should be discussed. Some cases require a psychiatrist’s assistance in treating disordered eating. Athletic trainers, coaches, and parents should be involved over the treatment course.1 Orthopedists must counsel women with osteopenia and osteoporosis about decreasing exercise to a safe level, improving nutritional intake, and supplementing with calcium (1200-1500 mg/d) and vitamin D (600-800 IU/d).3,7
2. Concussions
Increasing awareness of males’ sport-related concussions, particularly of concussions that occur during National Football League practice and games, has made physicians and researchers more aware of the rate of concussion in female athletes. That rate has increased, and, according to some reports, the risk for sport-related injury is higher for female athletes.9 A study of high school athletes found that the rate of concussion in girl’s soccer was second only to that in football.10
Concussions are categorized as mild traumatic brain injuries, and manifestations of the diagnosis are divided into physical, emotional, cognitive, and observed symptoms. The spectrum of symptoms is wide, ranging from difficulty concentrating and thinking clearly to headaches and dizziness.11 Compared with male athletes who sustain a concussion, female athletes report more of these concussive symptoms and have worse visual memory scores.12
Efforts to change sports at the player level have been resisted. Helmets have been proposed for field hockey and lacrosse but have not passed stringent concussion testing. In soccer, which has a high rate of concussion, a reform to eliminate heading the ball has been considered. Resistance to these suggestions stems from the thought that changes could alter the traditions of the games. Some individuals have indicated that helmets may give players a false sense of security and thereby cause them to play more aggressively.
Orthopedic surgeons must be aware of concussion symptoms. Multiple concussions may have a cumulative effect on functional ability and emotional well-being and may lead to chronic traumatic encephalopathy.13 Concern about the long-term effects of concussion has led to the implementation of universal “return to play” laws. These laws vary by state but have 3 steps in common: Educate coaches, players, and athletes; remove athletes from play; and obtain health care professionals’ permission to return to play.14 These guidelines set up an action plan for treating an athlete who has sustained a concussion.
Encouraging results of educating coaches have been noted. Coaches who were given Centers for Disease Control and Prevention–sponsored material on preventing, recognizing, and responding to concussions were able to effectively address concussions; 6 months later, 63% were better able to appreciate the severity of concussions.15 Continued education of athletic communities should help bring this injury to the attention of those treating female athletes.
3. Exercise safety in pregnancy
Women in sports can continue their athletic regimens during pregnancy. It is important to address challenges to the pregnant woman and to the fetus when assessing the risks of exercise.
The physiologic changes that occur during pregnancy may affect how a pregnant athlete responds to stress. Plasma volume, red blood cell volume, and cardiac function and output all increase during normal pregnancy.3,16 Abnormal heart rate during pregnancy can adversely affect the fetus. During and after exercise, fetal bradycardia can occur. Therefore, recommendations should include not exceeding pre-pregnancy activity levels.3 Careful monitoring of exercise intensity is recommended by the American College of Obstetrics and Gynecology; the guideline is to maintain less than 70% of maximal heart rate.17,18
The negative effects of exercise on the pregnant athlete are limited, but it is important to educate patients and to consider preventive strategies. One physiologic change that occurs during pregnancy is ligamentous laxity, which is caused by the hormone relaxin.16 Ligamentous laxity has the potential to put pregnant athletes at risk for soft-tissue and bony injury during impact sports. However, the positive effects of exercise during pregnancy include improved appetite, sleep, and emotional health.19 Aerobic exercise during pregnancy may reverse insulin resistance as demonstrated in animal studies; though this outcome has not been demonstrated in human studies,20 women should be reassured that moderate exercise has overall beneficial effects.
Some research suggests that exercise may expose the fetus to hyperthermia, blood sugar changes, physical injury, and premature labor.16 Typically, fetal heat is dissipated from the mother. After intense exercise, maternal body temperature rises and leads to some degree of fetal hyperthermia.16 Animal model studies have suggested that hyperthermia may result in a slightly higher rate of congenital abnormalities. Pregnant women should keep their exercise routines to less than 60 minutes, should exercise in a thermally regulated environment, and should keep themselves hydrated to avoid fetal hyperthermia.18
Reduced blood flow, accompanied by a deficit of oxygen to the uterus and the developing fetus, is another concern for pregnant athletes. During exercise, when more blood is flowing to the muscles, less is flowing to the uterus.16 Furthermore, during the third trimester, women should avoid supine exercise, as venous outflow is poor with the body in that position.21
Elite athletes who continue training during pregnancy should be carefully counseled about adjusting their training regimens. Because of increased cardiac output and blood volume, the heart rate will be lower than usual, demanding an adjustment in interpretation. Blood cell counts do not increase as much as plasma volume does—often leading to relative anemia. For elite athletes, this means iron supplementation is crucial.22 Thermal regulation may be more difficult, as training regimens may demand prolonged exercise. Physicians should recommend adequate hydration for these athletes.18
Although continued exercise is generally safe for a pregnant athlete and her fetus, caution is required when there is increased risk for premature delivery, or other special conditions exist. Multiple gestation, placenta previa, history of early labor or premature births, and incompetent cervix all contraindicate aerobic exercise during pregnancy.18 With these exceptions in mind, physicians can safely counsel pregnant women to do moderate exercise 30 minutes every day.17,18 Other recommendations are listed at the American College of Obstetricians and Gynecologists website.23
4. Anterior cruciate ligament injuries
ACL injuries affect a staggering number of athletes. In the United States, approximately 100,000 people sustain these injuries annually.24 As they occur up to 8 times more often in women than in men, ACL injuries are a top concern for physicians treating female athletes.
This disproportionate injury rate is influenced by differences between male and female anatomy. The width and shape of the femoral intercondylar notch have been studied as potential variables influencing the risk for ACL injury. Analysis of notch-view radiographs revealed a significant inverse relationship between notch width and ACL injury.25 A-shaped notches, notches with a significantly larger base and a narrowed roof, were more prevalent in women but did not correlate with increased risk for ACL injury. Studies have shown that female athletes with a noncontact ACL injury have a higher lateral tibial plateau posterior slope; this slope is associated with increased peak anteromedial ACL strain, which may contribute to injury.26 An analysis of magnetic resonance imaging scans in patients with and without ACL injury revealed that, for female patients, decreased femoral intercondylar notch width at the anterior outlet combined with increased lateral compartment posterior slope correlated best with risk for ACL injury.27
Although static anatomical factors contribute to ACL injuries in female athletes, dynamic neuromuscular influences are potential opportunities for intervention. Female athletes with high relative quadriceps strength and weak hamstring strength may be at increased risk for ACL injury.28 This “quadriceps dominance” becomes important in sports involving high-risk activities, such as running, cutting, pivoting, and jumping. In addition, compared with male athletes, female athletes demonstrate increased lateral trunk motion and knee valgus torque while landing during noncontact ACL tears, making core stability a factor in ACL injury.29
The collaborative efforts of physicians, physical therapists, athletic trainers, and coaches have yielded multifactorial neuromuscular training programs for the prevention of noncontact ACL injuries. Ideal ACL prevention protocols involve sessions that last for at least 10 minutes and take place 3 times a week. At these sessions, exercises are focused on strengthening, balance, and proprioceptive training.30 The programs last about 8 weeks, but sustained benefits require maintenance after the program has been completed and during the off-season. Program adherence must be encouraged and can be facilitated by varying workouts and raising risk awareness. The most effective programs have reduced the relative risk of noncontact ACL injuries by 75% to 100%.31 These promising results have led to increased focus on program implementation in an effort to prevent ACL injury.
5. Continued sex discrimination and social injustice
In 1972, Title IX was passed as part of the Education Amendments Act. Title IX states, “No person in the United States shall, on the basis of sex, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any educational program or activity receiving Federal financial assistance.” Passage of this law, which has implications outside of athletic participation, marked an important turning point in women’s ability to participate equally in college sports.32,33 The Civil Rights Restoration Act, passed in 1988, strengthened Title IX and made it applicable to all institutions receiving federal funding.34 Before the 1970s, women typically were restricted to club sports, and funding and participation opportunities were weighted heavily toward men. Over the past 40 years, women’s participation in high school, college, and professional sports has taken a huge leap forward.32 For example, the number of women participating in high school sports increased from 294,000 (7.4% of all athletes) in 1972 to 3.4 million (>41% of all athletes) in 2014.
Despite advances in women’s civil rights, examples of inequality in US schools remain, particularly in the distribution of funding, which still strongly favors men’s football.32 Men’s sports receive 90% of media coverage.33 In 2002, women represented 55% of college students but only 42% of varsity athletes.34 The schools that have complied the least with Title IX are schools in the Midwest and the South and those with football teams.34 Women are underrepresented as coaches, and funding continues to be disproportionately spent on men’s sports.
For women, the benefits of participating in sports are far-reaching and significant. These benefits include improvements in academic success, mental health, and responsible behavior.33 Women’s gaining acceptance and respect throughout the athletic world seems to have carried over elsewhere. Although many institutions remain noncompliant with Title IX, efforts continue to have a strongly positive effect on gender equality in the United States.
Since Title IX passed in 1972, women have become exponentially more involved in competitive sports, from high school to professional levels. With more women engaging in serious athletics, the specific challenges they face have come to the forefront of sports medicine. These problems include the female athlete triad, concussions, exercise safety in pregnancy, anterior cruciate ligament (ACL) injuries, and continued sex discrimination and social injustice. Orthopedists treating female athletes should be aware of these problems, each of which is discussed in this review.
1. Female athlete triad
In 1992, the term female athlete triad was coined to describe 3 problems that often coexist in high-intensity female athletes.1 Since then, the definition has evolved, but the problem has remained essentially the same. The modern definition incorporates menstrual abnormalities, low energy availability with or without disordered eating, and decreased bone mineral density (BMD).2
With intense exercise and weight loss comes a variety of menstrual disturbances.3 In affected athletes, the hypothalamus is underactivated, and changes in gonadotropin-releasing hormone and luteinizing hormone lead to decreased estrogen production. Research suggests abnormal menses result from having inadequate energy and insufficient caloric intake to support extensive exercise.1 This phenomenon can occur in athletes in any sport but is most prevalent in lean-body sports, such as swimming, gymnastics, and ballet. The incidence of abnormal menses is as high as 79% in ballet dancers but only 5% in the general population.3 Menstrual abnormalities indicate hormonal abnormalities that can interfere with growth and maturation in young athletes.
Although full-blown eating disorders are uncommon among female athletes, disordered eating patterns are often found among women in competitive sports. Disordered eating can involve a spectrum of inadequate caloric intake and purging behavior, such as vomiting or laxative abuse, and has been reported in up to 25% of collegiate female athletes.4 Physicians must recognize these conditions and initiate counseling and treatment when appropriate. Women with disordered eating are at risk for developing electrolyte imbalances, malnutrition syndromes, and osteopenia.
Although careful evaluation and counseling are important, physicians must note that, in most cases, athletics participation may also protect against disordered eating and body image difficulties. A study of 146 college-age women found better body satisfaction among athletes than among nonathletes.5 Lean-sport athletes (eg, swimmers, gymnasts) were at higher risk for disordered eating and body image problems than other athletes were. Similarly, other studies have found that a majority of athletes have healthy eating habits.4
For poorly nourished and hormonally imbalanced female athletes, decreased BMD poses substantial risk. One study found a significant difference in BMD between athletes with amenorrhea and athletes with normal menses.6 In a cohort of female Navy recruits, those with amenorrhea were at 91% higher risk for stress fractures; calcium and vitamin D supplementation reduced risk by 20%.7 Osteopenia may be a special problem for prepubescent athletes. Girls who engage in intense exercise and have delayed menarche may have a low estrogen state, predisposing them to low BMD.3 Osteopenia and osteoporosis are difficult to reverse and can put these athletes at risk for stress fractures the rest of their lives. If unrecognized, stress fractures can end an athlete’s career.
Recommendations for dual-energy X-ray absorptiometry (DXA) include testing female athletes who have a diagnosed eating disorder, body mass index under 17.5, history of delayed menarche, oligomenorrhea, 2 prior stress fractures, or prior abnormal DXA scan. Complete testing recommendations appear in the 2014 consensus statement on the female athlete triad and return to sport.2,8
Orthopedists performing physical examinations for sports participation can screen for the female athlete triad through thoughtful questioning about menstrual history, nutrition habits, and stress fracture symptoms. Best treatment for a diagnosed case of the triad is multidisciplinary care with strong social support. When abnormal menses are an issue, referral to a gynecologist or endocrinologist and consideration of estrogen replacement should be discussed. Some cases require a psychiatrist’s assistance in treating disordered eating. Athletic trainers, coaches, and parents should be involved over the treatment course.1 Orthopedists must counsel women with osteopenia and osteoporosis about decreasing exercise to a safe level, improving nutritional intake, and supplementing with calcium (1200-1500 mg/d) and vitamin D (600-800 IU/d).3,7
2. Concussions
Increasing awareness of males’ sport-related concussions, particularly of concussions that occur during National Football League practice and games, has made physicians and researchers more aware of the rate of concussion in female athletes. That rate has increased, and, according to some reports, the risk for sport-related injury is higher for female athletes.9 A study of high school athletes found that the rate of concussion in girl’s soccer was second only to that in football.10
Concussions are categorized as mild traumatic brain injuries, and manifestations of the diagnosis are divided into physical, emotional, cognitive, and observed symptoms. The spectrum of symptoms is wide, ranging from difficulty concentrating and thinking clearly to headaches and dizziness.11 Compared with male athletes who sustain a concussion, female athletes report more of these concussive symptoms and have worse visual memory scores.12
Efforts to change sports at the player level have been resisted. Helmets have been proposed for field hockey and lacrosse but have not passed stringent concussion testing. In soccer, which has a high rate of concussion, a reform to eliminate heading the ball has been considered. Resistance to these suggestions stems from the thought that changes could alter the traditions of the games. Some individuals have indicated that helmets may give players a false sense of security and thereby cause them to play more aggressively.
Orthopedic surgeons must be aware of concussion symptoms. Multiple concussions may have a cumulative effect on functional ability and emotional well-being and may lead to chronic traumatic encephalopathy.13 Concern about the long-term effects of concussion has led to the implementation of universal “return to play” laws. These laws vary by state but have 3 steps in common: Educate coaches, players, and athletes; remove athletes from play; and obtain health care professionals’ permission to return to play.14 These guidelines set up an action plan for treating an athlete who has sustained a concussion.
Encouraging results of educating coaches have been noted. Coaches who were given Centers for Disease Control and Prevention–sponsored material on preventing, recognizing, and responding to concussions were able to effectively address concussions; 6 months later, 63% were better able to appreciate the severity of concussions.15 Continued education of athletic communities should help bring this injury to the attention of those treating female athletes.
3. Exercise safety in pregnancy
Women in sports can continue their athletic regimens during pregnancy. It is important to address challenges to the pregnant woman and to the fetus when assessing the risks of exercise.
The physiologic changes that occur during pregnancy may affect how a pregnant athlete responds to stress. Plasma volume, red blood cell volume, and cardiac function and output all increase during normal pregnancy.3,16 Abnormal heart rate during pregnancy can adversely affect the fetus. During and after exercise, fetal bradycardia can occur. Therefore, recommendations should include not exceeding pre-pregnancy activity levels.3 Careful monitoring of exercise intensity is recommended by the American College of Obstetrics and Gynecology; the guideline is to maintain less than 70% of maximal heart rate.17,18
The negative effects of exercise on the pregnant athlete are limited, but it is important to educate patients and to consider preventive strategies. One physiologic change that occurs during pregnancy is ligamentous laxity, which is caused by the hormone relaxin.16 Ligamentous laxity has the potential to put pregnant athletes at risk for soft-tissue and bony injury during impact sports. However, the positive effects of exercise during pregnancy include improved appetite, sleep, and emotional health.19 Aerobic exercise during pregnancy may reverse insulin resistance as demonstrated in animal studies; though this outcome has not been demonstrated in human studies,20 women should be reassured that moderate exercise has overall beneficial effects.
Some research suggests that exercise may expose the fetus to hyperthermia, blood sugar changes, physical injury, and premature labor.16 Typically, fetal heat is dissipated from the mother. After intense exercise, maternal body temperature rises and leads to some degree of fetal hyperthermia.16 Animal model studies have suggested that hyperthermia may result in a slightly higher rate of congenital abnormalities. Pregnant women should keep their exercise routines to less than 60 minutes, should exercise in a thermally regulated environment, and should keep themselves hydrated to avoid fetal hyperthermia.18
Reduced blood flow, accompanied by a deficit of oxygen to the uterus and the developing fetus, is another concern for pregnant athletes. During exercise, when more blood is flowing to the muscles, less is flowing to the uterus.16 Furthermore, during the third trimester, women should avoid supine exercise, as venous outflow is poor with the body in that position.21
Elite athletes who continue training during pregnancy should be carefully counseled about adjusting their training regimens. Because of increased cardiac output and blood volume, the heart rate will be lower than usual, demanding an adjustment in interpretation. Blood cell counts do not increase as much as plasma volume does—often leading to relative anemia. For elite athletes, this means iron supplementation is crucial.22 Thermal regulation may be more difficult, as training regimens may demand prolonged exercise. Physicians should recommend adequate hydration for these athletes.18
Although continued exercise is generally safe for a pregnant athlete and her fetus, caution is required when there is increased risk for premature delivery, or other special conditions exist. Multiple gestation, placenta previa, history of early labor or premature births, and incompetent cervix all contraindicate aerobic exercise during pregnancy.18 With these exceptions in mind, physicians can safely counsel pregnant women to do moderate exercise 30 minutes every day.17,18 Other recommendations are listed at the American College of Obstetricians and Gynecologists website.23
4. Anterior cruciate ligament injuries
ACL injuries affect a staggering number of athletes. In the United States, approximately 100,000 people sustain these injuries annually.24 As they occur up to 8 times more often in women than in men, ACL injuries are a top concern for physicians treating female athletes.
This disproportionate injury rate is influenced by differences between male and female anatomy. The width and shape of the femoral intercondylar notch have been studied as potential variables influencing the risk for ACL injury. Analysis of notch-view radiographs revealed a significant inverse relationship between notch width and ACL injury.25 A-shaped notches, notches with a significantly larger base and a narrowed roof, were more prevalent in women but did not correlate with increased risk for ACL injury. Studies have shown that female athletes with a noncontact ACL injury have a higher lateral tibial plateau posterior slope; this slope is associated with increased peak anteromedial ACL strain, which may contribute to injury.26 An analysis of magnetic resonance imaging scans in patients with and without ACL injury revealed that, for female patients, decreased femoral intercondylar notch width at the anterior outlet combined with increased lateral compartment posterior slope correlated best with risk for ACL injury.27
Although static anatomical factors contribute to ACL injuries in female athletes, dynamic neuromuscular influences are potential opportunities for intervention. Female athletes with high relative quadriceps strength and weak hamstring strength may be at increased risk for ACL injury.28 This “quadriceps dominance” becomes important in sports involving high-risk activities, such as running, cutting, pivoting, and jumping. In addition, compared with male athletes, female athletes demonstrate increased lateral trunk motion and knee valgus torque while landing during noncontact ACL tears, making core stability a factor in ACL injury.29
The collaborative efforts of physicians, physical therapists, athletic trainers, and coaches have yielded multifactorial neuromuscular training programs for the prevention of noncontact ACL injuries. Ideal ACL prevention protocols involve sessions that last for at least 10 minutes and take place 3 times a week. At these sessions, exercises are focused on strengthening, balance, and proprioceptive training.30 The programs last about 8 weeks, but sustained benefits require maintenance after the program has been completed and during the off-season. Program adherence must be encouraged and can be facilitated by varying workouts and raising risk awareness. The most effective programs have reduced the relative risk of noncontact ACL injuries by 75% to 100%.31 These promising results have led to increased focus on program implementation in an effort to prevent ACL injury.
5. Continued sex discrimination and social injustice
In 1972, Title IX was passed as part of the Education Amendments Act. Title IX states, “No person in the United States shall, on the basis of sex, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any educational program or activity receiving Federal financial assistance.” Passage of this law, which has implications outside of athletic participation, marked an important turning point in women’s ability to participate equally in college sports.32,33 The Civil Rights Restoration Act, passed in 1988, strengthened Title IX and made it applicable to all institutions receiving federal funding.34 Before the 1970s, women typically were restricted to club sports, and funding and participation opportunities were weighted heavily toward men. Over the past 40 years, women’s participation in high school, college, and professional sports has taken a huge leap forward.32 For example, the number of women participating in high school sports increased from 294,000 (7.4% of all athletes) in 1972 to 3.4 million (>41% of all athletes) in 2014.
Despite advances in women’s civil rights, examples of inequality in US schools remain, particularly in the distribution of funding, which still strongly favors men’s football.32 Men’s sports receive 90% of media coverage.33 In 2002, women represented 55% of college students but only 42% of varsity athletes.34 The schools that have complied the least with Title IX are schools in the Midwest and the South and those with football teams.34 Women are underrepresented as coaches, and funding continues to be disproportionately spent on men’s sports.
For women, the benefits of participating in sports are far-reaching and significant. These benefits include improvements in academic success, mental health, and responsible behavior.33 Women’s gaining acceptance and respect throughout the athletic world seems to have carried over elsewhere. Although many institutions remain noncompliant with Title IX, efforts continue to have a strongly positive effect on gender equality in the United States.
1. Nattiv A, Loucks AB, Manore MM, Sanborn CF, Sundgot-Borgen J, Warren MP; American College of Sports Medicine. American College of Sports Medicine position stand. The female athlete triad. Med Sci Sports Exerc. 2007;39(10):1867-1882.
2. De Souza MJ, Nattiv A, Joy E, et al; Expert Panel. 2014 Female Athlete Triad Coalition consensus statement on treatment and return to play of the female athlete triad: 1st international conference held in San Francisco, California, May 2012 and 2nd international conference held in Indianapolis, Indiana, May 2013. Br J Sports Med. 2014;48(4):289.
3. Warren MP, Shantha S. The female athlete. Baillieres Best Pract Res Clin Endocrinol Metab. 2000;14(1):37-53.
4. Greenleaf C, Petrie TA, Carter J, Reel JJ. Female collegiate athletes: prevalence of eating disorders and disordered eating behaviors. J Am Coll Health. 2009;57(5):489-495.
5. Reinking MF, Alexander LE. Prevalence of disordered-eating behaviors in undergraduate female collegiate athletes and nonathletes. J Athl Train. 2005;40(1):47-51.
6. Rencken ML, Chesnut CH 3rd, Drinkwater BL. Bone density at multiple skeletal sites in amenorrheic athletes. JAMA. 1996;276(3):238-240.
7. Lappe J, Cullen D, Haynatzki G, Recker R, Ahlf R, Thompson K. Calcium and vitamin D supplementation decreases incidence of stress fractures in female Navy recruits. J Bone Miner Res. 2008;23(5):741-749.
8. De Souza MJ. 2014 Female athlete triad consensus statement on guidelines for treatment and return to play. National Collegiate Athletic Association (NCAA) website. http://www.ncaa.org/health-and-safety/nutrition-and-performance/2014-female-athlete-triad-consensus-statement-guidelines. Accessed November 24, 2015.
9. Preiss-Farzanegan SJ, Chapman B, Wong TM, Wu J, Bazarian JJ. The relationship between gender and postconcussion symptoms after sport-related mild traumatic brain injury. PM R. 2009;1(3):245-253.
10. Marar M, McIlvain NM, Fields SK, Comstock RD. Epidemiology of concussions among United States high school athletes in 20 sports. Am J Sports Med. 2012;40(4):747-755.
11. Uhl RL, Rosenbaum AJ, Czajka C, Mulligan M, King C. Minor traumatic brain injury: a primer for the orthopaedic surgeon. J Am Acad Orthop Surg. 2013;21(10):624-631.
12. Covassin T, Elbin RJ, Harris W, Parker T, Kontos A. The role of age and sex in symptoms, neurocognitive performance, and postural stability in athletes after concussion. Am J Sports Med. 2012;40(6):1303-1312.
13. Covassin T, Moran R, Wilhelm K. Concussion symptoms and neurocognitive performance of high school and college athletes who incur multiple concussions. Am J Sports Med. 2013;41(12):2885-2889.
14. Sports concussion policies and laws: information for parents, coaches, and school & sports professionals. Centers for Disease Control and Prevention website. http://www.cdc.gov/headsup/policy/index.html. Updated February 16, 2015. Accessed November 24, 2015.
15. Covassin T, Elbin RJ, Sarmiento K. Educating coaches about concussion in sports: evaluation of the CDC’s “Heads Up: concussion in youth sports” initiative. J Sch Health. 2012;82(5):233-238.
16. Lumbers ER. Exercise in pregnancy: physiological basis of exercise prescription for the pregnant woman. J Sci Med Sport. 2002;5(1):20-31.
17. ACOG Committee Obstetric Practice. ACOG Committee opinion. Number 267, January 2002: exercise during pregnancy and the postpartum period. Obstet Gynecol. 2002;99(1):171-173.
18. Artal R, O’Toole M. Guidelines of the American College of Obstetricians and Gynecologists for exercise during pregnancy and the postpartum period. Br J Sports Med. 2003;37(1):6-12.
19. Kramer MS. Regular aerobic exercise during pregnancy. Cochrane Database Syst Rev. 2000;(2):CD000180. Update in: Cochrane Database Syst Rev. 2002;(2):CD000180.
20. Stafne SN, Salvesen KA, Romundstad PR, Stuge B, Morkved S. Does regular exercise during pregnancy influence lumbopelvic pain? A randomized controlled trial. Acta Obstet Gynecol Scand. 2012;91(5):552-559.
21. Nascimento SL, Surita FG, Cecatti JG. Physical exercise during pregnancy: a systematic review. Curr Opin Obstet Gynecol. 2012;24(6):387-394.
22. Hale RW, Milne L. The elite athlete and exercise in pregnancy. Semin Perinatol. 1996;20(4):277-284.
23. Exercise during pregnancy. American College of Obstetricians and Gynecologists website. http://www.acog.org/Patients/FAQs/Exercise-During-Pregnancy. Published August 2011. Accessed November 24, 2015.
24. Giugliano DN, Solomon JL. ACL tears in female athletes. Phys Med Rehabil Clin North Am. 2007;18(3):417-438, viii.
25. Ireland ML, Ballantyne BT, Little K, McClay IS. A radiographic analysis of the relationship between the size and shape of the intercondylar notch and anterior cruciate ligament injury. Knee Surg Sports Traumatol Arthrosc. 2001;9(4):200-205.
26. Lipps DB, Oh YK, Ashton-Miller JA, Wojtys EM. Morphologic characteristics help explain the gender difference in peak anterior cruciate ligament strain during a simulated pivot landing. Am J Sports Med. 2012;40(1):32-40.
27. Sturnick DR, Vacek PM, DeSarno MJ, et al. Combined anatomic factors predicting risk of anterior cruciate ligament injury for males and females. Am J Sports Med. 2015;43(4):839-847.
28. Myer GD, Ford KR, Barber Foss KD, Liu C, Nick TG, Hewett TE. The relationship of hamstrings and quadriceps strength to anterior cruciate ligament injury in female athletes. Clin J Sport Med. 2009;19(1):3-8.
29. Hewett TE, Torg JS, Boden BP. Video analysis of trunk and knee motion during non-contact anterior cruciate ligament injury in female athletes: lateral trunk and knee abduction motion are combined components of the injury mechanism. Br J Sports Med. 2009;43(6):417-422.
30. Sutton KM, Bullock JM. Anterior cruciate ligament rupture: differences between males and females. J Am Acad Orthop Surg. 2013;21(1):41-50.
31. Noyes FR, Barber-Westin SD. Neuromuscular retraining intervention programs: do they reduce noncontact anterior cruciate ligament injury rates in adolescent female athletes? Arthroscopy. 2014;30(2):245-255.
32. Ladd AL. The sports bra, the ACL, and Title IX—the game in play. Clin Orthop Relat Res. 2014;472(6):1681-1684.
33. Lopiano DA. Modern history of women in sports. Twenty-five years of Title IX. Clin Sports Med. 2000;19(2):163-173, vii.
34. Anderson DJ, Cheslock JJ, Ehrenberg RG. Gender equity in intercollegiate athletics: determinants of Title IX compliance. J High Educ. 2006;77(2):225-250.
1. Nattiv A, Loucks AB, Manore MM, Sanborn CF, Sundgot-Borgen J, Warren MP; American College of Sports Medicine. American College of Sports Medicine position stand. The female athlete triad. Med Sci Sports Exerc. 2007;39(10):1867-1882.
2. De Souza MJ, Nattiv A, Joy E, et al; Expert Panel. 2014 Female Athlete Triad Coalition consensus statement on treatment and return to play of the female athlete triad: 1st international conference held in San Francisco, California, May 2012 and 2nd international conference held in Indianapolis, Indiana, May 2013. Br J Sports Med. 2014;48(4):289.
3. Warren MP, Shantha S. The female athlete. Baillieres Best Pract Res Clin Endocrinol Metab. 2000;14(1):37-53.
4. Greenleaf C, Petrie TA, Carter J, Reel JJ. Female collegiate athletes: prevalence of eating disorders and disordered eating behaviors. J Am Coll Health. 2009;57(5):489-495.
5. Reinking MF, Alexander LE. Prevalence of disordered-eating behaviors in undergraduate female collegiate athletes and nonathletes. J Athl Train. 2005;40(1):47-51.
6. Rencken ML, Chesnut CH 3rd, Drinkwater BL. Bone density at multiple skeletal sites in amenorrheic athletes. JAMA. 1996;276(3):238-240.
7. Lappe J, Cullen D, Haynatzki G, Recker R, Ahlf R, Thompson K. Calcium and vitamin D supplementation decreases incidence of stress fractures in female Navy recruits. J Bone Miner Res. 2008;23(5):741-749.
8. De Souza MJ. 2014 Female athlete triad consensus statement on guidelines for treatment and return to play. National Collegiate Athletic Association (NCAA) website. http://www.ncaa.org/health-and-safety/nutrition-and-performance/2014-female-athlete-triad-consensus-statement-guidelines. Accessed November 24, 2015.
9. Preiss-Farzanegan SJ, Chapman B, Wong TM, Wu J, Bazarian JJ. The relationship between gender and postconcussion symptoms after sport-related mild traumatic brain injury. PM R. 2009;1(3):245-253.
10. Marar M, McIlvain NM, Fields SK, Comstock RD. Epidemiology of concussions among United States high school athletes in 20 sports. Am J Sports Med. 2012;40(4):747-755.
11. Uhl RL, Rosenbaum AJ, Czajka C, Mulligan M, King C. Minor traumatic brain injury: a primer for the orthopaedic surgeon. J Am Acad Orthop Surg. 2013;21(10):624-631.
12. Covassin T, Elbin RJ, Harris W, Parker T, Kontos A. The role of age and sex in symptoms, neurocognitive performance, and postural stability in athletes after concussion. Am J Sports Med. 2012;40(6):1303-1312.
13. Covassin T, Moran R, Wilhelm K. Concussion symptoms and neurocognitive performance of high school and college athletes who incur multiple concussions. Am J Sports Med. 2013;41(12):2885-2889.
14. Sports concussion policies and laws: information for parents, coaches, and school & sports professionals. Centers for Disease Control and Prevention website. http://www.cdc.gov/headsup/policy/index.html. Updated February 16, 2015. Accessed November 24, 2015.
15. Covassin T, Elbin RJ, Sarmiento K. Educating coaches about concussion in sports: evaluation of the CDC’s “Heads Up: concussion in youth sports” initiative. J Sch Health. 2012;82(5):233-238.
16. Lumbers ER. Exercise in pregnancy: physiological basis of exercise prescription for the pregnant woman. J Sci Med Sport. 2002;5(1):20-31.
17. ACOG Committee Obstetric Practice. ACOG Committee opinion. Number 267, January 2002: exercise during pregnancy and the postpartum period. Obstet Gynecol. 2002;99(1):171-173.
18. Artal R, O’Toole M. Guidelines of the American College of Obstetricians and Gynecologists for exercise during pregnancy and the postpartum period. Br J Sports Med. 2003;37(1):6-12.
19. Kramer MS. Regular aerobic exercise during pregnancy. Cochrane Database Syst Rev. 2000;(2):CD000180. Update in: Cochrane Database Syst Rev. 2002;(2):CD000180.
20. Stafne SN, Salvesen KA, Romundstad PR, Stuge B, Morkved S. Does regular exercise during pregnancy influence lumbopelvic pain? A randomized controlled trial. Acta Obstet Gynecol Scand. 2012;91(5):552-559.
21. Nascimento SL, Surita FG, Cecatti JG. Physical exercise during pregnancy: a systematic review. Curr Opin Obstet Gynecol. 2012;24(6):387-394.
22. Hale RW, Milne L. The elite athlete and exercise in pregnancy. Semin Perinatol. 1996;20(4):277-284.
23. Exercise during pregnancy. American College of Obstetricians and Gynecologists website. http://www.acog.org/Patients/FAQs/Exercise-During-Pregnancy. Published August 2011. Accessed November 24, 2015.
24. Giugliano DN, Solomon JL. ACL tears in female athletes. Phys Med Rehabil Clin North Am. 2007;18(3):417-438, viii.
25. Ireland ML, Ballantyne BT, Little K, McClay IS. A radiographic analysis of the relationship between the size and shape of the intercondylar notch and anterior cruciate ligament injury. Knee Surg Sports Traumatol Arthrosc. 2001;9(4):200-205.
26. Lipps DB, Oh YK, Ashton-Miller JA, Wojtys EM. Morphologic characteristics help explain the gender difference in peak anterior cruciate ligament strain during a simulated pivot landing. Am J Sports Med. 2012;40(1):32-40.
27. Sturnick DR, Vacek PM, DeSarno MJ, et al. Combined anatomic factors predicting risk of anterior cruciate ligament injury for males and females. Am J Sports Med. 2015;43(4):839-847.
28. Myer GD, Ford KR, Barber Foss KD, Liu C, Nick TG, Hewett TE. The relationship of hamstrings and quadriceps strength to anterior cruciate ligament injury in female athletes. Clin J Sport Med. 2009;19(1):3-8.
29. Hewett TE, Torg JS, Boden BP. Video analysis of trunk and knee motion during non-contact anterior cruciate ligament injury in female athletes: lateral trunk and knee abduction motion are combined components of the injury mechanism. Br J Sports Med. 2009;43(6):417-422.
30. Sutton KM, Bullock JM. Anterior cruciate ligament rupture: differences between males and females. J Am Acad Orthop Surg. 2013;21(1):41-50.
31. Noyes FR, Barber-Westin SD. Neuromuscular retraining intervention programs: do they reduce noncontact anterior cruciate ligament injury rates in adolescent female athletes? Arthroscopy. 2014;30(2):245-255.
32. Ladd AL. The sports bra, the ACL, and Title IX—the game in play. Clin Orthop Relat Res. 2014;472(6):1681-1684.
33. Lopiano DA. Modern history of women in sports. Twenty-five years of Title IX. Clin Sports Med. 2000;19(2):163-173, vii.
34. Anderson DJ, Cheslock JJ, Ehrenberg RG. Gender equity in intercollegiate athletics: determinants of Title IX compliance. J High Educ. 2006;77(2):225-250.
Patient-Directed Valgus Stress Radiograph of the Knee: A New and Novel Technique
Medial-compartment partial knee arthroplasty (unicompartmental replacement) is an accepted surgical intervention for anteromedial osteoarthritis of the knee.1 The radiographic investigations required in the workup of these patients should include weight-bearing standing anteroposterior (AP), lateral, and sunrise (Merchant) views, as well as a valgus stress AP radiograph to assess the functionality of the lateral compartment. The method of properly obtaining the valgus stress film has been well described by the Oxford Group.2 Its recommended radiographic technique requires that a surgeon or a radiologic technologist perform the valgus stress maneuver, manually, while another technologist shoots the film. The 2 consequences of this technique are that it requires 2 individuals to obtain the film, and it subjects the individual who is applying the stress to some level of radiation exposure, which is undesirable. Because of this and the time inconvenience, many surgeons omit the valgus stress radiograph, which can lead to the adverse outcome of missing a lateral compartment that is functionally incompetent, resulting in the potential for early lateral compartment progression of osteoarthritis and the need for revision surgery, usually to a total knee arthroplasty.
In an attempt to mitigate these barriers to obtaining the necessary valgus stress radiograph, Dr. Mauerhan’s team developed a technique that could be done with the assistance of the patient and would require only 1 technologist to perform. Additionally, this project was a quality improvement initiative, because it lowered radiation exposure to all personnel involved in obtaining the correct films.
Materials and Methods
We initiated the project using weight-bearing strategies to impart the valgus stress view of the knee. After trying several different wedges and blocks, and varying patient instructions, we realized a different approach to this problem would be required to find an acceptable solution. We redirected our efforts to effectively performing the stress view with the patient in a supine position on the radiograph table. Ultimately, we decided that a much stiffer wedge and a denser object to squeeze would facilitate obtaining a proper film. Considering all available options, a youth size 4 soccer ball (diameter, 11 in) was introduced along with a slightly larger positioning wedge. The soccer ball was wrapped with 4-in Coban wrap (3M) to create a nonslip surface. This change in patient positioning, along with a standardized 7º to 10º cephalic radiographic tube angulation, helped to correct issues with tibial plateau visualization. Once these changes were enacted, we obtained fairly consistent positive results, and we instituted this patient-directed valgus stress view of the knee, along with a manual valgus stress view for comparison.
The protocol for obtaining the patient-directed valgus stress view of the knee is as follows: The patient lays supine with a dense 45º spine-positioning wedge (Burlington Medical Supplies) placed under both knees and the patient’s heels on the examining table. The radiographic tube is angled cephalad 7º to 10º centered on the inferior pole of the patella, using a 40-in source to image-receptor distance, collimated to part; the image receptor is placed under the affected knee, below the positioning wedge. The affected knee is rotated to the “true” AP position (the patella will be centered between the femoral condyles on the AP exposure), and the ball is placed between the patient’s legs just above the ankle joint. The technologist demonstrates to the patient how to squeeze the ball while maintaining contact of heels with the table. The technologist can exit the room and obtain the exposure, which is taken while the patient is squeezing the ball, as shown in Figures 1A and 1B. Examples of the standing AP, manual stress, and patient-directed valgus radiographs are shown in Figures 2A-2C. The entire technique is demonstrated in the Video.
Results
During the 9 months of this quality improvement project, 78 examinations were performed. Five studies did not show complete correction of the varus deformity. Of these, 3 showed complete correction on a manual valgus stress radiograph, and 2 did not, contraindicating the use of partial knee replacement. Three patients displayed collapse of the lateral compartment, indicating a nonfunctional lateral compartment, and, therefore, were also a contraindication to partial knee arthroplasty. The remaining 70 patients had identical radiographic results with both the manual and patient-directed valgus stress tests. There was no instance of examination failure or need to repeat as a result of difficulty of the examination for the patient. Repeat films because of positioning errors were very rare, usually early in the learning curve, and no more prevalent than when using the manual stress method. The technique was reproducible and easy to teach and adopt.
Discussion
In total, 73 patients (93.5%) with the patient-directed stress film showed the desired result, either correction of the medial compartment narrowing in conjunction with an intact lateral compartment or narrowing of the lateral compartment. Of the 5 patients (6.5%) whose patient-directed stress films did not show correction of the varus deformity, 3 patients displayed correction with a manually applied stress radiograph and 2 did not. Based on this observation, our recommendation would be for those patients who do not show adequate correction on the patient-directed stress radiograph to have a manual examination to establish the presence or absence of the desired correction.
Performing a valgus stress radiograph is an integral part of the investigation to determine if the patient is an appropriate candidate for partial knee arthroplasty.3 The historical, manually performed valgus stress radiograph requires 2 individuals, 1 to apply the stress with the patient on the table and 1 to shoot the exposure. For the individual or individuals applying this stress, there is an increased radiation exposure that would be undesirable over a long career. The authors developed a new technique using a commercially available spinal positioning wedge and 11-in youth soccer ball wrapped with Coban wrap, as described, which is economical and easy to obtain and use in the clinical setting. We believe this cost-effective method will offer surgeons who perform partial knee arthroplasty a novel method to obtain the important information gleaned from the valgus stress radiograph and to improve surgical outcomes through the preoperative assessment of the lateral compartment. Additionally, as a quality and safety improvement initiative, we believe this technique will reduce radiographic exposure for those performing these studies, and, because the examination can be carried out by a single technologist, it will significantly improve efficiency in the radiology suite.
Conclusion
We have developed a new method of obtaining the important valgus stress radiograph as part of the workup of patients with medial-compartment osteoarthritis of the knee. The technique can be performed with easily obtainable, commercially available products and is reliable 93.5% of the time. It also adds to the efficiency of the radiology suite and reduces radiographic exposure for technologists.
1. White SH, Ludkowski PF, Goodfellow JW. Anteromedial osteoarthritis of the knee. J Bone Joint Surg Br. 1991;73(4):582-586.
2. Goodfellow JW, O’Conner JJ, Dodd CA, Murray DW. Unicompartmental Arthroplasty with the Oxford Knee. Woodeaton, Oxford, England: Goodfellow Publishers Limited; 2006:38-39.
3. Gibson PH, Goodfellow JW. Stress radiography in degenerative arthritis of the knee. J Bone Joint Surg Br. 1986;68(4):608-609.
Medial-compartment partial knee arthroplasty (unicompartmental replacement) is an accepted surgical intervention for anteromedial osteoarthritis of the knee.1 The radiographic investigations required in the workup of these patients should include weight-bearing standing anteroposterior (AP), lateral, and sunrise (Merchant) views, as well as a valgus stress AP radiograph to assess the functionality of the lateral compartment. The method of properly obtaining the valgus stress film has been well described by the Oxford Group.2 Its recommended radiographic technique requires that a surgeon or a radiologic technologist perform the valgus stress maneuver, manually, while another technologist shoots the film. The 2 consequences of this technique are that it requires 2 individuals to obtain the film, and it subjects the individual who is applying the stress to some level of radiation exposure, which is undesirable. Because of this and the time inconvenience, many surgeons omit the valgus stress radiograph, which can lead to the adverse outcome of missing a lateral compartment that is functionally incompetent, resulting in the potential for early lateral compartment progression of osteoarthritis and the need for revision surgery, usually to a total knee arthroplasty.
In an attempt to mitigate these barriers to obtaining the necessary valgus stress radiograph, Dr. Mauerhan’s team developed a technique that could be done with the assistance of the patient and would require only 1 technologist to perform. Additionally, this project was a quality improvement initiative, because it lowered radiation exposure to all personnel involved in obtaining the correct films.
Materials and Methods
We initiated the project using weight-bearing strategies to impart the valgus stress view of the knee. After trying several different wedges and blocks, and varying patient instructions, we realized a different approach to this problem would be required to find an acceptable solution. We redirected our efforts to effectively performing the stress view with the patient in a supine position on the radiograph table. Ultimately, we decided that a much stiffer wedge and a denser object to squeeze would facilitate obtaining a proper film. Considering all available options, a youth size 4 soccer ball (diameter, 11 in) was introduced along with a slightly larger positioning wedge. The soccer ball was wrapped with 4-in Coban wrap (3M) to create a nonslip surface. This change in patient positioning, along with a standardized 7º to 10º cephalic radiographic tube angulation, helped to correct issues with tibial plateau visualization. Once these changes were enacted, we obtained fairly consistent positive results, and we instituted this patient-directed valgus stress view of the knee, along with a manual valgus stress view for comparison.
The protocol for obtaining the patient-directed valgus stress view of the knee is as follows: The patient lays supine with a dense 45º spine-positioning wedge (Burlington Medical Supplies) placed under both knees and the patient’s heels on the examining table. The radiographic tube is angled cephalad 7º to 10º centered on the inferior pole of the patella, using a 40-in source to image-receptor distance, collimated to part; the image receptor is placed under the affected knee, below the positioning wedge. The affected knee is rotated to the “true” AP position (the patella will be centered between the femoral condyles on the AP exposure), and the ball is placed between the patient’s legs just above the ankle joint. The technologist demonstrates to the patient how to squeeze the ball while maintaining contact of heels with the table. The technologist can exit the room and obtain the exposure, which is taken while the patient is squeezing the ball, as shown in Figures 1A and 1B. Examples of the standing AP, manual stress, and patient-directed valgus radiographs are shown in Figures 2A-2C. The entire technique is demonstrated in the Video.
Results
During the 9 months of this quality improvement project, 78 examinations were performed. Five studies did not show complete correction of the varus deformity. Of these, 3 showed complete correction on a manual valgus stress radiograph, and 2 did not, contraindicating the use of partial knee replacement. Three patients displayed collapse of the lateral compartment, indicating a nonfunctional lateral compartment, and, therefore, were also a contraindication to partial knee arthroplasty. The remaining 70 patients had identical radiographic results with both the manual and patient-directed valgus stress tests. There was no instance of examination failure or need to repeat as a result of difficulty of the examination for the patient. Repeat films because of positioning errors were very rare, usually early in the learning curve, and no more prevalent than when using the manual stress method. The technique was reproducible and easy to teach and adopt.
Discussion
In total, 73 patients (93.5%) with the patient-directed stress film showed the desired result, either correction of the medial compartment narrowing in conjunction with an intact lateral compartment or narrowing of the lateral compartment. Of the 5 patients (6.5%) whose patient-directed stress films did not show correction of the varus deformity, 3 patients displayed correction with a manually applied stress radiograph and 2 did not. Based on this observation, our recommendation would be for those patients who do not show adequate correction on the patient-directed stress radiograph to have a manual examination to establish the presence or absence of the desired correction.
Performing a valgus stress radiograph is an integral part of the investigation to determine if the patient is an appropriate candidate for partial knee arthroplasty.3 The historical, manually performed valgus stress radiograph requires 2 individuals, 1 to apply the stress with the patient on the table and 1 to shoot the exposure. For the individual or individuals applying this stress, there is an increased radiation exposure that would be undesirable over a long career. The authors developed a new technique using a commercially available spinal positioning wedge and 11-in youth soccer ball wrapped with Coban wrap, as described, which is economical and easy to obtain and use in the clinical setting. We believe this cost-effective method will offer surgeons who perform partial knee arthroplasty a novel method to obtain the important information gleaned from the valgus stress radiograph and to improve surgical outcomes through the preoperative assessment of the lateral compartment. Additionally, as a quality and safety improvement initiative, we believe this technique will reduce radiographic exposure for those performing these studies, and, because the examination can be carried out by a single technologist, it will significantly improve efficiency in the radiology suite.
Conclusion
We have developed a new method of obtaining the important valgus stress radiograph as part of the workup of patients with medial-compartment osteoarthritis of the knee. The technique can be performed with easily obtainable, commercially available products and is reliable 93.5% of the time. It also adds to the efficiency of the radiology suite and reduces radiographic exposure for technologists.
Medial-compartment partial knee arthroplasty (unicompartmental replacement) is an accepted surgical intervention for anteromedial osteoarthritis of the knee.1 The radiographic investigations required in the workup of these patients should include weight-bearing standing anteroposterior (AP), lateral, and sunrise (Merchant) views, as well as a valgus stress AP radiograph to assess the functionality of the lateral compartment. The method of properly obtaining the valgus stress film has been well described by the Oxford Group.2 Its recommended radiographic technique requires that a surgeon or a radiologic technologist perform the valgus stress maneuver, manually, while another technologist shoots the film. The 2 consequences of this technique are that it requires 2 individuals to obtain the film, and it subjects the individual who is applying the stress to some level of radiation exposure, which is undesirable. Because of this and the time inconvenience, many surgeons omit the valgus stress radiograph, which can lead to the adverse outcome of missing a lateral compartment that is functionally incompetent, resulting in the potential for early lateral compartment progression of osteoarthritis and the need for revision surgery, usually to a total knee arthroplasty.
In an attempt to mitigate these barriers to obtaining the necessary valgus stress radiograph, Dr. Mauerhan’s team developed a technique that could be done with the assistance of the patient and would require only 1 technologist to perform. Additionally, this project was a quality improvement initiative, because it lowered radiation exposure to all personnel involved in obtaining the correct films.
Materials and Methods
We initiated the project using weight-bearing strategies to impart the valgus stress view of the knee. After trying several different wedges and blocks, and varying patient instructions, we realized a different approach to this problem would be required to find an acceptable solution. We redirected our efforts to effectively performing the stress view with the patient in a supine position on the radiograph table. Ultimately, we decided that a much stiffer wedge and a denser object to squeeze would facilitate obtaining a proper film. Considering all available options, a youth size 4 soccer ball (diameter, 11 in) was introduced along with a slightly larger positioning wedge. The soccer ball was wrapped with 4-in Coban wrap (3M) to create a nonslip surface. This change in patient positioning, along with a standardized 7º to 10º cephalic radiographic tube angulation, helped to correct issues with tibial plateau visualization. Once these changes were enacted, we obtained fairly consistent positive results, and we instituted this patient-directed valgus stress view of the knee, along with a manual valgus stress view for comparison.
The protocol for obtaining the patient-directed valgus stress view of the knee is as follows: The patient lays supine with a dense 45º spine-positioning wedge (Burlington Medical Supplies) placed under both knees and the patient’s heels on the examining table. The radiographic tube is angled cephalad 7º to 10º centered on the inferior pole of the patella, using a 40-in source to image-receptor distance, collimated to part; the image receptor is placed under the affected knee, below the positioning wedge. The affected knee is rotated to the “true” AP position (the patella will be centered between the femoral condyles on the AP exposure), and the ball is placed between the patient’s legs just above the ankle joint. The technologist demonstrates to the patient how to squeeze the ball while maintaining contact of heels with the table. The technologist can exit the room and obtain the exposure, which is taken while the patient is squeezing the ball, as shown in Figures 1A and 1B. Examples of the standing AP, manual stress, and patient-directed valgus radiographs are shown in Figures 2A-2C. The entire technique is demonstrated in the Video.
Results
During the 9 months of this quality improvement project, 78 examinations were performed. Five studies did not show complete correction of the varus deformity. Of these, 3 showed complete correction on a manual valgus stress radiograph, and 2 did not, contraindicating the use of partial knee replacement. Three patients displayed collapse of the lateral compartment, indicating a nonfunctional lateral compartment, and, therefore, were also a contraindication to partial knee arthroplasty. The remaining 70 patients had identical radiographic results with both the manual and patient-directed valgus stress tests. There was no instance of examination failure or need to repeat as a result of difficulty of the examination for the patient. Repeat films because of positioning errors were very rare, usually early in the learning curve, and no more prevalent than when using the manual stress method. The technique was reproducible and easy to teach and adopt.
Discussion
In total, 73 patients (93.5%) with the patient-directed stress film showed the desired result, either correction of the medial compartment narrowing in conjunction with an intact lateral compartment or narrowing of the lateral compartment. Of the 5 patients (6.5%) whose patient-directed stress films did not show correction of the varus deformity, 3 patients displayed correction with a manually applied stress radiograph and 2 did not. Based on this observation, our recommendation would be for those patients who do not show adequate correction on the patient-directed stress radiograph to have a manual examination to establish the presence or absence of the desired correction.
Performing a valgus stress radiograph is an integral part of the investigation to determine if the patient is an appropriate candidate for partial knee arthroplasty.3 The historical, manually performed valgus stress radiograph requires 2 individuals, 1 to apply the stress with the patient on the table and 1 to shoot the exposure. For the individual or individuals applying this stress, there is an increased radiation exposure that would be undesirable over a long career. The authors developed a new technique using a commercially available spinal positioning wedge and 11-in youth soccer ball wrapped with Coban wrap, as described, which is economical and easy to obtain and use in the clinical setting. We believe this cost-effective method will offer surgeons who perform partial knee arthroplasty a novel method to obtain the important information gleaned from the valgus stress radiograph and to improve surgical outcomes through the preoperative assessment of the lateral compartment. Additionally, as a quality and safety improvement initiative, we believe this technique will reduce radiographic exposure for those performing these studies, and, because the examination can be carried out by a single technologist, it will significantly improve efficiency in the radiology suite.
Conclusion
We have developed a new method of obtaining the important valgus stress radiograph as part of the workup of patients with medial-compartment osteoarthritis of the knee. The technique can be performed with easily obtainable, commercially available products and is reliable 93.5% of the time. It also adds to the efficiency of the radiology suite and reduces radiographic exposure for technologists.
1. White SH, Ludkowski PF, Goodfellow JW. Anteromedial osteoarthritis of the knee. J Bone Joint Surg Br. 1991;73(4):582-586.
2. Goodfellow JW, O’Conner JJ, Dodd CA, Murray DW. Unicompartmental Arthroplasty with the Oxford Knee. Woodeaton, Oxford, England: Goodfellow Publishers Limited; 2006:38-39.
3. Gibson PH, Goodfellow JW. Stress radiography in degenerative arthritis of the knee. J Bone Joint Surg Br. 1986;68(4):608-609.
1. White SH, Ludkowski PF, Goodfellow JW. Anteromedial osteoarthritis of the knee. J Bone Joint Surg Br. 1991;73(4):582-586.
2. Goodfellow JW, O’Conner JJ, Dodd CA, Murray DW. Unicompartmental Arthroplasty with the Oxford Knee. Woodeaton, Oxford, England: Goodfellow Publishers Limited; 2006:38-39.
3. Gibson PH, Goodfellow JW. Stress radiography in degenerative arthritis of the knee. J Bone Joint Surg Br. 1986;68(4):608-609.
Concomitant Ulnar Styloid Fracture and Distal Radius Fracture Portend Poorer Outcome
Distal radius fracture is a common injury treated by orthopedic surgeons. Fifty percent or more of distal radius fractures (DRFs) occur with concomitant ulnar styloid fractures (USFs)1-3 (Figure). The base of the ulnar styloid is the insertion site for portions of the triangular fibrocartilaginous complex (TFCC), which is a primary stabilizer of the distal radioulnar joint (DRUJ).4,5
Although the topic has received significant attention in the literature, there remains a lack of consensus on the prognostic and clinical significance of USF occurring with DRF. In a series reported by May and colleagues,6 all patients with DRUJ instability after DRF also had an USF. Some authors have reported USF as a poor prognostic indicator for DRF, as the occurrence of USF was taken as a proxy for DRUJ instability.7,8 Conversely, other authors have reported that USF nonunion has no effect on the outcome of volar plating of DRF.9-11 In a retrospective cohort study of 182 patients, Li and colleagues12 found no clinically significant difference in outcome between presence or absence of USF with DRF. They also reported that the quality of the DRF reduction was the main determinant of clinical outcome in patients with USF.
We examined a large cohort of patients treated for DRF to identify any possible effect of an associated USF on clinical outcome. All patients provided written informed consent for study inclusion.
Materials and Methods
We retrospectively evaluated 315 cases of DRFs treated (184 operatively, 131 nonoperatively) by members of the Trauma and Hand divisions at our institution over a 7-year period. All cases had sufficient follow-up. In each group, patients with concomitant USF were identified.
At presentation, all displaced fractures underwent closed reduction and immobilization with a sugar-tong splint. Baseline demographic data, injury information, and baseline functional scores on the Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire and the 36-Item Short Form Health Survey (SF-36) were recorded. Complete histories were taken and physical examinations performed. Standard radiographs of the injured and contralateral wrists were obtained at time of initial injury.13
Surgery was indicated in patients with an open fracture and in patients with an inherently unstable fracture pattern, using the instability criteria of Cooney and colleagues.14 According to these criteria, unstable fractures have lost alignment after closed reduction or have more than 20° of dorsal angulation, more than 10 mm of longitudinal shortening, or more than 2 mm of articular displacement.14 Patients were treated with either a volar locked plate or bridging external fixation with supplemental Kirschner-wire fixation (usually 2 or 3 wires). Patients in both groups (operative, nonoperative) participated in a formal outpatient therapy program that emphasized active and passive range of motion (ROM) of the finger, wrist motion (if clinically appropriate), and forearm motion. Mean clinical follow-up was 12 months (range, 8-18 months). At each clinic visit, we used a handheld dynamometer to measure ROM, grip strength, and other parameters and compared them with the same parameters on the uninjured side, along with functional outcome.
Differences in demographic characteristics were evaluated with 2 tests—the χ2 test for categorical variables (eg, USF incidence, sex, hand dominance, fracture pattern) and the Student t test for continuous variables. Mann-Whitney U tests were used to assess differences between groups in DASH and SF-36 scores at long-term follow-up, as well as differences in ROM and radiographic measurements. Statistical significance was set at P < .05.
Results
DRFs occurred in the dominant-side wrist more commonly (P < .05) in the nonoperative group than in the operative group, though there was no difference in hand dominance and presence or absence of USF. There was a significant correlation of intra-articular fractures in the operative group (70%) compared with the nonoperative group (34%), though no association was found between presence of USF and intra-articular fracture location.
The percentage of concomitant USF was higher (P< .0002) in patients treated operatively (64.1%) than in those treated nonoperatively (38.9%). Mean (SD) pain score was higher (P = .0001) for patients with USF, 1.80 (2.43), than for patients without USF, 0.80 (1.55). This relationship held in both the operative group, 1.95 (2.48) versus 1.04 (1.58) (P = .027), and the nonoperative group, 1.29 (2.09) versus 0.66 (1.53) (P = .048). Similarly, at long-term follow-up for the entire patient cohort, mean (SD) DASH score was negatively affected by presence of USF, 17.03 (18.94) versus 9.21 (14.06) (P = .001), as was mean (SD) SF-36 score, 77.16 (17.69) versus 82.68 (16.10) (P = .022). This relationship also held in the operative and nonoperative groups with respect to pain and DASH scores, though there were only trends in this direction with respect to SF-36 scores. At final follow-up, there was no significant correlation of pain, SF-36, or DASH scores with presence of an intra-articular fracture as compared with an extra-articular fracture.
Time to radiographic healing was not influenced by presence of USF compared with absence of USF (11 vs 10.06 weeks; P > .05). Similarly, healing was no different in intra-articular fractures compared with extra-articular fractures (11 vs 10 weeks; P > .05).
Wrist ROM at final follow-up was not affected by presence of USF; there was no significant difference in wrist flexion, extension, or forearm rotation. In addition, mean (SD) grip strength was unaffected (P = .132) by presence or absence of USF with DRF overall, 45.45% (31.92) of contralateral versus 52.88% (30.03). However, grip strength was negatively affected (P = .035) by presence of USF in the nonoperative group, 37.79% (20.58) versus 54.52% (31.89) (Table).
Discussion
In this study, we determined that presence of USF was a negative predictor for clinical outcomes after DRF. Given the higher incidence of USF in operatively treated DRFs, USF likely represents a higher-energy mechanism of injury. We think these inferior clinical results are attributable to other wrist pathologies that commonly occur with these injuries. These pathologies, identified in the past, include stylocarpal impaction, extensor carpi ulnaris tendinitis, and pain at USF site.6,10,15 In addition, intracarpal ligamentous injuries, including damage to scapholunate and lunotriquetral ligaments, have been shown to occur in roughly 80% of patients who sustain DRFs, with TFCC injuries occurring at a rate of 60%.16
Patient outcome is multifactorial and depends on initial injury characteristics, reduction quality, associated injuries, and patient demographics and lifestyle factors. Li and colleagues12 showed that the quality of the DRF reduction influenced outcomes in these injuries, as the ulnar styloid and its associated TFCC are in turn reduced more anatomically with a restored DRF reduction. This concept applies to injuries treated both operatively and nonoperatively. Similarly, Xarchas and colleagues17 identified malunion of the ulnar styloid as causing chronic wrist pain because of triquetral impingement, which was treated successfully with ulnar styloidectomy. The poor results at final follow-up in their study may reflect severity of the initial injury, as reported by Frykman.18
Additional factors may compromise clinical outcomes after such injuries. For example, the effect of USF fragment size on outcome has been suggested and debated. In a retrospective series, May and colleagues6 identified fractures involving the base of the ulnar styloid or fovea as potentially destabilizing the DRUJ and in turn leading to chronic instability. This mechanism should be considered a potential contributor to protracted clinical recovery. Other studies have shown that, irrespective of USF fragment size, presence of USF with DRF is not a reliable predictor of DRUJ instability.2,10,19 In the present study, we simply identified presence or absence of USF, irrespective of either stability or fragment size. In cases in which there was an USF without instability, we fixed the DRF in isolation, without surgically addressing the USF. Our data demonstrated that, even in the absence of DRUJ instability, presence of USF was a negative prognostic indicator for patient outcome.
This study had several limitations. First, its design was retrospective. A prospective study would have been ideal for eliminating certain inherent bias. Second, USF represents a higher association with DRUJ instability.6 As there are no validated tests for this clinical entity, identification is somewhat subjective. We did not separate patients by presence or absence of DRUJ instability and thus were not able to directly correlate the connection between USF, DRUJ instability, and poor outcomes in association with DRF. In addition, management of an unstable DRUJ after operative fixation of DRF is controversial, with techniques ranging from splinting in supination to pinning the DRUJ. This inconsistency likely contributed to some error between groups of patients in this study. Last, we did not stratify patients by USF fragment size, as previously discussed, which may have affected outcomes within patient groups.
Our data add to the evidence showing that USF in association with DRF portends poorer clinical outcomes. Concomitant USF should alert the treating physician to a higher-energy mechanism of injury and raise the index of suspicion for other associated injuries in the carpus.
1. Richards RS, Bennett JD, Roth JH, Milne K Jr. Arthroscopic diagnosis of intra-articular soft tissue injuries associated with distal radial fractures. J Hand Surg Am. 1997;22(5):772-776.
2. Sammer DM, Shah HM, Shauver MJ, Chung KC. The effect of ulnar styloid fractures on patient-rated outcomes after volar locking plating of distal radius fractures. J Hand Surg Am. 2009;34(9):1595-1602.
3. Villar RN, Marsh D, Rushton N, Greatorex RA. Three years after Colles’ fracture. A prospective review. J Bone Joint Surg Br. 1987;69(4):635-638.
4. Palmer AK, Werner FW. The triangular fibrocartilage complex of the wrist—anatomy and function. J Hand Surg Am. 1981;6(2):153-162.
5. Stuart PR, Berger RA, Linscheid RL, An KN. The dorsopalmar stability of the distal radioulnar joint. J Hand Surg Am. 2000;25(4):689-699.
6. May MM, Lawton JN, Blazar PE. Ulnar styloid fractures associated with distal radius fractures: incidence and implications for distal radioulnar joint instability. J Hand Surg Am. 2002;27(6):965-971.
7. Oskarsson GV, Aaser P, Hjall A. Do we underestimate the predictive value of the ulnar styloid affection in Colles fractures? Arch Orthop Trauma Surg. 1997;116(6-7):341-344.
8. Stoffelen D, De Smet L, Broos P. The importance of the distal radioulnar joint in distal radial fractures. J Hand Surg Br. 1998;23(4):507-511.
9. Buijze GA, Ring D. Clinical impact of united versus nonunited fractures of the proximal half of the ulnar styloid following volar plate fixation of the distal radius. J Hand Surg Am. 2010;35(2):223-227.
10. Kim JK, Yun YH, Kim DJ, Yun GU. Comparison of united and nonunited fractures of the ulnar styloid following volar-plate fixation of distal radius fractures. Injury. 2011;42(4):371-375.
11. Wijffels M, Ring D. The influence of non-union of the ulnar styloid on pain, wrist function and instability after distal radius fracture. J Hand Microsurg. 2011;3(1):11-14.
12. Li S, Chen Y, Lin Z, Fan Q, Cui W, Feng Z. Effect of associated ulnar styloid fracture on wrist function after distal radius [in Chinese]. Zhongguo Xiu Fu Chong Jian Wai Ke Za Zhi. 2012;26(6):666-670.
13. Egol KA, Walsh M, Romo-Cardoso S, Dorsky S, Paksima N. Distal radial fractures in the elderly: operative compared with nonoperative treatment. J Bone Joint Surg Am. 2010;92(9):1851-1857.
14. Cooney WP 3rd, Linscheid RL, Dobyns JH. External pin fixation for unstable Colles’ fractures. J Bone Joint Surg Am. 1979;61(6):840-845.
15. Cerezal L, del Piñal F, Abascal F, García-Valtuille R, Pereda T, Canga A. Imaging findings in ulnar-sided wrist impaction syndromes. Radiographics. 2002;22(1):105-121.
16. Ogawa T, Tanaka T, Yanai T, Kumagai H, Ochiai N. Analysis of soft tissue injuries associated with distal radius fractures. BMC Sports Sci Med Rehabil. 2013;5(1):19.
17. Xarchas KC, Yfandithis P, Kazakos K. Malunion of the ulnar styloid as a cause of ulnar wrist pain. Clin Anat. 2004;17(5):418-422.
18. Frykman G. Fracture of the distal radius including sequelae—shoulder–hand–finger syndrome, disturbance in the distal radio-ulnar joint and impairment of nerve function. A clinical and experimental study. Acta Orthop Scand. 1967:(suppl 108):3+.
19. Fujitani R, Omokawa S, Akahane M, Iida A, Ono H, Tanaka Y. Predictors of distal radioulnar joint instability in distal radius fractures. J Hand Surg Am. 2011;36(12):1919-1925.
Distal radius fracture is a common injury treated by orthopedic surgeons. Fifty percent or more of distal radius fractures (DRFs) occur with concomitant ulnar styloid fractures (USFs)1-3 (Figure). The base of the ulnar styloid is the insertion site for portions of the triangular fibrocartilaginous complex (TFCC), which is a primary stabilizer of the distal radioulnar joint (DRUJ).4,5
Although the topic has received significant attention in the literature, there remains a lack of consensus on the prognostic and clinical significance of USF occurring with DRF. In a series reported by May and colleagues,6 all patients with DRUJ instability after DRF also had an USF. Some authors have reported USF as a poor prognostic indicator for DRF, as the occurrence of USF was taken as a proxy for DRUJ instability.7,8 Conversely, other authors have reported that USF nonunion has no effect on the outcome of volar plating of DRF.9-11 In a retrospective cohort study of 182 patients, Li and colleagues12 found no clinically significant difference in outcome between presence or absence of USF with DRF. They also reported that the quality of the DRF reduction was the main determinant of clinical outcome in patients with USF.
We examined a large cohort of patients treated for DRF to identify any possible effect of an associated USF on clinical outcome. All patients provided written informed consent for study inclusion.
Materials and Methods
We retrospectively evaluated 315 cases of DRFs treated (184 operatively, 131 nonoperatively) by members of the Trauma and Hand divisions at our institution over a 7-year period. All cases had sufficient follow-up. In each group, patients with concomitant USF were identified.
At presentation, all displaced fractures underwent closed reduction and immobilization with a sugar-tong splint. Baseline demographic data, injury information, and baseline functional scores on the Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire and the 36-Item Short Form Health Survey (SF-36) were recorded. Complete histories were taken and physical examinations performed. Standard radiographs of the injured and contralateral wrists were obtained at time of initial injury.13
Surgery was indicated in patients with an open fracture and in patients with an inherently unstable fracture pattern, using the instability criteria of Cooney and colleagues.14 According to these criteria, unstable fractures have lost alignment after closed reduction or have more than 20° of dorsal angulation, more than 10 mm of longitudinal shortening, or more than 2 mm of articular displacement.14 Patients were treated with either a volar locked plate or bridging external fixation with supplemental Kirschner-wire fixation (usually 2 or 3 wires). Patients in both groups (operative, nonoperative) participated in a formal outpatient therapy program that emphasized active and passive range of motion (ROM) of the finger, wrist motion (if clinically appropriate), and forearm motion. Mean clinical follow-up was 12 months (range, 8-18 months). At each clinic visit, we used a handheld dynamometer to measure ROM, grip strength, and other parameters and compared them with the same parameters on the uninjured side, along with functional outcome.
Differences in demographic characteristics were evaluated with 2 tests—the χ2 test for categorical variables (eg, USF incidence, sex, hand dominance, fracture pattern) and the Student t test for continuous variables. Mann-Whitney U tests were used to assess differences between groups in DASH and SF-36 scores at long-term follow-up, as well as differences in ROM and radiographic measurements. Statistical significance was set at P < .05.
Results
DRFs occurred in the dominant-side wrist more commonly (P < .05) in the nonoperative group than in the operative group, though there was no difference in hand dominance and presence or absence of USF. There was a significant correlation of intra-articular fractures in the operative group (70%) compared with the nonoperative group (34%), though no association was found between presence of USF and intra-articular fracture location.
The percentage of concomitant USF was higher (P< .0002) in patients treated operatively (64.1%) than in those treated nonoperatively (38.9%). Mean (SD) pain score was higher (P = .0001) for patients with USF, 1.80 (2.43), than for patients without USF, 0.80 (1.55). This relationship held in both the operative group, 1.95 (2.48) versus 1.04 (1.58) (P = .027), and the nonoperative group, 1.29 (2.09) versus 0.66 (1.53) (P = .048). Similarly, at long-term follow-up for the entire patient cohort, mean (SD) DASH score was negatively affected by presence of USF, 17.03 (18.94) versus 9.21 (14.06) (P = .001), as was mean (SD) SF-36 score, 77.16 (17.69) versus 82.68 (16.10) (P = .022). This relationship also held in the operative and nonoperative groups with respect to pain and DASH scores, though there were only trends in this direction with respect to SF-36 scores. At final follow-up, there was no significant correlation of pain, SF-36, or DASH scores with presence of an intra-articular fracture as compared with an extra-articular fracture.
Time to radiographic healing was not influenced by presence of USF compared with absence of USF (11 vs 10.06 weeks; P > .05). Similarly, healing was no different in intra-articular fractures compared with extra-articular fractures (11 vs 10 weeks; P > .05).
Wrist ROM at final follow-up was not affected by presence of USF; there was no significant difference in wrist flexion, extension, or forearm rotation. In addition, mean (SD) grip strength was unaffected (P = .132) by presence or absence of USF with DRF overall, 45.45% (31.92) of contralateral versus 52.88% (30.03). However, grip strength was negatively affected (P = .035) by presence of USF in the nonoperative group, 37.79% (20.58) versus 54.52% (31.89) (Table).
Discussion
In this study, we determined that presence of USF was a negative predictor for clinical outcomes after DRF. Given the higher incidence of USF in operatively treated DRFs, USF likely represents a higher-energy mechanism of injury. We think these inferior clinical results are attributable to other wrist pathologies that commonly occur with these injuries. These pathologies, identified in the past, include stylocarpal impaction, extensor carpi ulnaris tendinitis, and pain at USF site.6,10,15 In addition, intracarpal ligamentous injuries, including damage to scapholunate and lunotriquetral ligaments, have been shown to occur in roughly 80% of patients who sustain DRFs, with TFCC injuries occurring at a rate of 60%.16
Patient outcome is multifactorial and depends on initial injury characteristics, reduction quality, associated injuries, and patient demographics and lifestyle factors. Li and colleagues12 showed that the quality of the DRF reduction influenced outcomes in these injuries, as the ulnar styloid and its associated TFCC are in turn reduced more anatomically with a restored DRF reduction. This concept applies to injuries treated both operatively and nonoperatively. Similarly, Xarchas and colleagues17 identified malunion of the ulnar styloid as causing chronic wrist pain because of triquetral impingement, which was treated successfully with ulnar styloidectomy. The poor results at final follow-up in their study may reflect severity of the initial injury, as reported by Frykman.18
Additional factors may compromise clinical outcomes after such injuries. For example, the effect of USF fragment size on outcome has been suggested and debated. In a retrospective series, May and colleagues6 identified fractures involving the base of the ulnar styloid or fovea as potentially destabilizing the DRUJ and in turn leading to chronic instability. This mechanism should be considered a potential contributor to protracted clinical recovery. Other studies have shown that, irrespective of USF fragment size, presence of USF with DRF is not a reliable predictor of DRUJ instability.2,10,19 In the present study, we simply identified presence or absence of USF, irrespective of either stability or fragment size. In cases in which there was an USF without instability, we fixed the DRF in isolation, without surgically addressing the USF. Our data demonstrated that, even in the absence of DRUJ instability, presence of USF was a negative prognostic indicator for patient outcome.
This study had several limitations. First, its design was retrospective. A prospective study would have been ideal for eliminating certain inherent bias. Second, USF represents a higher association with DRUJ instability.6 As there are no validated tests for this clinical entity, identification is somewhat subjective. We did not separate patients by presence or absence of DRUJ instability and thus were not able to directly correlate the connection between USF, DRUJ instability, and poor outcomes in association with DRF. In addition, management of an unstable DRUJ after operative fixation of DRF is controversial, with techniques ranging from splinting in supination to pinning the DRUJ. This inconsistency likely contributed to some error between groups of patients in this study. Last, we did not stratify patients by USF fragment size, as previously discussed, which may have affected outcomes within patient groups.
Our data add to the evidence showing that USF in association with DRF portends poorer clinical outcomes. Concomitant USF should alert the treating physician to a higher-energy mechanism of injury and raise the index of suspicion for other associated injuries in the carpus.
Distal radius fracture is a common injury treated by orthopedic surgeons. Fifty percent or more of distal radius fractures (DRFs) occur with concomitant ulnar styloid fractures (USFs)1-3 (Figure). The base of the ulnar styloid is the insertion site for portions of the triangular fibrocartilaginous complex (TFCC), which is a primary stabilizer of the distal radioulnar joint (DRUJ).4,5
Although the topic has received significant attention in the literature, there remains a lack of consensus on the prognostic and clinical significance of USF occurring with DRF. In a series reported by May and colleagues,6 all patients with DRUJ instability after DRF also had an USF. Some authors have reported USF as a poor prognostic indicator for DRF, as the occurrence of USF was taken as a proxy for DRUJ instability.7,8 Conversely, other authors have reported that USF nonunion has no effect on the outcome of volar plating of DRF.9-11 In a retrospective cohort study of 182 patients, Li and colleagues12 found no clinically significant difference in outcome between presence or absence of USF with DRF. They also reported that the quality of the DRF reduction was the main determinant of clinical outcome in patients with USF.
We examined a large cohort of patients treated for DRF to identify any possible effect of an associated USF on clinical outcome. All patients provided written informed consent for study inclusion.
Materials and Methods
We retrospectively evaluated 315 cases of DRFs treated (184 operatively, 131 nonoperatively) by members of the Trauma and Hand divisions at our institution over a 7-year period. All cases had sufficient follow-up. In each group, patients with concomitant USF were identified.
At presentation, all displaced fractures underwent closed reduction and immobilization with a sugar-tong splint. Baseline demographic data, injury information, and baseline functional scores on the Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire and the 36-Item Short Form Health Survey (SF-36) were recorded. Complete histories were taken and physical examinations performed. Standard radiographs of the injured and contralateral wrists were obtained at time of initial injury.13
Surgery was indicated in patients with an open fracture and in patients with an inherently unstable fracture pattern, using the instability criteria of Cooney and colleagues.14 According to these criteria, unstable fractures have lost alignment after closed reduction or have more than 20° of dorsal angulation, more than 10 mm of longitudinal shortening, or more than 2 mm of articular displacement.14 Patients were treated with either a volar locked plate or bridging external fixation with supplemental Kirschner-wire fixation (usually 2 or 3 wires). Patients in both groups (operative, nonoperative) participated in a formal outpatient therapy program that emphasized active and passive range of motion (ROM) of the finger, wrist motion (if clinically appropriate), and forearm motion. Mean clinical follow-up was 12 months (range, 8-18 months). At each clinic visit, we used a handheld dynamometer to measure ROM, grip strength, and other parameters and compared them with the same parameters on the uninjured side, along with functional outcome.
Differences in demographic characteristics were evaluated with 2 tests—the χ2 test for categorical variables (eg, USF incidence, sex, hand dominance, fracture pattern) and the Student t test for continuous variables. Mann-Whitney U tests were used to assess differences between groups in DASH and SF-36 scores at long-term follow-up, as well as differences in ROM and radiographic measurements. Statistical significance was set at P < .05.
Results
DRFs occurred in the dominant-side wrist more commonly (P < .05) in the nonoperative group than in the operative group, though there was no difference in hand dominance and presence or absence of USF. There was a significant correlation of intra-articular fractures in the operative group (70%) compared with the nonoperative group (34%), though no association was found between presence of USF and intra-articular fracture location.
The percentage of concomitant USF was higher (P< .0002) in patients treated operatively (64.1%) than in those treated nonoperatively (38.9%). Mean (SD) pain score was higher (P = .0001) for patients with USF, 1.80 (2.43), than for patients without USF, 0.80 (1.55). This relationship held in both the operative group, 1.95 (2.48) versus 1.04 (1.58) (P = .027), and the nonoperative group, 1.29 (2.09) versus 0.66 (1.53) (P = .048). Similarly, at long-term follow-up for the entire patient cohort, mean (SD) DASH score was negatively affected by presence of USF, 17.03 (18.94) versus 9.21 (14.06) (P = .001), as was mean (SD) SF-36 score, 77.16 (17.69) versus 82.68 (16.10) (P = .022). This relationship also held in the operative and nonoperative groups with respect to pain and DASH scores, though there were only trends in this direction with respect to SF-36 scores. At final follow-up, there was no significant correlation of pain, SF-36, or DASH scores with presence of an intra-articular fracture as compared with an extra-articular fracture.
Time to radiographic healing was not influenced by presence of USF compared with absence of USF (11 vs 10.06 weeks; P > .05). Similarly, healing was no different in intra-articular fractures compared with extra-articular fractures (11 vs 10 weeks; P > .05).
Wrist ROM at final follow-up was not affected by presence of USF; there was no significant difference in wrist flexion, extension, or forearm rotation. In addition, mean (SD) grip strength was unaffected (P = .132) by presence or absence of USF with DRF overall, 45.45% (31.92) of contralateral versus 52.88% (30.03). However, grip strength was negatively affected (P = .035) by presence of USF in the nonoperative group, 37.79% (20.58) versus 54.52% (31.89) (Table).
Discussion
In this study, we determined that presence of USF was a negative predictor for clinical outcomes after DRF. Given the higher incidence of USF in operatively treated DRFs, USF likely represents a higher-energy mechanism of injury. We think these inferior clinical results are attributable to other wrist pathologies that commonly occur with these injuries. These pathologies, identified in the past, include stylocarpal impaction, extensor carpi ulnaris tendinitis, and pain at USF site.6,10,15 In addition, intracarpal ligamentous injuries, including damage to scapholunate and lunotriquetral ligaments, have been shown to occur in roughly 80% of patients who sustain DRFs, with TFCC injuries occurring at a rate of 60%.16
Patient outcome is multifactorial and depends on initial injury characteristics, reduction quality, associated injuries, and patient demographics and lifestyle factors. Li and colleagues12 showed that the quality of the DRF reduction influenced outcomes in these injuries, as the ulnar styloid and its associated TFCC are in turn reduced more anatomically with a restored DRF reduction. This concept applies to injuries treated both operatively and nonoperatively. Similarly, Xarchas and colleagues17 identified malunion of the ulnar styloid as causing chronic wrist pain because of triquetral impingement, which was treated successfully with ulnar styloidectomy. The poor results at final follow-up in their study may reflect severity of the initial injury, as reported by Frykman.18
Additional factors may compromise clinical outcomes after such injuries. For example, the effect of USF fragment size on outcome has been suggested and debated. In a retrospective series, May and colleagues6 identified fractures involving the base of the ulnar styloid or fovea as potentially destabilizing the DRUJ and in turn leading to chronic instability. This mechanism should be considered a potential contributor to protracted clinical recovery. Other studies have shown that, irrespective of USF fragment size, presence of USF with DRF is not a reliable predictor of DRUJ instability.2,10,19 In the present study, we simply identified presence or absence of USF, irrespective of either stability or fragment size. In cases in which there was an USF without instability, we fixed the DRF in isolation, without surgically addressing the USF. Our data demonstrated that, even in the absence of DRUJ instability, presence of USF was a negative prognostic indicator for patient outcome.
This study had several limitations. First, its design was retrospective. A prospective study would have been ideal for eliminating certain inherent bias. Second, USF represents a higher association with DRUJ instability.6 As there are no validated tests for this clinical entity, identification is somewhat subjective. We did not separate patients by presence or absence of DRUJ instability and thus were not able to directly correlate the connection between USF, DRUJ instability, and poor outcomes in association with DRF. In addition, management of an unstable DRUJ after operative fixation of DRF is controversial, with techniques ranging from splinting in supination to pinning the DRUJ. This inconsistency likely contributed to some error between groups of patients in this study. Last, we did not stratify patients by USF fragment size, as previously discussed, which may have affected outcomes within patient groups.
Our data add to the evidence showing that USF in association with DRF portends poorer clinical outcomes. Concomitant USF should alert the treating physician to a higher-energy mechanism of injury and raise the index of suspicion for other associated injuries in the carpus.
1. Richards RS, Bennett JD, Roth JH, Milne K Jr. Arthroscopic diagnosis of intra-articular soft tissue injuries associated with distal radial fractures. J Hand Surg Am. 1997;22(5):772-776.
2. Sammer DM, Shah HM, Shauver MJ, Chung KC. The effect of ulnar styloid fractures on patient-rated outcomes after volar locking plating of distal radius fractures. J Hand Surg Am. 2009;34(9):1595-1602.
3. Villar RN, Marsh D, Rushton N, Greatorex RA. Three years after Colles’ fracture. A prospective review. J Bone Joint Surg Br. 1987;69(4):635-638.
4. Palmer AK, Werner FW. The triangular fibrocartilage complex of the wrist—anatomy and function. J Hand Surg Am. 1981;6(2):153-162.
5. Stuart PR, Berger RA, Linscheid RL, An KN. The dorsopalmar stability of the distal radioulnar joint. J Hand Surg Am. 2000;25(4):689-699.
6. May MM, Lawton JN, Blazar PE. Ulnar styloid fractures associated with distal radius fractures: incidence and implications for distal radioulnar joint instability. J Hand Surg Am. 2002;27(6):965-971.
7. Oskarsson GV, Aaser P, Hjall A. Do we underestimate the predictive value of the ulnar styloid affection in Colles fractures? Arch Orthop Trauma Surg. 1997;116(6-7):341-344.
8. Stoffelen D, De Smet L, Broos P. The importance of the distal radioulnar joint in distal radial fractures. J Hand Surg Br. 1998;23(4):507-511.
9. Buijze GA, Ring D. Clinical impact of united versus nonunited fractures of the proximal half of the ulnar styloid following volar plate fixation of the distal radius. J Hand Surg Am. 2010;35(2):223-227.
10. Kim JK, Yun YH, Kim DJ, Yun GU. Comparison of united and nonunited fractures of the ulnar styloid following volar-plate fixation of distal radius fractures. Injury. 2011;42(4):371-375.
11. Wijffels M, Ring D. The influence of non-union of the ulnar styloid on pain, wrist function and instability after distal radius fracture. J Hand Microsurg. 2011;3(1):11-14.
12. Li S, Chen Y, Lin Z, Fan Q, Cui W, Feng Z. Effect of associated ulnar styloid fracture on wrist function after distal radius [in Chinese]. Zhongguo Xiu Fu Chong Jian Wai Ke Za Zhi. 2012;26(6):666-670.
13. Egol KA, Walsh M, Romo-Cardoso S, Dorsky S, Paksima N. Distal radial fractures in the elderly: operative compared with nonoperative treatment. J Bone Joint Surg Am. 2010;92(9):1851-1857.
14. Cooney WP 3rd, Linscheid RL, Dobyns JH. External pin fixation for unstable Colles’ fractures. J Bone Joint Surg Am. 1979;61(6):840-845.
15. Cerezal L, del Piñal F, Abascal F, García-Valtuille R, Pereda T, Canga A. Imaging findings in ulnar-sided wrist impaction syndromes. Radiographics. 2002;22(1):105-121.
16. Ogawa T, Tanaka T, Yanai T, Kumagai H, Ochiai N. Analysis of soft tissue injuries associated with distal radius fractures. BMC Sports Sci Med Rehabil. 2013;5(1):19.
17. Xarchas KC, Yfandithis P, Kazakos K. Malunion of the ulnar styloid as a cause of ulnar wrist pain. Clin Anat. 2004;17(5):418-422.
18. Frykman G. Fracture of the distal radius including sequelae—shoulder–hand–finger syndrome, disturbance in the distal radio-ulnar joint and impairment of nerve function. A clinical and experimental study. Acta Orthop Scand. 1967:(suppl 108):3+.
19. Fujitani R, Omokawa S, Akahane M, Iida A, Ono H, Tanaka Y. Predictors of distal radioulnar joint instability in distal radius fractures. J Hand Surg Am. 2011;36(12):1919-1925.
1. Richards RS, Bennett JD, Roth JH, Milne K Jr. Arthroscopic diagnosis of intra-articular soft tissue injuries associated with distal radial fractures. J Hand Surg Am. 1997;22(5):772-776.
2. Sammer DM, Shah HM, Shauver MJ, Chung KC. The effect of ulnar styloid fractures on patient-rated outcomes after volar locking plating of distal radius fractures. J Hand Surg Am. 2009;34(9):1595-1602.
3. Villar RN, Marsh D, Rushton N, Greatorex RA. Three years after Colles’ fracture. A prospective review. J Bone Joint Surg Br. 1987;69(4):635-638.
4. Palmer AK, Werner FW. The triangular fibrocartilage complex of the wrist—anatomy and function. J Hand Surg Am. 1981;6(2):153-162.
5. Stuart PR, Berger RA, Linscheid RL, An KN. The dorsopalmar stability of the distal radioulnar joint. J Hand Surg Am. 2000;25(4):689-699.
6. May MM, Lawton JN, Blazar PE. Ulnar styloid fractures associated with distal radius fractures: incidence and implications for distal radioulnar joint instability. J Hand Surg Am. 2002;27(6):965-971.
7. Oskarsson GV, Aaser P, Hjall A. Do we underestimate the predictive value of the ulnar styloid affection in Colles fractures? Arch Orthop Trauma Surg. 1997;116(6-7):341-344.
8. Stoffelen D, De Smet L, Broos P. The importance of the distal radioulnar joint in distal radial fractures. J Hand Surg Br. 1998;23(4):507-511.
9. Buijze GA, Ring D. Clinical impact of united versus nonunited fractures of the proximal half of the ulnar styloid following volar plate fixation of the distal radius. J Hand Surg Am. 2010;35(2):223-227.
10. Kim JK, Yun YH, Kim DJ, Yun GU. Comparison of united and nonunited fractures of the ulnar styloid following volar-plate fixation of distal radius fractures. Injury. 2011;42(4):371-375.
11. Wijffels M, Ring D. The influence of non-union of the ulnar styloid on pain, wrist function and instability after distal radius fracture. J Hand Microsurg. 2011;3(1):11-14.
12. Li S, Chen Y, Lin Z, Fan Q, Cui W, Feng Z. Effect of associated ulnar styloid fracture on wrist function after distal radius [in Chinese]. Zhongguo Xiu Fu Chong Jian Wai Ke Za Zhi. 2012;26(6):666-670.
13. Egol KA, Walsh M, Romo-Cardoso S, Dorsky S, Paksima N. Distal radial fractures in the elderly: operative compared with nonoperative treatment. J Bone Joint Surg Am. 2010;92(9):1851-1857.
14. Cooney WP 3rd, Linscheid RL, Dobyns JH. External pin fixation for unstable Colles’ fractures. J Bone Joint Surg Am. 1979;61(6):840-845.
15. Cerezal L, del Piñal F, Abascal F, García-Valtuille R, Pereda T, Canga A. Imaging findings in ulnar-sided wrist impaction syndromes. Radiographics. 2002;22(1):105-121.
16. Ogawa T, Tanaka T, Yanai T, Kumagai H, Ochiai N. Analysis of soft tissue injuries associated with distal radius fractures. BMC Sports Sci Med Rehabil. 2013;5(1):19.
17. Xarchas KC, Yfandithis P, Kazakos K. Malunion of the ulnar styloid as a cause of ulnar wrist pain. Clin Anat. 2004;17(5):418-422.
18. Frykman G. Fracture of the distal radius including sequelae—shoulder–hand–finger syndrome, disturbance in the distal radio-ulnar joint and impairment of nerve function. A clinical and experimental study. Acta Orthop Scand. 1967:(suppl 108):3+.
19. Fujitani R, Omokawa S, Akahane M, Iida A, Ono H, Tanaka Y. Predictors of distal radioulnar joint instability in distal radius fractures. J Hand Surg Am. 2011;36(12):1919-1925.
