The American Journal of Orthopedics

According to their classic article, Kozinn and Scott¹² outlined the indications for UKA that, in their experience, led to the most successful outcomes, including age >60 years, weight <82 kg, low physical demand, localized arthritis with no full-thickness chondromalacia elsewhere in the joint, intact anterior cruciate ligament, minimal deformity, and flexion >90°. More recently, indications have been expanded to include younger and more active patients, higher body mass index, and some patterns of patellofemoral chondromalacia, with an increasing number of publications reporting successful clinical outcomes in these cohorts as well.^13-17 Taken together, it is clear that the “classic” strict indications for UKA can be safely expanded, which have and will result in an increased number of these procedures being performed above and beyond that which might be predicted based on demographic trends alone.

A growing body of literature has been published on the prevalence and projections of orthopedic procedures in the United States.^18-20 Several studies have focused their analysis on 1 of several large administrative databases, including the Nationwide Inpatient Sample, the 5% Medicare Part B database, and the National Hospital Discharge Survey.^18,20-23 A concern with limiting an analysis of the prevalence of unicompartmental knee arthroplasty to these particular databases is that it may underestimate the “true” number of cases performed in the United States, given that several UKA patients are <65 years and have private insurance, and therefore, would not be captured statistically by a database that collects data on patients ≥65 years.

The purpose of this study was to quantify the current prevalence and epidemiology of UKA in the U.S. patient population. Our hypothesis was that the number of procedures and the procedural rate of UKA are increasing over time. Furthermore, this increase may be attributed to an increase in select age- or gender-based segments of the population. To test this hypothesis, we analyzed 2 separate large claims databases to capture patients over a spectrum of age and inclusive of both private and public payers, including the 5% Medicare Part B database (2002–2011) for patients ≥65 years and the MarketScan database (2004 to June 2011) for patients <65 years. Understanding the accurate trends in the use of UKA on a national scale is important for legislative bodies, healthcare administrators, and physicians.

MATERIALS AND METHODS

The 2002 to 2011 5% sample of the Medicare data (Part B) and the 2004 to June 2012 MarketScan Commercial and Medicare Supplemental databases were used to evaluate the prevalence of UKA in elderly (≥65 years) and younger (<65 years) populations, respectively. The UKA procedures were identified using the CPT code 27446.

The prevalence of UKA was stratified by age, gender, census region, Charlson Comorbidity Index, Medicare buy-in status, and diagnosis. The buy-in status is a proxy for the socioeconomic status as it reflects the state subsidizing the health insurance premium for the beneficiary. The Charlson Comorbidity Index is a composite score that has been used to assess the comorbidity level of a patient by taking into account the number and the severity of comorbid conditions.²⁴ For the elderly population, the rate of UKA was subsequently evaluated based on the number of beneficiaries for that particular age-gender group and year in both databases. Poisson regression was used to evaluate the annual rate of change in the UKA rate for assessing temporal changes considering year as a covariate. Age and gender, as well as 2-way interaction terms for age, gender, and year, were also considered as covariates. 

Continue to: RESULTS...

RESULTS

For the time periods analyzed, a total of 5235 and 23,310 UKA procedures were identified from the 5% Medicare and MarketScan databases, respectively. A peak in the prevalence appeared around 2008 for the elderly population and in 2009 for the younger population (Figure 1). When normalized by the size of the population segment, the rate of UKA was found to be approximately 5 times greater in the elderly population, increasing from 369 in 2002 to 639 in 2008, but plateauing to 561 in 2011. Extrapolating to the 100% Medicare population, these numbers increased to 7380, 12,780, and 11,220, respectively. Temporal changes in the UKA rate were significant, increasing from 24.5 UKAs per 100,000 persons in 2002 to 43.1 UKAs in 2008, followed by a decline to 36.5 in 2011 (P < .0001) (Figure 2). The rates of UKA generally increased from 2002 to 2008 for both males and females in the Medicare cohort; however, the rates of UKA in female patients continuously declined from 2008 onward, whereas the UKA rates in male patients decreased in 2009, followed by an increase in 2010 and 2011 (Figure 2). For the younger population, there was a slight increase in the rate of UKA from 2004 to approximately 2009, after which the rates for both males and females remained relatively steady. When put in the context of the prevalence of TKA, the prevalence of UKA fluctuated during the same time period. In the Medicare population, the prevalence of UKA ranged from 4.3% (2005) to 5.9% (2008) of the TKA prevalence between 2002 and 2011. In the younger MarketScan population, the prevalence of UKA ranged from 6.7% (2005) to 8.9% (2008) between 2004 and June 2012.

The UKA rate differed significantly according to gender (P = .0209), with higher rates for males. Although there were no age-related differences (P = .3723), age–gender interactions were found to be significant (P < .0001). For males, the largest rate of UKA in the most recent year of data was observed in the 70- to 74-year-old group, followed by the 75- to 79- and the 65- to 69-year-old groups (Figure 3). For females, those in the 65- to 69- and the 70- to 74-year-old groups had the highest rate of UKA. In the younger cohort, there were increases in the UKA rates since 2004. These rates appeared to be relatively stable from the 2008 or 2009 period onward, except for females 55–64 years, which demonstrated a steady decline since 2008. Analysis of data obtained over the past few years showed that males 55–64, 65–69, and 70–74 years were the only age–gender groups whose UKA rates appeared to be trending upward.

The vast majority of elderly UKA patients were white (95.5%), and when stratified by census region, the highest proportion of UKA procedures was observed in the South and the Midwest (Figure 4). Furthermore, among patients <65 years, 64.2% had a Charlson score of 0 compared to 40.8% in the elderly group (Figure 5). For the Medicare population, based on their receipt of state subsidies for their insurance premiums, 5.1% of patients were of lower socioeconomic status. Osteoarthritis was diagnosed in 99.4% and 97.3% of the MarketScan and Medicare cohorts, respectively.

In the Medicare cohort, gender (P = .0209) and year of operation (P < .0001) were found to be significantly associated with the rate of UKA, along with age-gender (P < .0001) and gender-year (P = .0202) interaction terms. In the MarketScan cohort, age (P = .0173), gender (P = .0017), and year of operation (P = .0002) were found to be significantly associated with UKA rate. Two-way interactions between age-gender (P = .0018), age–year (P = .0207), and gender-year (P = .0017) were also found to be statistically significant factors.

Continue to: DISCUSSION...

The results of our study indicate that between 2002 and 2011, a steadily increasing number of UKA procedures was performed in the United States, and a significant proportion of the surgeries was performed on patients <65 years. Without the MarketScan database data, we would have missed more than 23,000 UKA cases performed during this 10-year time period. This finding validates our research methodology that incorporated data on privately insured younger (<65 years) patients, which is something that has not been done when examining the epidemiology of UKA.

To our knowledge, there are only 2 other publications attempting to quantify the incidence of UKA procedures performed in the United States. Bolognesi and colleagues²³ used the Medicare 5% sample to assess trends in the use of knee arthroplasty from 2000 to 2009. The authors reported that a total of 68,603 patients underwent unilateral total knee arthroplasty (n = 65,505) or unicompartmental knee arthroplasty (n = 3098) over this 10-year time period. Given that there is substantial overlap of our time periods, it is not surprising that our Medicare numbers are similar (3098 vs 5235). In their study, the use of TKA increased 1.7-fold, whereas the use of UKA increased 6.2-fold²³. In our analysis of the Medicare (2011 vs 2002) and MarketScan (2011 vs 2004) databases, there was a 1.3-fold and a 3.4-fold increase in the number of TKAs performed. Concomitantly, the use of UKA increased 1.5-fold and 2.8-fold, respectively, in these databases over the same time periods. The reason for the slight discrepancy in the numbers may be attributable to the peak occurring in 2008. The other publication on the subject by Riddle and colleagues⁸ focused on the time period 1998 to 2005 and used implant manufacturer’s sales data cross-referenced to a database of 44 hospitals to derive their national estimates. Using their unique methodology, the authors calculated an incidence of UKA, ranging from 6570 implants in 1998 to 44,990 in 2005. They reported that UKA use during the study period increased by 3 times the rate of TKA in the United States, with an average yearly percentage increase in the number of UKA procedures of 32.5% compared to 9.4% for TKA procedures. It is difficult to account for the discrepancy in the number of UKAs performed reported between our current study and that of Riddle and colleagues;⁸ however, the fact that the authors used implant manufacturer’s individual sales numbers may indicate that a portion of UKA patients was not captured in either the Medicare 5% or the MarketScan database. Nonetheless, in our analysis, the annual increase in the number of UKA procedures performed during the time periods studied averaged 5.8% in the older population and 25.4% in the younger population compared to the increase in the number of TKA procedures, which averaged 3.6% and 33.9% in the older and younger populations, respectively. In addition, in our study, the percentage of UKAs performed relative to the number of TKAs during the time intervals studied varied from a low of 4.3% to a high of 5.9% in the older population and from a low of 6.7% to a high of 8.9% in the younger population.

During the 10-year period of this study, a general upward trend appeared in the total number of unicompartmental knee arthroplasties performed in both the Medicare and the MarketScan databases. The rate at which the procedure was performed increased in the Medicare population from 24.5 to 36.5 (per 100,000 persons) over a 10-year time period and in the MarketScan cohort from 5.9 to 7.4 (per 100,000 persons) over an 8.5-year time period. This indicates both a larger absolute and a relative rate increase in UKA procedures in the elderly population. Around 2008 and 2009, the data showed a slight dip in the rate of UKA in the Medicare population and a plateau in the rate in the MarketScan database. Although this may be a spurious finding in the data that would be smoothed out with a longer time period investigated, it is interesting that this finding coincided with a national economic downturn. Although it might be expected that macroeconomics may affect the utilization of elective surgery such as total joint replacement, Kurtz and colleagues²⁵ investigated this particular question and found that neither the economic downturns of 2001 or those of 2008 and 2009 had a significant impact on the incidence of total joint replacement surgeries.

Incorporation of the MarketScan database data indicated that a significant proportion of patients undergoing UKA were <65 years and that there was a slight but increasing rate of procedures performed on this age cohort over the past decade. A similar finding has been reported in the Finnish Arthroplasty Registry. Leskinen and colleagues²⁶ reported that the incidence of UKAs among individuals 30 to 59 years increased from 0.2 (per 100,000 persons) to 10 (per 100,000 persons) from 1980 to 2006 and that most of the increase occurred among patients 50 to 59 years. The fact that younger age is no longer observed as a relative contraindication to this procedure is supported by several clinical investigations. Cartier and colleagues²⁷ reported 93% survival at 10 years in patients with a mean age of 65 years, but included patients as young as 28 years, claiming that the results for younger patients were no worse than those for older patients in the series. Pandit and colleagues¹⁷ compared the results of 245 young patients (<60 years) to those of 755 older patients (>60 years) and found a survival rate of 97% at 10 years, with no significant difference in mean functional outcomes, failure rate, or survival between the groups at >5 years of follow-up. Given that patients <65 years now account for approximately half of the TKAs performed each year, with the greatest increase in volume among patients between 45 and 54, it is clear that investigations on the epidemiology of UKA must take into account this increasingly relevant younger patient cohort.²⁸

Continue to: Our data indicate...

Our data indicate that only approximately 5% of UKA patients were non-white and another 5% were from lower socioeconomic status. These findings have been observed in multiple other studies looking at the epidemiology of total joint replacement in the United States.²⁹ Bolognesi and colleagues²³ reported that although “non-white race” patients made up 12% of the general Medicare sample they were analyzing, these patients accounted for only 5% and 3% of the total knee arthroplasty and unicompartmental knee arthroplasty populations, respectively. Although it is beyond the scope of this paper to delve into the reasons for this discrepancy, it may be related to differences in access to care, healthcare literacy, and trust of patients in the healthcare system.^30,31

Our study, like all those based on administrative claims, has several notable inherent limitations. Coding inaccuracies as well as the potential for systematic bias (eg, underreporting) may affect the accuracy of our results. Although the MarketScan Commercial Research Database (Truven Health Analytics) includes nationally representative information for >180 million patients covered with private insurance, it is possible that we have missed some patients who underwent UKA during the time period investigated. However, we feel that the number missed is probably small and does not affect our conclusions in any meaningful manner.

CONCLUSION

This novel analysis of 2 separate administrative claims databases, which more accurately captures all patients undergoing UKA, indicates that there has been a steady increase in the rate of the procedure over the past decade and that a significant proportion of the surgeries were performed in younger (<65 years) patients. Understanding the accurate trends in the use of UKA on a national scale is important for legislative bodies, healthcare administrators, as well as physicians. Furthermore, given the increasing rates of UKA in patients <65 years old, and the increased burden on implants for withstanding increased activities and repetitive loads, it remains imperative to strive to optimize materials, implant designs, and surgical techniques to enhance implant durability.

According to their classic article, Kozinn and Scott¹² outlined the indications for UKA that, in their experience, led to the most successful outcomes, including age >60 years, weight <82 kg, low physical demand, localized arthritis with no full-thickness chondromalacia elsewhere in the joint, intact anterior cruciate ligament, minimal deformity, and flexion >90°. More recently, indications have been expanded to include younger and more active patients, higher body mass index, and some patterns of patellofemoral chondromalacia, with an increasing number of publications reporting successful clinical outcomes in these cohorts as well.^13-17 Taken together, it is clear that the “classic” strict indications for UKA can be safely expanded, which have and will result in an increased number of these procedures being performed above and beyond that which might be predicted based on demographic trends alone.

A growing body of literature has been published on the prevalence and projections of orthopedic procedures in the United States.^18-20 Several studies have focused their analysis on 1 of several large administrative databases, including the Nationwide Inpatient Sample, the 5% Medicare Part B database, and the National Hospital Discharge Survey.^18,20-23 A concern with limiting an analysis of the prevalence of unicompartmental knee arthroplasty to these particular databases is that it may underestimate the “true” number of cases performed in the United States, given that several UKA patients are <65 years and have private insurance, and therefore, would not be captured statistically by a database that collects data on patients ≥65 years.

The purpose of this study was to quantify the current prevalence and epidemiology of UKA in the U.S. patient population. Our hypothesis was that the number of procedures and the procedural rate of UKA are increasing over time. Furthermore, this increase may be attributed to an increase in select age- or gender-based segments of the population. To test this hypothesis, we analyzed 2 separate large claims databases to capture patients over a spectrum of age and inclusive of both private and public payers, including the 5% Medicare Part B database (2002–2011) for patients ≥65 years and the MarketScan database (2004 to June 2011) for patients <65 years. Understanding the accurate trends in the use of UKA on a national scale is important for legislative bodies, healthcare administrators, and physicians.

MATERIALS AND METHODS

The 2002 to 2011 5% sample of the Medicare data (Part B) and the 2004 to June 2012 MarketScan Commercial and Medicare Supplemental databases were used to evaluate the prevalence of UKA in elderly (≥65 years) and younger (<65 years) populations, respectively. The UKA procedures were identified using the CPT code 27446.

The prevalence of UKA was stratified by age, gender, census region, Charlson Comorbidity Index, Medicare buy-in status, and diagnosis. The buy-in status is a proxy for the socioeconomic status as it reflects the state subsidizing the health insurance premium for the beneficiary. The Charlson Comorbidity Index is a composite score that has been used to assess the comorbidity level of a patient by taking into account the number and the severity of comorbid conditions.²⁴ For the elderly population, the rate of UKA was subsequently evaluated based on the number of beneficiaries for that particular age-gender group and year in both databases. Poisson regression was used to evaluate the annual rate of change in the UKA rate for assessing temporal changes considering year as a covariate. Age and gender, as well as 2-way interaction terms for age, gender, and year, were also considered as covariates. 

Continue to: RESULTS...

RESULTS

For the time periods analyzed, a total of 5235 and 23,310 UKA procedures were identified from the 5% Medicare and MarketScan databases, respectively. A peak in the prevalence appeared around 2008 for the elderly population and in 2009 for the younger population (Figure 1). When normalized by the size of the population segment, the rate of UKA was found to be approximately 5 times greater in the elderly population, increasing from 369 in 2002 to 639 in 2008, but plateauing to 561 in 2011. Extrapolating to the 100% Medicare population, these numbers increased to 7380, 12,780, and 11,220, respectively. Temporal changes in the UKA rate were significant, increasing from 24.5 UKAs per 100,000 persons in 2002 to 43.1 UKAs in 2008, followed by a decline to 36.5 in 2011 (P < .0001) (Figure 2). The rates of UKA generally increased from 2002 to 2008 for both males and females in the Medicare cohort; however, the rates of UKA in female patients continuously declined from 2008 onward, whereas the UKA rates in male patients decreased in 2009, followed by an increase in 2010 and 2011 (Figure 2). For the younger population, there was a slight increase in the rate of UKA from 2004 to approximately 2009, after which the rates for both males and females remained relatively steady. When put in the context of the prevalence of TKA, the prevalence of UKA fluctuated during the same time period. In the Medicare population, the prevalence of UKA ranged from 4.3% (2005) to 5.9% (2008) of the TKA prevalence between 2002 and 2011. In the younger MarketScan population, the prevalence of UKA ranged from 6.7% (2005) to 8.9% (2008) between 2004 and June 2012.

The UKA rate differed significantly according to gender (P = .0209), with higher rates for males. Although there were no age-related differences (P = .3723), age–gender interactions were found to be significant (P < .0001). For males, the largest rate of UKA in the most recent year of data was observed in the 70- to 74-year-old group, followed by the 75- to 79- and the 65- to 69-year-old groups (Figure 3). For females, those in the 65- to 69- and the 70- to 74-year-old groups had the highest rate of UKA. In the younger cohort, there were increases in the UKA rates since 2004. These rates appeared to be relatively stable from the 2008 or 2009 period onward, except for females 55–64 years, which demonstrated a steady decline since 2008. Analysis of data obtained over the past few years showed that males 55–64, 65–69, and 70–74 years were the only age–gender groups whose UKA rates appeared to be trending upward.

The vast majority of elderly UKA patients were white (95.5%), and when stratified by census region, the highest proportion of UKA procedures was observed in the South and the Midwest (Figure 4). Furthermore, among patients <65 years, 64.2% had a Charlson score of 0 compared to 40.8% in the elderly group (Figure 5). For the Medicare population, based on their receipt of state subsidies for their insurance premiums, 5.1% of patients were of lower socioeconomic status. Osteoarthritis was diagnosed in 99.4% and 97.3% of the MarketScan and Medicare cohorts, respectively.

In the Medicare cohort, gender (P = .0209) and year of operation (P < .0001) were found to be significantly associated with the rate of UKA, along with age-gender (P < .0001) and gender-year (P = .0202) interaction terms. In the MarketScan cohort, age (P = .0173), gender (P = .0017), and year of operation (P = .0002) were found to be significantly associated with UKA rate. Two-way interactions between age-gender (P = .0018), age–year (P = .0207), and gender-year (P = .0017) were also found to be statistically significant factors.

Continue to: DISCUSSION...

The results of our study indicate that between 2002 and 2011, a steadily increasing number of UKA procedures was performed in the United States, and a significant proportion of the surgeries was performed on patients <65 years. Without the MarketScan database data, we would have missed more than 23,000 UKA cases performed during this 10-year time period. This finding validates our research methodology that incorporated data on privately insured younger (<65 years) patients, which is something that has not been done when examining the epidemiology of UKA.

To our knowledge, there are only 2 other publications attempting to quantify the incidence of UKA procedures performed in the United States. Bolognesi and colleagues²³ used the Medicare 5% sample to assess trends in the use of knee arthroplasty from 2000 to 2009. The authors reported that a total of 68,603 patients underwent unilateral total knee arthroplasty (n = 65,505) or unicompartmental knee arthroplasty (n = 3098) over this 10-year time period. Given that there is substantial overlap of our time periods, it is not surprising that our Medicare numbers are similar (3098 vs 5235). In their study, the use of TKA increased 1.7-fold, whereas the use of UKA increased 6.2-fold²³. In our analysis of the Medicare (2011 vs 2002) and MarketScan (2011 vs 2004) databases, there was a 1.3-fold and a 3.4-fold increase in the number of TKAs performed. Concomitantly, the use of UKA increased 1.5-fold and 2.8-fold, respectively, in these databases over the same time periods. The reason for the slight discrepancy in the numbers may be attributable to the peak occurring in 2008. The other publication on the subject by Riddle and colleagues⁸ focused on the time period 1998 to 2005 and used implant manufacturer’s sales data cross-referenced to a database of 44 hospitals to derive their national estimates. Using their unique methodology, the authors calculated an incidence of UKA, ranging from 6570 implants in 1998 to 44,990 in 2005. They reported that UKA use during the study period increased by 3 times the rate of TKA in the United States, with an average yearly percentage increase in the number of UKA procedures of 32.5% compared to 9.4% for TKA procedures. It is difficult to account for the discrepancy in the number of UKAs performed reported between our current study and that of Riddle and colleagues;⁸ however, the fact that the authors used implant manufacturer’s individual sales numbers may indicate that a portion of UKA patients was not captured in either the Medicare 5% or the MarketScan database. Nonetheless, in our analysis, the annual increase in the number of UKA procedures performed during the time periods studied averaged 5.8% in the older population and 25.4% in the younger population compared to the increase in the number of TKA procedures, which averaged 3.6% and 33.9% in the older and younger populations, respectively. In addition, in our study, the percentage of UKAs performed relative to the number of TKAs during the time intervals studied varied from a low of 4.3% to a high of 5.9% in the older population and from a low of 6.7% to a high of 8.9% in the younger population.

During the 10-year period of this study, a general upward trend appeared in the total number of unicompartmental knee arthroplasties performed in both the Medicare and the MarketScan databases. The rate at which the procedure was performed increased in the Medicare population from 24.5 to 36.5 (per 100,000 persons) over a 10-year time period and in the MarketScan cohort from 5.9 to 7.4 (per 100,000 persons) over an 8.5-year time period. This indicates both a larger absolute and a relative rate increase in UKA procedures in the elderly population. Around 2008 and 2009, the data showed a slight dip in the rate of UKA in the Medicare population and a plateau in the rate in the MarketScan database. Although this may be a spurious finding in the data that would be smoothed out with a longer time period investigated, it is interesting that this finding coincided with a national economic downturn. Although it might be expected that macroeconomics may affect the utilization of elective surgery such as total joint replacement, Kurtz and colleagues²⁵ investigated this particular question and found that neither the economic downturns of 2001 or those of 2008 and 2009 had a significant impact on the incidence of total joint replacement surgeries.

Incorporation of the MarketScan database data indicated that a significant proportion of patients undergoing UKA were <65 years and that there was a slight but increasing rate of procedures performed on this age cohort over the past decade. A similar finding has been reported in the Finnish Arthroplasty Registry. Leskinen and colleagues²⁶ reported that the incidence of UKAs among individuals 30 to 59 years increased from 0.2 (per 100,000 persons) to 10 (per 100,000 persons) from 1980 to 2006 and that most of the increase occurred among patients 50 to 59 years. The fact that younger age is no longer observed as a relative contraindication to this procedure is supported by several clinical investigations. Cartier and colleagues²⁷ reported 93% survival at 10 years in patients with a mean age of 65 years, but included patients as young as 28 years, claiming that the results for younger patients were no worse than those for older patients in the series. Pandit and colleagues¹⁷ compared the results of 245 young patients (<60 years) to those of 755 older patients (>60 years) and found a survival rate of 97% at 10 years, with no significant difference in mean functional outcomes, failure rate, or survival between the groups at >5 years of follow-up. Given that patients <65 years now account for approximately half of the TKAs performed each year, with the greatest increase in volume among patients between 45 and 54, it is clear that investigations on the epidemiology of UKA must take into account this increasingly relevant younger patient cohort.²⁸

Continue to: Our data indicate...

Our data indicate that only approximately 5% of UKA patients were non-white and another 5% were from lower socioeconomic status. These findings have been observed in multiple other studies looking at the epidemiology of total joint replacement in the United States.²⁹ Bolognesi and colleagues²³ reported that although “non-white race” patients made up 12% of the general Medicare sample they were analyzing, these patients accounted for only 5% and 3% of the total knee arthroplasty and unicompartmental knee arthroplasty populations, respectively. Although it is beyond the scope of this paper to delve into the reasons for this discrepancy, it may be related to differences in access to care, healthcare literacy, and trust of patients in the healthcare system.^30,31

Our study, like all those based on administrative claims, has several notable inherent limitations. Coding inaccuracies as well as the potential for systematic bias (eg, underreporting) may affect the accuracy of our results. Although the MarketScan Commercial Research Database (Truven Health Analytics) includes nationally representative information for >180 million patients covered with private insurance, it is possible that we have missed some patients who underwent UKA during the time period investigated. However, we feel that the number missed is probably small and does not affect our conclusions in any meaningful manner.

CONCLUSION

This novel analysis of 2 separate administrative claims databases, which more accurately captures all patients undergoing UKA, indicates that there has been a steady increase in the rate of the procedure over the past decade and that a significant proportion of the surgeries were performed in younger (<65 years) patients. Understanding the accurate trends in the use of UKA on a national scale is important for legislative bodies, healthcare administrators, as well as physicians. Furthermore, given the increasing rates of UKA in patients <65 years old, and the increased burden on implants for withstanding increased activities and repetitive loads, it remains imperative to strive to optimize materials, implant designs, and surgical techniques to enhance implant durability.

According to their classic article, Kozinn and Scott¹² outlined the indications for UKA that, in their experience, led to the most successful outcomes, including age >60 years, weight <82 kg, low physical demand, localized arthritis with no full-thickness chondromalacia elsewhere in the joint, intact anterior cruciate ligament, minimal deformity, and flexion >90°. More recently, indications have been expanded to include younger and more active patients, higher body mass index, and some patterns of patellofemoral chondromalacia, with an increasing number of publications reporting successful clinical outcomes in these cohorts as well.^13-17 Taken together, it is clear that the “classic” strict indications for UKA can be safely expanded, which have and will result in an increased number of these procedures being performed above and beyond that which might be predicted based on demographic trends alone.

A growing body of literature has been published on the prevalence and projections of orthopedic procedures in the United States.^18-20 Several studies have focused their analysis on 1 of several large administrative databases, including the Nationwide Inpatient Sample, the 5% Medicare Part B database, and the National Hospital Discharge Survey.^18,20-23 A concern with limiting an analysis of the prevalence of unicompartmental knee arthroplasty to these particular databases is that it may underestimate the “true” number of cases performed in the United States, given that several UKA patients are <65 years and have private insurance, and therefore, would not be captured statistically by a database that collects data on patients ≥65 years.

The purpose of this study was to quantify the current prevalence and epidemiology of UKA in the U.S. patient population. Our hypothesis was that the number of procedures and the procedural rate of UKA are increasing over time. Furthermore, this increase may be attributed to an increase in select age- or gender-based segments of the population. To test this hypothesis, we analyzed 2 separate large claims databases to capture patients over a spectrum of age and inclusive of both private and public payers, including the 5% Medicare Part B database (2002–2011) for patients ≥65 years and the MarketScan database (2004 to June 2011) for patients <65 years. Understanding the accurate trends in the use of UKA on a national scale is important for legislative bodies, healthcare administrators, and physicians.

MATERIALS AND METHODS

The 2002 to 2011 5% sample of the Medicare data (Part B) and the 2004 to June 2012 MarketScan Commercial and Medicare Supplemental databases were used to evaluate the prevalence of UKA in elderly (≥65 years) and younger (<65 years) populations, respectively. The UKA procedures were identified using the CPT code 27446.

The prevalence of UKA was stratified by age, gender, census region, Charlson Comorbidity Index, Medicare buy-in status, and diagnosis. The buy-in status is a proxy for the socioeconomic status as it reflects the state subsidizing the health insurance premium for the beneficiary. The Charlson Comorbidity Index is a composite score that has been used to assess the comorbidity level of a patient by taking into account the number and the severity of comorbid conditions.²⁴ For the elderly population, the rate of UKA was subsequently evaluated based on the number of beneficiaries for that particular age-gender group and year in both databases. Poisson regression was used to evaluate the annual rate of change in the UKA rate for assessing temporal changes considering year as a covariate. Age and gender, as well as 2-way interaction terms for age, gender, and year, were also considered as covariates. 

Continue to: RESULTS...

RESULTS

For the time periods analyzed, a total of 5235 and 23,310 UKA procedures were identified from the 5% Medicare and MarketScan databases, respectively. A peak in the prevalence appeared around 2008 for the elderly population and in 2009 for the younger population (Figure 1). When normalized by the size of the population segment, the rate of UKA was found to be approximately 5 times greater in the elderly population, increasing from 369 in 2002 to 639 in 2008, but plateauing to 561 in 2011. Extrapolating to the 100% Medicare population, these numbers increased to 7380, 12,780, and 11,220, respectively. Temporal changes in the UKA rate were significant, increasing from 24.5 UKAs per 100,000 persons in 2002 to 43.1 UKAs in 2008, followed by a decline to 36.5 in 2011 (P < .0001) (Figure 2). The rates of UKA generally increased from 2002 to 2008 for both males and females in the Medicare cohort; however, the rates of UKA in female patients continuously declined from 2008 onward, whereas the UKA rates in male patients decreased in 2009, followed by an increase in 2010 and 2011 (Figure 2). For the younger population, there was a slight increase in the rate of UKA from 2004 to approximately 2009, after which the rates for both males and females remained relatively steady. When put in the context of the prevalence of TKA, the prevalence of UKA fluctuated during the same time period. In the Medicare population, the prevalence of UKA ranged from 4.3% (2005) to 5.9% (2008) of the TKA prevalence between 2002 and 2011. In the younger MarketScan population, the prevalence of UKA ranged from 6.7% (2005) to 8.9% (2008) between 2004 and June 2012.

The UKA rate differed significantly according to gender (P = .0209), with higher rates for males. Although there were no age-related differences (P = .3723), age–gender interactions were found to be significant (P < .0001). For males, the largest rate of UKA in the most recent year of data was observed in the 70- to 74-year-old group, followed by the 75- to 79- and the 65- to 69-year-old groups (Figure 3). For females, those in the 65- to 69- and the 70- to 74-year-old groups had the highest rate of UKA. In the younger cohort, there were increases in the UKA rates since 2004. These rates appeared to be relatively stable from the 2008 or 2009 period onward, except for females 55–64 years, which demonstrated a steady decline since 2008. Analysis of data obtained over the past few years showed that males 55–64, 65–69, and 70–74 years were the only age–gender groups whose UKA rates appeared to be trending upward.

The vast majority of elderly UKA patients were white (95.5%), and when stratified by census region, the highest proportion of UKA procedures was observed in the South and the Midwest (Figure 4). Furthermore, among patients <65 years, 64.2% had a Charlson score of 0 compared to 40.8% in the elderly group (Figure 5). For the Medicare population, based on their receipt of state subsidies for their insurance premiums, 5.1% of patients were of lower socioeconomic status. Osteoarthritis was diagnosed in 99.4% and 97.3% of the MarketScan and Medicare cohorts, respectively.

In the Medicare cohort, gender (P = .0209) and year of operation (P < .0001) were found to be significantly associated with the rate of UKA, along with age-gender (P < .0001) and gender-year (P = .0202) interaction terms. In the MarketScan cohort, age (P = .0173), gender (P = .0017), and year of operation (P = .0002) were found to be significantly associated with UKA rate. Two-way interactions between age-gender (P = .0018), age–year (P = .0207), and gender-year (P = .0017) were also found to be statistically significant factors.

Continue to: DISCUSSION...

The results of our study indicate that between 2002 and 2011, a steadily increasing number of UKA procedures was performed in the United States, and a significant proportion of the surgeries was performed on patients <65 years. Without the MarketScan database data, we would have missed more than 23,000 UKA cases performed during this 10-year time period. This finding validates our research methodology that incorporated data on privately insured younger (<65 years) patients, which is something that has not been done when examining the epidemiology of UKA.

To our knowledge, there are only 2 other publications attempting to quantify the incidence of UKA procedures performed in the United States. Bolognesi and colleagues²³ used the Medicare 5% sample to assess trends in the use of knee arthroplasty from 2000 to 2009. The authors reported that a total of 68,603 patients underwent unilateral total knee arthroplasty (n = 65,505) or unicompartmental knee arthroplasty (n = 3098) over this 10-year time period. Given that there is substantial overlap of our time periods, it is not surprising that our Medicare numbers are similar (3098 vs 5235). In their study, the use of TKA increased 1.7-fold, whereas the use of UKA increased 6.2-fold²³. In our analysis of the Medicare (2011 vs 2002) and MarketScan (2011 vs 2004) databases, there was a 1.3-fold and a 3.4-fold increase in the number of TKAs performed. Concomitantly, the use of UKA increased 1.5-fold and 2.8-fold, respectively, in these databases over the same time periods. The reason for the slight discrepancy in the numbers may be attributable to the peak occurring in 2008. The other publication on the subject by Riddle and colleagues⁸ focused on the time period 1998 to 2005 and used implant manufacturer’s sales data cross-referenced to a database of 44 hospitals to derive their national estimates. Using their unique methodology, the authors calculated an incidence of UKA, ranging from 6570 implants in 1998 to 44,990 in 2005. They reported that UKA use during the study period increased by 3 times the rate of TKA in the United States, with an average yearly percentage increase in the number of UKA procedures of 32.5% compared to 9.4% for TKA procedures. It is difficult to account for the discrepancy in the number of UKAs performed reported between our current study and that of Riddle and colleagues;⁸ however, the fact that the authors used implant manufacturer’s individual sales numbers may indicate that a portion of UKA patients was not captured in either the Medicare 5% or the MarketScan database. Nonetheless, in our analysis, the annual increase in the number of UKA procedures performed during the time periods studied averaged 5.8% in the older population and 25.4% in the younger population compared to the increase in the number of TKA procedures, which averaged 3.6% and 33.9% in the older and younger populations, respectively. In addition, in our study, the percentage of UKAs performed relative to the number of TKAs during the time intervals studied varied from a low of 4.3% to a high of 5.9% in the older population and from a low of 6.7% to a high of 8.9% in the younger population.

During the 10-year period of this study, a general upward trend appeared in the total number of unicompartmental knee arthroplasties performed in both the Medicare and the MarketScan databases. The rate at which the procedure was performed increased in the Medicare population from 24.5 to 36.5 (per 100,000 persons) over a 10-year time period and in the MarketScan cohort from 5.9 to 7.4 (per 100,000 persons) over an 8.5-year time period. This indicates both a larger absolute and a relative rate increase in UKA procedures in the elderly population. Around 2008 and 2009, the data showed a slight dip in the rate of UKA in the Medicare population and a plateau in the rate in the MarketScan database. Although this may be a spurious finding in the data that would be smoothed out with a longer time period investigated, it is interesting that this finding coincided with a national economic downturn. Although it might be expected that macroeconomics may affect the utilization of elective surgery such as total joint replacement, Kurtz and colleagues²⁵ investigated this particular question and found that neither the economic downturns of 2001 or those of 2008 and 2009 had a significant impact on the incidence of total joint replacement surgeries.

Incorporation of the MarketScan database data indicated that a significant proportion of patients undergoing UKA were <65 years and that there was a slight but increasing rate of procedures performed on this age cohort over the past decade. A similar finding has been reported in the Finnish Arthroplasty Registry. Leskinen and colleagues²⁶ reported that the incidence of UKAs among individuals 30 to 59 years increased from 0.2 (per 100,000 persons) to 10 (per 100,000 persons) from 1980 to 2006 and that most of the increase occurred among patients 50 to 59 years. The fact that younger age is no longer observed as a relative contraindication to this procedure is supported by several clinical investigations. Cartier and colleagues²⁷ reported 93% survival at 10 years in patients with a mean age of 65 years, but included patients as young as 28 years, claiming that the results for younger patients were no worse than those for older patients in the series. Pandit and colleagues¹⁷ compared the results of 245 young patients (<60 years) to those of 755 older patients (>60 years) and found a survival rate of 97% at 10 years, with no significant difference in mean functional outcomes, failure rate, or survival between the groups at >5 years of follow-up. Given that patients <65 years now account for approximately half of the TKAs performed each year, with the greatest increase in volume among patients between 45 and 54, it is clear that investigations on the epidemiology of UKA must take into account this increasingly relevant younger patient cohort.²⁸

Continue to: Our data indicate...

Our data indicate that only approximately 5% of UKA patients were non-white and another 5% were from lower socioeconomic status. These findings have been observed in multiple other studies looking at the epidemiology of total joint replacement in the United States.²⁹ Bolognesi and colleagues²³ reported that although “non-white race” patients made up 12% of the general Medicare sample they were analyzing, these patients accounted for only 5% and 3% of the total knee arthroplasty and unicompartmental knee arthroplasty populations, respectively. Although it is beyond the scope of this paper to delve into the reasons for this discrepancy, it may be related to differences in access to care, healthcare literacy, and trust of patients in the healthcare system.^30,31

Our study, like all those based on administrative claims, has several notable inherent limitations. Coding inaccuracies as well as the potential for systematic bias (eg, underreporting) may affect the accuracy of our results. Although the MarketScan Commercial Research Database (Truven Health Analytics) includes nationally representative information for >180 million patients covered with private insurance, it is possible that we have missed some patients who underwent UKA during the time period investigated. However, we feel that the number missed is probably small and does not affect our conclusions in any meaningful manner.

CONCLUSION

This novel analysis of 2 separate administrative claims databases, which more accurately captures all patients undergoing UKA, indicates that there has been a steady increase in the rate of the procedure over the past decade and that a significant proportion of the surgeries were performed in younger (<65 years) patients. Understanding the accurate trends in the use of UKA on a national scale is important for legislative bodies, healthcare administrators, as well as physicians. Furthermore, given the increasing rates of UKA in patients <65 years old, and the increased burden on implants for withstanding increased activities and repetitive loads, it remains imperative to strive to optimize materials, implant designs, and surgical techniques to enhance implant durability.

Fragility fractures are not only traumatizing for patients; they are also associated with significantly increased mortality. A study by Gosch and colleagues found that 70.6% of patients died during the normal follow-up period, and 29.4% of patients died within the first year of suffering a fracture.³ Also, the mean life expectancy post-fragility fracture was only 527 days.³ Diagnosis and treatment of osteoporosis is imperative to prevent fragility fractures before they occur.

RISK FACTORS AND CAUSES

The incidence of fragility fractures increases in patients with comorbidities such as thyroid disease, diabetes, hypertension, and heart disease.⁴ Hyperthyroidism and treated hypothyroidism cause an imbalance between osteoblast and osteoclast activity, resulting in osteoporosis.⁵ A thyroid-stimulating hormone level < 0.1 increases the risk of vertebral and non-vertebral fractures by a factor of 4.5 and 3.2 mIU/L respectively.⁴ Patients with diabetes also have an increased risk of fragility fractures, which is due to impaired healing capabilities, especially that of bone healing. Approximately 2 million people are affected by type 1 diabetes in the United States, and 20% of those patients will develop osteoporosis.⁶

Hypertension and osteoporosis are 2 diseases that occur often in the elderly. Common etiological factors believed to cause both hypertension and osteoporosis are low calcium intake, high consumption of salt, and vitamin D and vitamin K deficiency. Also, hypertension treated with loop diuretics has been found to cause negative effects on bone and increase the risk of osteoporosis.⁷ The only antihypertensive medications that preserve bone mineral density (BMD) and reduce fracture risk are thiazide diuretics.⁷ Lastly, an association between coronary artery disease and osteoporosis has been hypothesized. The link is not completely understood, but it is believed that oxidative stress and inflammation are the culprits in both diseases.⁸ In contrast to previous hypotheses, Sosa and colleagues found an independent association between beta blockers and fragility fractures.⁹ The idea that beta blockers and fragility fractures are linked is still controversial and needs more study. Unlike beta blockers, statins provide a protective effect on bone. They increase BMD and reduce fracture risk by inhibiting osteoclastogenesis.¹⁰

In addition to loop diuretics and beta blockers, inhaled glucocorticoids, oral glucocorticoids, proton pump inhibitors (PPIs), H₂ receptor antagonists, and anticonvulsants decrease bone density and increase the incidence of fragility fractures.¹¹ Chronic glucocorticoid therapy is the most common cause of secondary osteoporosis. Osteoblasts and osteocytes undergo apoptosis in the presence of glucocorticoids.¹² Patients on glucocorticoid therapy have an increased risk of fracture, even with higher BMD values.¹³ Bone changes that occur while a patient is taking glucocorticoids may not be detected during BMD testing. Therefore, a high level of suspicion of osteoporosis in patients on long-term glucocorticoids is imperative.

Proton pump inhibitors are among the most prescribed medications in the world; they reduce bone resorption, increasing the risk of fracture.¹⁴ Proton pump inhibitors and H₂ receptor antagonists are hypothesized to cause malabsorption of calcium and indirectly cause osteoporosis. The risk of osteoporosis increases with the length of PPI treatment.¹⁵ However, exposure lasting <7 years does not increase the risk of fracture.¹⁶ It is recommended that patients on long-term PPIs be referred for BMD testing.

An association between anticonvulsants and osteoporosis has been found in observational studies. The mechanism of this association is not yet fully understood, but it is believed that exacerbation of vitamin D deficiency leads to increased bone metabolism.¹⁷ Gastrointestinal (GI) calcium absorption also decreases with anticonvulsant use. Prolonged antiepileptic therapy and high-dose therapy rapidly decrease BMD. Primidone, carbamazepine, phenobarbital, and phenytoin are the drugs most often associated with decreased BMD. Osteoporosis and fragility fracture in these patients can be prevented with calcium, vitamin D, and the bisphosphonate risedronate. These medications have been shown to improve BMD by 69%.¹⁸

Continue to: DIAGNOSIS...

Osteoporosis is diagnosed by the presence of a fragility fracture or by dual-energy x-ray absorptiometry (DXA) in the absence of a fragility fracture.¹⁹ Measurements of the femoral neck by DXA are used to diagnose osteoporosis, although DXA can also be used to measure the bone density of the spine and peripheral skeleton.²⁰

The World Health Organization developed a set of T score criteria to diagnose osteoporosis in postmenopausal women (Table 1). A T score >-1 is normal, <-1 but >-2.5 signifies osteopenia, <-2.5 is osteoporosis, and <-2.5 with fragility fracture is severe osteoporosis.¹⁹ The Z score, not the T score, should be used to assess osteoporosis in premenopausal women, men <50 years, and children (Table 2). The Z score is calculated by comparing the patient’s BMD with the mean BMD of their peers of a similar age, race, and gender.¹⁹ Z scores <-2.0 indicate low BMD for chronological age. A Z score > -2.0 is considered within the expected range for age.²⁰ Bone mineral density testing is the rate- limiting step to starting osteoporosis treatment.²¹ Without testing, treatment of osteoporosis is very unlikely.

Table 1. T Score Criteria

Table 2. Z Score Criteria

The World Health Organization also developed a tool to predict fracture risk. The Fracture Risk Assessment Tool uses fracture history in addition to other risk factors to predict a patient’s 10-year risk of major fracture.²² Risk factors used to assess fracture risk include age, sex, weight, height, previous fracture, parental hip fracture history, current smoker, glucocorticoid use, rheumatoid arthritis, secondary osteoporosis, excessive alcohol use, and femoral neck BMD.

In 2011, the United States Preventive Services Task Force (USPSTF) recommended that all women ≥65 years should be screened for osteoporosis by DXA. Women <65 years with a 10-year fracture risk =/> than that of a 65-year-old white woman should also be screened for osteoporosis. These recommendations are different for men. It was concluded that the evidence was insufficient to support osteoporosis screening in men.²³ As of April 2017, Centers for Medicare and Medicaid Services current reimbursement rates for DXA scans are, on average, $123.10 in the hospital setting and $41.63 in the office setting. The axial DXA CPT code is 77080.

Continue to: TREATMENT...

NONPHARMACOLOGIC

Patients with mild osteoporosis may be treated first non-pharmacologically. Lifestyle changes such as calcium and vitamin D supplementation, exercise, and smoking cessation are non-pharmacologic treatment options. Calcium carbonate and calcium citrate are common supplements. Calcium carbonate is 40% elemental calcium, whereas calcium citrate supplements are only 21% elemental calcium. Calcium supplements are best absorbed when taken with food.²⁴ The recommended daily total calcium intake is 1200 mg.²⁵ Only 500 to 600 milligrams of calcium can be absorbed by the GI tract at a time. Therefore, calcium supplements should be taken at least 4 to 5 hours apart.²⁴Patients should also be counseled that calcium supplements may cause GI side effects such as bloating and constipation. To reduce side effects, patients can slowly increase the dose of calcium to a therapeutic level.

Vitamin D supplementation works best in conjunction with calcium supplementation. Vitamin D functions to regulate calcium absorption in the intestine and stimulate bone resorption and maintain the serum calcium concentration. The National Osteoporosis Foundation recommends 800 to 1000 international units of vitamin D daily.²⁴ Lifestyle changes may be sufficient to stop the progression of osteoporosis in its early stages. Once osteoporosis becomes severe enough, pharmacotherapy is needed to stop further bone destruction and improve BMD.

PHARMACOLOGIC

After an initial fragility fracture, the risk of additional ones increases significantly, making treatment of osteoporosis essential. The National Osteoporosis Foundation recommends treating osteoporosis with pharmacotherapy in patients with a high risk of fracture (T score <-2.5) or history of fragility fracture.²⁶ Bisphosphonates inhibit bone resorption and are considered the first-line therapy for postmenopausal women with osteoporosis. A common side effect of oral bisphosphonates is GI toxicity. Patients are advised to avoid lying down for at least 30 minutes after medication administration to avoid esophageal irritation. Oral bisphosphonates should also be taken in the morning on an empty stomach with at least 8 ounces of water. Recurrent bisphosphonate use should be avoided in patients with chronic kidney disease. Oral alendronate and risedronate are typically discontinued after 5 years of use.²⁷ Long-term bisphosphonate use may cause an increased risk of fragility fracture due to oversuppression of bone turnover. To avoid this risk, bisphosphonate “drug holidays” are an option. Bisphosphonates accumulate over time, creating reservoirs. Even after therapy is stopped, patients continue to have therapeutic effects for 2 to 5 years.²⁸

Bisphosphonates are available in both oral and intravenous forms. Alendronate is available in doses of 10 mg and 70 mg for daily and weekly administration, respectively. Both are available in tablet form, but the 70 mg weekly dose is also available in a dissolvable formulation. Alendronate is available in a reduced dose for osteoporosis prevention. Alendronate dosing for osteoporosis prevention is 5 mg daily or 35 mg weekly. Risedronate is dosed as 5 mg daily, 35 mg weekly, or 150 mg monthly. Intravenous bisphosphonates are indicated when oral bisphosphonates are not tolerated, only after vitamin D has been assessed and is within the normal range. Zoledronic acid is administered as a 15-minute infusion once a year.

Teriparatide (Forteo; PTH-1-34) is available for glucocorticoid-induced osteoporosis, postmenopausal women, and men with severe osteoporosis. It is indicated for patients in whom bisphosphonate treatment has failed or those who do not tolerate bisphosphonates. Teriparatide is a synthetic parathyroid hormone (PTH) that acts as an anabolic agent, stimulating bone formation, maturation, and remodeling.²⁹ In addition to its application as a bone-building hormone, teriparatide has gained popularity for various off-label uses. These include accelerated osteosynthesis, stress fracture healing, and in the nonoperative treatment of osteoarthritis.²⁹ Parathyroid hormone has been shown to stimulate the maturation, proliferation, and maintenance of osteoblast progenitor cells. More recently, PTH has been shown to regulate chondrocyte signaling, as well as differentiation and maturation. Further study on the chondroregenerative potential of PTH has demonstrated its efficacy as a novel disease-modifying agent in the treatment of osteoarthritis.²⁹ Teriparatide is administered as a daily subcutaneous injection. The United States dosing is 600 mcg/2.4 mL. Adverse effects such as orthostatic hypotension and osteosarcoma may occur. BMD testing should be performed 1 to 2 years after initiation of teriparatide and every 2 years thereafter.²⁶

Abaloparatide (Tymlos), a human parathyroid hormone, is another treatment option for postmenopausal women at risk of osteoporotic fracture. In a study comparing the efficacy of abaloparatide and teriparatide, treatment with abaloparatide was found to induce higher BMD levels in a time frame of 12 months. The BMD differences could be attributed to many factors, such as an enhanced net anabolic effect or a reduced osteoblast expression. Furthermore, the risk of developing new vertebral and nonvertebral fractures decreased in the abaloparatide group compared with the placebo group over a period of 18 months.³⁰

Continue to: The recommended daily dose for abaloparatide...

The recommended daily dose for abaloparatide is 80 mcg via subcutaneous injection with calcium and vitamin D supplements.³¹ Adverse reactions were consistent between abaloparatide and teriparatide, and included hypercalcemia, hypercalciuria, and orthostatic hypotension.³⁰ The use of parathyroid analogs for >2 years is not recommended due to the risk of osteosarcoma.

Denosumab (Prolia) is a monoclonal antibody that stops osteoclastogenesis by blocking the binding of RANKL to RANK.³¹ It is indicated for patients intolerant to bisphosphonates or with impaired kidney function. Prolia is administered subcutaneously in 60 mg doses every 6 months in men and postmenopausal women with osteoporosis. Prolia is contraindicated in patients with hypersensitivity to any component of the medication, pregnancy, and hypocalcemia.

Selective estrogen receptor modulators (SERMs), such as raloxifene and tamoxifen, can treat osteoporosis effectively in postmenopausal women. Raloxifene is considered the SERM of choice due to the availability of more robust safety and efficacy data. Raloxifene increases BMD while decreasing bone resorption and bone turnover.³² It is also used to reduce breast cancer risk; however, it increases the risk of thromboembolic events and hot flashes. Tamoxifen is not typically used to treat osteoporosis, but women treated for breast cancer with tamoxifen receive some bone protection.

Lastly, calcitonin and strontium ranelate are also options to treat osteoporosis. However, both calcitonin and strontium ranelate have weak effects on BMD. Calcitonin only transiently inhibits osteoclast activity.³³ Therefore, medications like bisphosphonates, teriparatide, denosumab, and SERMs are preferred.

A summary of medications used to treat osteoporosis can be found in Table 3.

Table 3. Overview of Common Medications Used in the Treatment and Prevention of Osteoporosis

Abbreviation: CrCl, creatinine clearance.

CONCLUSION

With a growing aging population, the prevalence of osteoporosis is expected to increase. By 2025, experts estimate that there will be 2 million fractures yearly, costing the United States upwards of $25 billion.^34,35 This estimate does not include the cost of lost productivity or disability, which will likely cost billions more.^34,35 Understanding risk factors and eliminating medications known to cause decreased BMD are vital. Obtaining a BMD measurement is the rate-limiting step for treatment initiation. Without an appropriate diagnosis, treatment is unlikely. As providers, it us our responsibility to maintain a high level of suspicion of osteoporosis in the elderly and promptly diagnose and treat them.

Fragility fractures are not only traumatizing for patients; they are also associated with significantly increased mortality. A study by Gosch and colleagues found that 70.6% of patients died during the normal follow-up period, and 29.4% of patients died within the first year of suffering a fracture.³ Also, the mean life expectancy post-fragility fracture was only 527 days.³ Diagnosis and treatment of osteoporosis is imperative to prevent fragility fractures before they occur.

RISK FACTORS AND CAUSES

The incidence of fragility fractures increases in patients with comorbidities such as thyroid disease, diabetes, hypertension, and heart disease.⁴ Hyperthyroidism and treated hypothyroidism cause an imbalance between osteoblast and osteoclast activity, resulting in osteoporosis.⁵ A thyroid-stimulating hormone level < 0.1 increases the risk of vertebral and non-vertebral fractures by a factor of 4.5 and 3.2 mIU/L respectively.⁴ Patients with diabetes also have an increased risk of fragility fractures, which is due to impaired healing capabilities, especially that of bone healing. Approximately 2 million people are affected by type 1 diabetes in the United States, and 20% of those patients will develop osteoporosis.⁶

Hypertension and osteoporosis are 2 diseases that occur often in the elderly. Common etiological factors believed to cause both hypertension and osteoporosis are low calcium intake, high consumption of salt, and vitamin D and vitamin K deficiency. Also, hypertension treated with loop diuretics has been found to cause negative effects on bone and increase the risk of osteoporosis.⁷ The only antihypertensive medications that preserve bone mineral density (BMD) and reduce fracture risk are thiazide diuretics.⁷ Lastly, an association between coronary artery disease and osteoporosis has been hypothesized. The link is not completely understood, but it is believed that oxidative stress and inflammation are the culprits in both diseases.⁸ In contrast to previous hypotheses, Sosa and colleagues found an independent association between beta blockers and fragility fractures.⁹ The idea that beta blockers and fragility fractures are linked is still controversial and needs more study. Unlike beta blockers, statins provide a protective effect on bone. They increase BMD and reduce fracture risk by inhibiting osteoclastogenesis.¹⁰

In addition to loop diuretics and beta blockers, inhaled glucocorticoids, oral glucocorticoids, proton pump inhibitors (PPIs), H₂ receptor antagonists, and anticonvulsants decrease bone density and increase the incidence of fragility fractures.¹¹ Chronic glucocorticoid therapy is the most common cause of secondary osteoporosis. Osteoblasts and osteocytes undergo apoptosis in the presence of glucocorticoids.¹² Patients on glucocorticoid therapy have an increased risk of fracture, even with higher BMD values.¹³ Bone changes that occur while a patient is taking glucocorticoids may not be detected during BMD testing. Therefore, a high level of suspicion of osteoporosis in patients on long-term glucocorticoids is imperative.

Proton pump inhibitors are among the most prescribed medications in the world; they reduce bone resorption, increasing the risk of fracture.¹⁴ Proton pump inhibitors and H₂ receptor antagonists are hypothesized to cause malabsorption of calcium and indirectly cause osteoporosis. The risk of osteoporosis increases with the length of PPI treatment.¹⁵ However, exposure lasting <7 years does not increase the risk of fracture.¹⁶ It is recommended that patients on long-term PPIs be referred for BMD testing.

An association between anticonvulsants and osteoporosis has been found in observational studies. The mechanism of this association is not yet fully understood, but it is believed that exacerbation of vitamin D deficiency leads to increased bone metabolism.¹⁷ Gastrointestinal (GI) calcium absorption also decreases with anticonvulsant use. Prolonged antiepileptic therapy and high-dose therapy rapidly decrease BMD. Primidone, carbamazepine, phenobarbital, and phenytoin are the drugs most often associated with decreased BMD. Osteoporosis and fragility fracture in these patients can be prevented with calcium, vitamin D, and the bisphosphonate risedronate. These medications have been shown to improve BMD by 69%.¹⁸

Continue to: DIAGNOSIS...

Osteoporosis is diagnosed by the presence of a fragility fracture or by dual-energy x-ray absorptiometry (DXA) in the absence of a fragility fracture.¹⁹ Measurements of the femoral neck by DXA are used to diagnose osteoporosis, although DXA can also be used to measure the bone density of the spine and peripheral skeleton.²⁰

The World Health Organization developed a set of T score criteria to diagnose osteoporosis in postmenopausal women (Table 1). A T score >-1 is normal, <-1 but >-2.5 signifies osteopenia, <-2.5 is osteoporosis, and <-2.5 with fragility fracture is severe osteoporosis.¹⁹ The Z score, not the T score, should be used to assess osteoporosis in premenopausal women, men <50 years, and children (Table 2). The Z score is calculated by comparing the patient’s BMD with the mean BMD of their peers of a similar age, race, and gender.¹⁹ Z scores <-2.0 indicate low BMD for chronological age. A Z score > -2.0 is considered within the expected range for age.²⁰ Bone mineral density testing is the rate- limiting step to starting osteoporosis treatment.²¹ Without testing, treatment of osteoporosis is very unlikely.

Table 1. T Score Criteria

Table 2. Z Score Criteria

The World Health Organization also developed a tool to predict fracture risk. The Fracture Risk Assessment Tool uses fracture history in addition to other risk factors to predict a patient’s 10-year risk of major fracture.²² Risk factors used to assess fracture risk include age, sex, weight, height, previous fracture, parental hip fracture history, current smoker, glucocorticoid use, rheumatoid arthritis, secondary osteoporosis, excessive alcohol use, and femoral neck BMD.

In 2011, the United States Preventive Services Task Force (USPSTF) recommended that all women ≥65 years should be screened for osteoporosis by DXA. Women <65 years with a 10-year fracture risk =/> than that of a 65-year-old white woman should also be screened for osteoporosis. These recommendations are different for men. It was concluded that the evidence was insufficient to support osteoporosis screening in men.²³ As of April 2017, Centers for Medicare and Medicaid Services current reimbursement rates for DXA scans are, on average, $123.10 in the hospital setting and $41.63 in the office setting. The axial DXA CPT code is 77080.

Continue to: TREATMENT...

NONPHARMACOLOGIC

Patients with mild osteoporosis may be treated first non-pharmacologically. Lifestyle changes such as calcium and vitamin D supplementation, exercise, and smoking cessation are non-pharmacologic treatment options. Calcium carbonate and calcium citrate are common supplements. Calcium carbonate is 40% elemental calcium, whereas calcium citrate supplements are only 21% elemental calcium. Calcium supplements are best absorbed when taken with food.²⁴ The recommended daily total calcium intake is 1200 mg.²⁵ Only 500 to 600 milligrams of calcium can be absorbed by the GI tract at a time. Therefore, calcium supplements should be taken at least 4 to 5 hours apart.²⁴Patients should also be counseled that calcium supplements may cause GI side effects such as bloating and constipation. To reduce side effects, patients can slowly increase the dose of calcium to a therapeutic level.

Vitamin D supplementation works best in conjunction with calcium supplementation. Vitamin D functions to regulate calcium absorption in the intestine and stimulate bone resorption and maintain the serum calcium concentration. The National Osteoporosis Foundation recommends 800 to 1000 international units of vitamin D daily.²⁴ Lifestyle changes may be sufficient to stop the progression of osteoporosis in its early stages. Once osteoporosis becomes severe enough, pharmacotherapy is needed to stop further bone destruction and improve BMD.

PHARMACOLOGIC

After an initial fragility fracture, the risk of additional ones increases significantly, making treatment of osteoporosis essential. The National Osteoporosis Foundation recommends treating osteoporosis with pharmacotherapy in patients with a high risk of fracture (T score <-2.5) or history of fragility fracture.²⁶ Bisphosphonates inhibit bone resorption and are considered the first-line therapy for postmenopausal women with osteoporosis. A common side effect of oral bisphosphonates is GI toxicity. Patients are advised to avoid lying down for at least 30 minutes after medication administration to avoid esophageal irritation. Oral bisphosphonates should also be taken in the morning on an empty stomach with at least 8 ounces of water. Recurrent bisphosphonate use should be avoided in patients with chronic kidney disease. Oral alendronate and risedronate are typically discontinued after 5 years of use.²⁷ Long-term bisphosphonate use may cause an increased risk of fragility fracture due to oversuppression of bone turnover. To avoid this risk, bisphosphonate “drug holidays” are an option. Bisphosphonates accumulate over time, creating reservoirs. Even after therapy is stopped, patients continue to have therapeutic effects for 2 to 5 years.²⁸

Bisphosphonates are available in both oral and intravenous forms. Alendronate is available in doses of 10 mg and 70 mg for daily and weekly administration, respectively. Both are available in tablet form, but the 70 mg weekly dose is also available in a dissolvable formulation. Alendronate is available in a reduced dose for osteoporosis prevention. Alendronate dosing for osteoporosis prevention is 5 mg daily or 35 mg weekly. Risedronate is dosed as 5 mg daily, 35 mg weekly, or 150 mg monthly. Intravenous bisphosphonates are indicated when oral bisphosphonates are not tolerated, only after vitamin D has been assessed and is within the normal range. Zoledronic acid is administered as a 15-minute infusion once a year.

Teriparatide (Forteo; PTH-1-34) is available for glucocorticoid-induced osteoporosis, postmenopausal women, and men with severe osteoporosis. It is indicated for patients in whom bisphosphonate treatment has failed or those who do not tolerate bisphosphonates. Teriparatide is a synthetic parathyroid hormone (PTH) that acts as an anabolic agent, stimulating bone formation, maturation, and remodeling.²⁹ In addition to its application as a bone-building hormone, teriparatide has gained popularity for various off-label uses. These include accelerated osteosynthesis, stress fracture healing, and in the nonoperative treatment of osteoarthritis.²⁹ Parathyroid hormone has been shown to stimulate the maturation, proliferation, and maintenance of osteoblast progenitor cells. More recently, PTH has been shown to regulate chondrocyte signaling, as well as differentiation and maturation. Further study on the chondroregenerative potential of PTH has demonstrated its efficacy as a novel disease-modifying agent in the treatment of osteoarthritis.²⁹ Teriparatide is administered as a daily subcutaneous injection. The United States dosing is 600 mcg/2.4 mL. Adverse effects such as orthostatic hypotension and osteosarcoma may occur. BMD testing should be performed 1 to 2 years after initiation of teriparatide and every 2 years thereafter.²⁶

Abaloparatide (Tymlos), a human parathyroid hormone, is another treatment option for postmenopausal women at risk of osteoporotic fracture. In a study comparing the efficacy of abaloparatide and teriparatide, treatment with abaloparatide was found to induce higher BMD levels in a time frame of 12 months. The BMD differences could be attributed to many factors, such as an enhanced net anabolic effect or a reduced osteoblast expression. Furthermore, the risk of developing new vertebral and nonvertebral fractures decreased in the abaloparatide group compared with the placebo group over a period of 18 months.³⁰

Continue to: The recommended daily dose for abaloparatide...

The recommended daily dose for abaloparatide is 80 mcg via subcutaneous injection with calcium and vitamin D supplements.³¹ Adverse reactions were consistent between abaloparatide and teriparatide, and included hypercalcemia, hypercalciuria, and orthostatic hypotension.³⁰ The use of parathyroid analogs for >2 years is not recommended due to the risk of osteosarcoma.

Denosumab (Prolia) is a monoclonal antibody that stops osteoclastogenesis by blocking the binding of RANKL to RANK.³¹ It is indicated for patients intolerant to bisphosphonates or with impaired kidney function. Prolia is administered subcutaneously in 60 mg doses every 6 months in men and postmenopausal women with osteoporosis. Prolia is contraindicated in patients with hypersensitivity to any component of the medication, pregnancy, and hypocalcemia.

Selective estrogen receptor modulators (SERMs), such as raloxifene and tamoxifen, can treat osteoporosis effectively in postmenopausal women. Raloxifene is considered the SERM of choice due to the availability of more robust safety and efficacy data. Raloxifene increases BMD while decreasing bone resorption and bone turnover.³² It is also used to reduce breast cancer risk; however, it increases the risk of thromboembolic events and hot flashes. Tamoxifen is not typically used to treat osteoporosis, but women treated for breast cancer with tamoxifen receive some bone protection.

Lastly, calcitonin and strontium ranelate are also options to treat osteoporosis. However, both calcitonin and strontium ranelate have weak effects on BMD. Calcitonin only transiently inhibits osteoclast activity.³³ Therefore, medications like bisphosphonates, teriparatide, denosumab, and SERMs are preferred.

A summary of medications used to treat osteoporosis can be found in Table 3.

Table 3. Overview of Common Medications Used in the Treatment and Prevention of Osteoporosis

Abbreviation: CrCl, creatinine clearance.

CONCLUSION

With a growing aging population, the prevalence of osteoporosis is expected to increase. By 2025, experts estimate that there will be 2 million fractures yearly, costing the United States upwards of $25 billion.^34,35 This estimate does not include the cost of lost productivity or disability, which will likely cost billions more.^34,35 Understanding risk factors and eliminating medications known to cause decreased BMD are vital. Obtaining a BMD measurement is the rate-limiting step for treatment initiation. Without an appropriate diagnosis, treatment is unlikely. As providers, it us our responsibility to maintain a high level of suspicion of osteoporosis in the elderly and promptly diagnose and treat them.

Fragility fractures are not only traumatizing for patients; they are also associated with significantly increased mortality. A study by Gosch and colleagues found that 70.6% of patients died during the normal follow-up period, and 29.4% of patients died within the first year of suffering a fracture.³ Also, the mean life expectancy post-fragility fracture was only 527 days.³ Diagnosis and treatment of osteoporosis is imperative to prevent fragility fractures before they occur.

RISK FACTORS AND CAUSES

The incidence of fragility fractures increases in patients with comorbidities such as thyroid disease, diabetes, hypertension, and heart disease.⁴ Hyperthyroidism and treated hypothyroidism cause an imbalance between osteoblast and osteoclast activity, resulting in osteoporosis.⁵ A thyroid-stimulating hormone level < 0.1 increases the risk of vertebral and non-vertebral fractures by a factor of 4.5 and 3.2 mIU/L respectively.⁴ Patients with diabetes also have an increased risk of fragility fractures, which is due to impaired healing capabilities, especially that of bone healing. Approximately 2 million people are affected by type 1 diabetes in the United States, and 20% of those patients will develop osteoporosis.⁶

Hypertension and osteoporosis are 2 diseases that occur often in the elderly. Common etiological factors believed to cause both hypertension and osteoporosis are low calcium intake, high consumption of salt, and vitamin D and vitamin K deficiency. Also, hypertension treated with loop diuretics has been found to cause negative effects on bone and increase the risk of osteoporosis.⁷ The only antihypertensive medications that preserve bone mineral density (BMD) and reduce fracture risk are thiazide diuretics.⁷ Lastly, an association between coronary artery disease and osteoporosis has been hypothesized. The link is not completely understood, but it is believed that oxidative stress and inflammation are the culprits in both diseases.⁸ In contrast to previous hypotheses, Sosa and colleagues found an independent association between beta blockers and fragility fractures.⁹ The idea that beta blockers and fragility fractures are linked is still controversial and needs more study. Unlike beta blockers, statins provide a protective effect on bone. They increase BMD and reduce fracture risk by inhibiting osteoclastogenesis.¹⁰

In addition to loop diuretics and beta blockers, inhaled glucocorticoids, oral glucocorticoids, proton pump inhibitors (PPIs), H₂ receptor antagonists, and anticonvulsants decrease bone density and increase the incidence of fragility fractures.¹¹ Chronic glucocorticoid therapy is the most common cause of secondary osteoporosis. Osteoblasts and osteocytes undergo apoptosis in the presence of glucocorticoids.¹² Patients on glucocorticoid therapy have an increased risk of fracture, even with higher BMD values.¹³ Bone changes that occur while a patient is taking glucocorticoids may not be detected during BMD testing. Therefore, a high level of suspicion of osteoporosis in patients on long-term glucocorticoids is imperative.

Proton pump inhibitors are among the most prescribed medications in the world; they reduce bone resorption, increasing the risk of fracture.¹⁴ Proton pump inhibitors and H₂ receptor antagonists are hypothesized to cause malabsorption of calcium and indirectly cause osteoporosis. The risk of osteoporosis increases with the length of PPI treatment.¹⁵ However, exposure lasting <7 years does not increase the risk of fracture.¹⁶ It is recommended that patients on long-term PPIs be referred for BMD testing.

An association between anticonvulsants and osteoporosis has been found in observational studies. The mechanism of this association is not yet fully understood, but it is believed that exacerbation of vitamin D deficiency leads to increased bone metabolism.¹⁷ Gastrointestinal (GI) calcium absorption also decreases with anticonvulsant use. Prolonged antiepileptic therapy and high-dose therapy rapidly decrease BMD. Primidone, carbamazepine, phenobarbital, and phenytoin are the drugs most often associated with decreased BMD. Osteoporosis and fragility fracture in these patients can be prevented with calcium, vitamin D, and the bisphosphonate risedronate. These medications have been shown to improve BMD by 69%.¹⁸

Continue to: DIAGNOSIS...

Osteoporosis is diagnosed by the presence of a fragility fracture or by dual-energy x-ray absorptiometry (DXA) in the absence of a fragility fracture.¹⁹ Measurements of the femoral neck by DXA are used to diagnose osteoporosis, although DXA can also be used to measure the bone density of the spine and peripheral skeleton.²⁰

The World Health Organization developed a set of T score criteria to diagnose osteoporosis in postmenopausal women (Table 1). A T score >-1 is normal, <-1 but >-2.5 signifies osteopenia, <-2.5 is osteoporosis, and <-2.5 with fragility fracture is severe osteoporosis.¹⁹ The Z score, not the T score, should be used to assess osteoporosis in premenopausal women, men <50 years, and children (Table 2). The Z score is calculated by comparing the patient’s BMD with the mean BMD of their peers of a similar age, race, and gender.¹⁹ Z scores <-2.0 indicate low BMD for chronological age. A Z score > -2.0 is considered within the expected range for age.²⁰ Bone mineral density testing is the rate- limiting step to starting osteoporosis treatment.²¹ Without testing, treatment of osteoporosis is very unlikely.

Table 1. T Score Criteria

Table 2. Z Score Criteria

The World Health Organization also developed a tool to predict fracture risk. The Fracture Risk Assessment Tool uses fracture history in addition to other risk factors to predict a patient’s 10-year risk of major fracture.²² Risk factors used to assess fracture risk include age, sex, weight, height, previous fracture, parental hip fracture history, current smoker, glucocorticoid use, rheumatoid arthritis, secondary osteoporosis, excessive alcohol use, and femoral neck BMD.

In 2011, the United States Preventive Services Task Force (USPSTF) recommended that all women ≥65 years should be screened for osteoporosis by DXA. Women <65 years with a 10-year fracture risk =/> than that of a 65-year-old white woman should also be screened for osteoporosis. These recommendations are different for men. It was concluded that the evidence was insufficient to support osteoporosis screening in men.²³ As of April 2017, Centers for Medicare and Medicaid Services current reimbursement rates for DXA scans are, on average, $123.10 in the hospital setting and $41.63 in the office setting. The axial DXA CPT code is 77080.

Continue to: TREATMENT...

NONPHARMACOLOGIC

Patients with mild osteoporosis may be treated first non-pharmacologically. Lifestyle changes such as calcium and vitamin D supplementation, exercise, and smoking cessation are non-pharmacologic treatment options. Calcium carbonate and calcium citrate are common supplements. Calcium carbonate is 40% elemental calcium, whereas calcium citrate supplements are only 21% elemental calcium. Calcium supplements are best absorbed when taken with food.²⁴ The recommended daily total calcium intake is 1200 mg.²⁵ Only 500 to 600 milligrams of calcium can be absorbed by the GI tract at a time. Therefore, calcium supplements should be taken at least 4 to 5 hours apart.²⁴Patients should also be counseled that calcium supplements may cause GI side effects such as bloating and constipation. To reduce side effects, patients can slowly increase the dose of calcium to a therapeutic level.

Vitamin D supplementation works best in conjunction with calcium supplementation. Vitamin D functions to regulate calcium absorption in the intestine and stimulate bone resorption and maintain the serum calcium concentration. The National Osteoporosis Foundation recommends 800 to 1000 international units of vitamin D daily.²⁴ Lifestyle changes may be sufficient to stop the progression of osteoporosis in its early stages. Once osteoporosis becomes severe enough, pharmacotherapy is needed to stop further bone destruction and improve BMD.

PHARMACOLOGIC

After an initial fragility fracture, the risk of additional ones increases significantly, making treatment of osteoporosis essential. The National Osteoporosis Foundation recommends treating osteoporosis with pharmacotherapy in patients with a high risk of fracture (T score <-2.5) or history of fragility fracture.²⁶ Bisphosphonates inhibit bone resorption and are considered the first-line therapy for postmenopausal women with osteoporosis. A common side effect of oral bisphosphonates is GI toxicity. Patients are advised to avoid lying down for at least 30 minutes after medication administration to avoid esophageal irritation. Oral bisphosphonates should also be taken in the morning on an empty stomach with at least 8 ounces of water. Recurrent bisphosphonate use should be avoided in patients with chronic kidney disease. Oral alendronate and risedronate are typically discontinued after 5 years of use.²⁷ Long-term bisphosphonate use may cause an increased risk of fragility fracture due to oversuppression of bone turnover. To avoid this risk, bisphosphonate “drug holidays” are an option. Bisphosphonates accumulate over time, creating reservoirs. Even after therapy is stopped, patients continue to have therapeutic effects for 2 to 5 years.²⁸

Bisphosphonates are available in both oral and intravenous forms. Alendronate is available in doses of 10 mg and 70 mg for daily and weekly administration, respectively. Both are available in tablet form, but the 70 mg weekly dose is also available in a dissolvable formulation. Alendronate is available in a reduced dose for osteoporosis prevention. Alendronate dosing for osteoporosis prevention is 5 mg daily or 35 mg weekly. Risedronate is dosed as 5 mg daily, 35 mg weekly, or 150 mg monthly. Intravenous bisphosphonates are indicated when oral bisphosphonates are not tolerated, only after vitamin D has been assessed and is within the normal range. Zoledronic acid is administered as a 15-minute infusion once a year.

Teriparatide (Forteo; PTH-1-34) is available for glucocorticoid-induced osteoporosis, postmenopausal women, and men with severe osteoporosis. It is indicated for patients in whom bisphosphonate treatment has failed or those who do not tolerate bisphosphonates. Teriparatide is a synthetic parathyroid hormone (PTH) that acts as an anabolic agent, stimulating bone formation, maturation, and remodeling.²⁹ In addition to its application as a bone-building hormone, teriparatide has gained popularity for various off-label uses. These include accelerated osteosynthesis, stress fracture healing, and in the nonoperative treatment of osteoarthritis.²⁹ Parathyroid hormone has been shown to stimulate the maturation, proliferation, and maintenance of osteoblast progenitor cells. More recently, PTH has been shown to regulate chondrocyte signaling, as well as differentiation and maturation. Further study on the chondroregenerative potential of PTH has demonstrated its efficacy as a novel disease-modifying agent in the treatment of osteoarthritis.²⁹ Teriparatide is administered as a daily subcutaneous injection. The United States dosing is 600 mcg/2.4 mL. Adverse effects such as orthostatic hypotension and osteosarcoma may occur. BMD testing should be performed 1 to 2 years after initiation of teriparatide and every 2 years thereafter.²⁶

Abaloparatide (Tymlos), a human parathyroid hormone, is another treatment option for postmenopausal women at risk of osteoporotic fracture. In a study comparing the efficacy of abaloparatide and teriparatide, treatment with abaloparatide was found to induce higher BMD levels in a time frame of 12 months. The BMD differences could be attributed to many factors, such as an enhanced net anabolic effect or a reduced osteoblast expression. Furthermore, the risk of developing new vertebral and nonvertebral fractures decreased in the abaloparatide group compared with the placebo group over a period of 18 months.³⁰

Continue to: The recommended daily dose for abaloparatide...

The recommended daily dose for abaloparatide is 80 mcg via subcutaneous injection with calcium and vitamin D supplements.³¹ Adverse reactions were consistent between abaloparatide and teriparatide, and included hypercalcemia, hypercalciuria, and orthostatic hypotension.³⁰ The use of parathyroid analogs for >2 years is not recommended due to the risk of osteosarcoma.

Denosumab (Prolia) is a monoclonal antibody that stops osteoclastogenesis by blocking the binding of RANKL to RANK.³¹ It is indicated for patients intolerant to bisphosphonates or with impaired kidney function. Prolia is administered subcutaneously in 60 mg doses every 6 months in men and postmenopausal women with osteoporosis. Prolia is contraindicated in patients with hypersensitivity to any component of the medication, pregnancy, and hypocalcemia.

Selective estrogen receptor modulators (SERMs), such as raloxifene and tamoxifen, can treat osteoporosis effectively in postmenopausal women. Raloxifene is considered the SERM of choice due to the availability of more robust safety and efficacy data. Raloxifene increases BMD while decreasing bone resorption and bone turnover.³² It is also used to reduce breast cancer risk; however, it increases the risk of thromboembolic events and hot flashes. Tamoxifen is not typically used to treat osteoporosis, but women treated for breast cancer with tamoxifen receive some bone protection.

Lastly, calcitonin and strontium ranelate are also options to treat osteoporosis. However, both calcitonin and strontium ranelate have weak effects on BMD. Calcitonin only transiently inhibits osteoclast activity.³³ Therefore, medications like bisphosphonates, teriparatide, denosumab, and SERMs are preferred.

A summary of medications used to treat osteoporosis can be found in Table 3.

Table 3. Overview of Common Medications Used in the Treatment and Prevention of Osteoporosis

Abbreviation: CrCl, creatinine clearance.

CONCLUSION

With a growing aging population, the prevalence of osteoporosis is expected to increase. By 2025, experts estimate that there will be 2 million fractures yearly, costing the United States upwards of $25 billion.^34,35 This estimate does not include the cost of lost productivity or disability, which will likely cost billions more.^34,35 Understanding risk factors and eliminating medications known to cause decreased BMD are vital. Obtaining a BMD measurement is the rate-limiting step for treatment initiation. Without an appropriate diagnosis, treatment is unlikely. As providers, it us our responsibility to maintain a high level of suspicion of osteoporosis in the elderly and promptly diagnose and treat them.

Prior reviews have demonstrated a strong link between surgeon and hospital TSA volume and outcomes of the procedure.^8–10 Somerson and colleagues¹¹ investigated fellowship training among surgeons performing TSA in 2012 and found that only 28% had completed a shoulder and elbow fellowship. In addition to prior analyses2, ¹², Somerson and colleagues confirmed a persistent geographic variation in utilization of TSA.¹¹ In conjunction with the evolution of shoulder arthroplasty, dedicated shoulder and elbow fellowship training has expanded. With a shift toward specialization in care, nearly 90% of orthopedic surgery residents plan to pursue shoulder and elbow fellowships, comprising 4.6% of (42/897) of available positions.¹³  

What remains unknown is the specialization of surgeons performing TSA, the regularity of their arthroplasty volume, and trends in TSA specialization over time. Therefore, this study aims to (a) identify surgeons performing shoulder arthroplasty and cohort changes over time, (b) determine the case profile of surgeons consistently performing shoulder arthroplasty, and (c) establish the characteristics of shoulder arthroplasty surgeons with a specific focus on fellowship training. 

METHODS

Prior to collecting surgeon-specific data, we identified surgeons performing TSA through the Centers for Medicare and Medicaid Services’ public release of “Medicare Provider Utilization and Payment Data: Physician and Other Supplier.”¹⁴ Datasets from 2012, 2013, and 2014 were used to identify all surgeons performing >10 TSAs (Current Procedural Terminology [CPT] Code 23472) during at least 1 of those years. This dataset provides the name, identification number, address, and all billing (by volume) for each unique CPT code submitted ≥10 times in a calendar year.

Once the cohort of surgeons had been generated, the number of surgeons consistently performing TSA year-over-year was determined. This allowed for an analysis of the consistency with which surgeons are performing moderate- to high-volume TSA. To form a case profile of surgeons performing TSA and observe how this shifted over time, a count and a description of each CPT code submitted by each surgeon was identified. To maintain patient privacy, only those claims made >10 times are reported for a provider (both physicians and physician-extenders are included in this dataset). First, all CPT codes were reviewed and tagged as surgical or non-surgical events. Then, every procedural CPT code identified was reviewed and categorized based upon anatomic location and procedure (eg, total knee arthroplasty [TKA]). It is important to note that all claims in this dataset are limited to those patients participating in Medicare’s fee-for-service program. 

Specialization was defined as the number of categorized procedures as a percentage of all procedures performed on Medicare patients. The trends for national, regional, and individual specialization of TSA, arthroplasty (major joint), and shoulder procedures were determined.

Continue to: To investigate the characteristics of surgeons...

To investigate the characteristics of surgeons consistently performing TSA, all surgeons performing a minimum of 11 TSA in Medicare fee-for-service beneficiaries in all years between 2012 and 2014 were identified. Such surgeons were defined as consistent TSA surgeons. Investigation of this cohort included a web-based search of their self-reported post-graduate fellowship training and year of graduation from medical school. Using these data, the percentage of surgeons performing TSA who underwent formal shoulder and elbow training was determined. In addition, the impact of fellowship training on shoulder specialization and practice location was determined. Surgeons who had completed multiple fellowships were categorized under all of them. As such, there may be some duplication of surgeons in the comparisons. In addition, other potential characteristics of shoulder and elbow fellowship-trained surgeons were investigated: number of regional shoulder surgeons, urban area, total number of Medicare beneficiaries, average reimbursement for TSA, ethnicity of Medicare beneficiaries, and percentage of Medicare patients eligible for Medicaid. Geographic regions were defined by the Dartmouth Atlas and assigned by hospital referral region.¹⁵ These defined regions were used to assess the beneficiaries (number and characteristics) that individual surgeons were likely serving. The United States Census Bureau characterization of zip code-based regions as urban areas (population >50,000), urban clusters (2500 to 50,000), and rural region (<2500) was used to categorize practice location.¹⁶ 

Descriptive statistics were used initially to report these findings. To analyze predictors of utilization and specialization, comparative statistics were undertaken. For comparison of binomial variables between groups, a χ2analysis was utilized. For continuous variables, data normality was assessed. A skewness and kurtosis <2 and 12, respectively, was considered to represent parametric data. For parametric data, the mean was reported; conversely, the median is reported for non-parametric data. To assess continuous variables between groups, a t test or a Wilcoxon rank-sum test was used for parametric and non-parametric distributions, respectively. 

RESULTS

Between 2012 and 2014, 1374 surgeons (39 females [2.8%]) performed >10 TSA in Medicare patients in at least 1 year, for a combined total of 71,973 TSAs (Table 1). In 2012, only 834 surgeons (13 females [1.6%]) performed a minimum of 10 TSA in Medicare patients (21,137 arthroplasties; 25.3 per surgeon). This increased to 1078 surgeons (33 females [3.1%]; P = .04) performing 26,865 TSA (24.92 per surgeon) in 2014. Utilization of non-physician assistants in TSA also increased significantly over this period, with 307 assisting in 6885 TSAs (22.4 per provider) in 2012 and 465 assisting 10,433 TSA (22.4 per provider) in 2014. When all procedures were considered, including those performed at outpatient visits, 1319 physicians (95.9% of cohort) were active in 2012—providing either surgical procedures or outpatient consults to the Medicare population. Yet, only 63.2% performed >10 TSA in Medicare patients. The number of active surgeons performing TSA increased to 79.6% (1078/1353) in 2014 (P < .001). 

Table 1. Trends in the Number of Providers Performing TSA between 2012 and 2014*

* Included are the number of arthroplasties and total procedures over time among this cohort. The number of active providers, determined by billing Medicare for office or surgical procedures within that year, is reported.

Abbreviation: TSA, total shoulder arthroplasty.  

In addition to TSA, this cohort of surgeons submitted 240 unique CPT codes with case volumes >10 annually over the 3-year study period. Of these, 80.2% (1102/1374) of surgeons performed non-arthroplasty shoulder procedures on Medicare patients, for a combined total of 202,335 procedures over the 3-year study period (Table 2). A significant proportion of these procedures were arthroscopic debridement (60,014 procedures performed by 908 surgeons) and arthroscopic rotator cuff repair (47,089 procedures performed by 809 surgeons). Just under half (49.1%; 674/1374) of surgeons performing TSA also performed TKA during this period (77,873 arthroplasties). Fewer surgeons (27.8%; 382/1374) performed total hip arthroplasty during this period (27,322 arthroplasties). Other procedure types that this group of surgeons routinely performed on Medicare patients were repairs of traumatic injuries (29.8%), non-arthroplasty knee surgeries (27.2%), elbow procedures (19.6%), and hand surgery (15.4%). By case load, non-arthroplasty shoulder procedures consisted of 43% of Medicare volume over the study period (Figure 1). Between 2012 and 2014, the average proportion of Medicare cases that were shoulder arthroplasties increased from 13.8% (21,137/152,862) to 16.7% (26,865/160,851; P = .001). Shoulder arthroplasty constituted 100% of the Medicare surgical case volume for 67 (4.9%; 67/1374) of the surgeons.    

Table 2. Case Volumes over Time with All Procedures Categorized by Anatomic Region and Arthroplasty vs Non-arthroplasty*

* Procedures of interest with high case volumes are reported individually.

Only 44.3% (609/1374) of surgeons performed TSA in a minimum of 11 Medicare patients in all 3 years of the study period (consistent providers of TSA), providing a total of 55,538 TSA (77.2%; 55,538/71,973). When fellowship training was evaluated, 191 (31.4%; 191/609) of these surgeons were shoulder and elbow fellowship trained (21,444 TSA; 38.6%; Table 3). More than one-third (36.6%; 223/609) had completed a sports surgery fellowship (18,899 TSA; 34.0%). Surgeons trained in hand surgery (12.5%; 76/609) and adult reconstruction (5.3%; 32/610) also made contributions to meeting the TSA demand with 6971 (12.6%) and 2485 (4.5%) TSA, respectively. One-fifth of this cohort (18.1%; 110/609) had unknown fellowship training: they either reported no fellowship (13.6%; 83/609) or did not specify the type of training (4.4%; 27/609). Shoulder and elbow fellowship-trained surgeons performed more TSA (median: 89.0 TSA per surgeon between 2012 and 2014) than surgeons without shoulder and elbow fellowship training (median: 67.0 TSA per surgeon; P < 0.001). More than one-third (37%) of shoulder and elbow fellowship-trained surgeons’ surgical case volume was comprised of TSA, with an additional 35% from non-arthroplasty shoulder procedures (Figure 2). In order for the current supply of shoulder and elbow fellowship-trained surgeons to meet the Medicare TSA demand, each fellowship graduate would have to perform 140.6 TSA in Medicare patients annually. Shoulder and elbow fellowship-trained surgeons were more likely to practice in referral regions with an increased Medicare population (P < .001), an increased number of surgeons performing TSA (P < .001), and a higher proportion of Medicaid-eligible patients (P = .01; Table 4). Shoulder and elbow fellowship-trained surgeons (18.7 years post-medical school graduation) were also earlier in their careers than other consistent TSA surgeons (23.1 years post-graduation; P < .001). 

Table 3. A Representation of Fellowships Among TSA Surgeons and Their Shoulder Arthroplasty Case Load*

* Not all fellowships (eg, oncology) included due to small numbers. Also, many surgeons performed multiple fellowships.

Abbreviations: SA, shoulder arthroplasty; TSA, total shoulder arthroplasty.

Table 4. Breakdown of Geographic Characteristics of Orthopedic Surgeons Consistently

Performing TSA between 2012 and 2014 Stratified by Fellowship Training

Abbreviations: HRR, hospital referral region; TSA, total shoulder arthroplasty.

Continue to: DISCUSSION...

DISCUSSION

Utilization of TSA has continued to rise; however, access to this cost-effective procedure was recently demonstrated to be limited.¹¹ In a separate analysis, we established the continued rise in use of TSA in the Medicare population, coupled with an increase in the number of surgeons routinely performing TSA.² Multiple analyses have demonstrated the importance of high-volume surgeons and hospitals familiar with the intricacies of shoulder arthroplasty concepts in minimizing complications, improving the quality and decreasing the cost of TSA.^6,10,17 Specifically, Singh and colleagues18 demonstrated from a multi-center registry that surgeons and hospitals with greater shoulder arthroplasty volumes had decreased intra-operative blood loss, operative time, and hospital length of stay. As the demand for TSA, both anatomic and reverse, continues to rise, it is imperative that the healthcare delivery system is optimized to provide the best possible care. Before we can determine whether specialized training in shoulder arthroplasty influences surgical outcomes, characteristics and training of surgeons performing TSA should be described. 

The number of surgeons performing >10 TSA in the Medicare population rose significantly between 2012 and 2014 (29.3%). However, the number of TSAs per surgeon over this time period remained consistent (approximately 25 per surgeon). Furthermore, the increase in the number of surgeons performing a reportable volume of TSA by 2014 was from the addition of already active surgeons (ie, the growth in TSA was not from the addition of newly trained arthroplasty surgeons but originated from the existing orthopedic surgeon workforce). In a recently published analysis, Somerson and colleagues, ¹¹ using this same dataset, demonstrated persistent limitations in access to high-volume TSA surgeons. In a more recent analysis, we showed that while still lacking for some patients, access to a high-volume TSA surgeon has improved significantly over the past 3 years, with 96.9% of the United States population residing within 200 kilometers of a high-volume TSA surgeon (>20 Medicare cases).² This analysis validates those findings, with the caveat that the average annual volume per surgeon is not increasing. What remains unknown, due to limitations of this dataset, is how many surgeons are not identified because they are performing ≤10 TSA each year or are performing TSA in non-Medicare patients. 

With the specialization of healthcare delivery, specifically in orthopedics, it is imperative that mechanisms for providing specialty-focused care be established. However, the proportion of their practice that surgeons dedicate to TSA was unknown. This study demonstrates that this proportion is increasing. Including non-arthroplasty procedures, more than half (58%) of the procedures performed by this surgeon cohort were shoulder-specific. Furthermore, this analysis demonstrates that surgeons performing TSA have significant case diversity, including nearly half of the cohort performing TKA. Repeated evidence has demonstrated the effect of case volume on improved outcomes following orthopedic procedures.^8,19–21 The pre-existing location-based model for delivering orthopedic care supports case diversity; however, this model continues to be challenged with high-volume centers of excellence and patient travel.^22–24 Hip and knee arthroplasty experienced a similar surge in demand, with a subsequent shift in care to high-volume surgeons and centers.²⁵ Shoulder and elbow fellowship-trained surgeons would need to nearly quadruple their current Medicare TSA volume to meet the entire current demand for TSA in the Medicare population (and this does not account for TSA performed by very low-volume surgeons not included in this cohort). With increased utilization of TSA, policymakers and the orthopedic community must determine the structure of delivery (centers of excellence or medium-volume disseminated throughout the country) that is optimal. 

For those surgeons consistently performing TSA over the study period, fellowship training was diverse. While the current focus in orthopedics is on case volume, research in other specialties, namely general surgery, has provided repeated evidence that surgical specialization (more so than high case volume) provides improved outcomes.^26–29Furthermore, Leopold and colleagues³⁰ demonstrated an inverse relationship between competency in performing a procedure and confidence in one’s ability to do so. In their study, educational intervention provided improved competency in the procedure. Less than one-third (29.8%) of TSA in this cohort were performed by a shoulder and elbow fellowship-trained surgeon consistently performing this procedure. Approximately another quarter (26.2%) were performed by consistent TSA surgeons trained in sports surgery. Meanwhile, 34.6% of TSA in this study cohort were performed by a surgeon who did not consistently meet the minimum threshold in all study years (16,435 TSA; 22.8%) or by a surgeon performing TSA without fellowship training (8,489 TSA; 11.8%). There has been a trend toward orthopedic subspecialty training with an increased demand for fellowship-trained surgeons.³¹ Despite this and the complexities of TSA, many continue to be performed by surgeons with an inconsistent volume and those without arthroplasty-specific fellowship training. The available evidence supports a push toward the fellowship-trained, high-volume TSA surgeon in providing reproducible high-quality shoulder arthroplasty care. For now, that surgeon is more likely to be earlier in his/her career and reside in large, referral-based centers surrounded by other surgeons performing TSA.

These findings must be considered in the light of the study limitations. First, this is a large publicly available database. While this type of database provides a unique opportunity to assess the geographic distributions and characteristics of orthopedic surgeons, specifically those performing TSA, it completely prevents any assessment of the relationship between these findings and quality. As such, while the reader may generate hypotheses regarding the implications of our findings on the quality of TSA delivery, the true effects cannot be determined. In the same vein, for the purpose of privacy, surgeons performing ≤10 TSA were not included in this dataset. This limitation prevents the identification of low-volume TSA surgeons. Also, it is likely that the observed increase in surgeons over time is likely a reflection of small increases in volume for surgeons already performing TSA. Lastly, a web-based search was undertaken to identify surgeons’ self-reported fellowship training. The results of this web-based search could not be validated, and it is possible that fellowship training, or the lack thereof, was mischaracterized and simply not obtainable through a web-based search. Furthermore, it is not possible to fully assess the extent of high-quality TSA training in these various fellowships.

CONCLUSION

In just the past decade, the utilization of TSA in the Medicare population has increased significantly. However, this increase was not achieved by the addition of highly specialized, high-volume surgeons but by the addition of many surgeons performing lower numbers of TSA surgeries. Furthermore, for those performing this cost-effective procedure, TSA constitutes a relatively small proportion of the surgeries they perform. Shoulder and elbow fellowship-trained surgeons currently account for a low percentage of the overall number of surgeons performing TSA. The implications of these findings must be considered and investigated.

Prior reviews have demonstrated a strong link between surgeon and hospital TSA volume and outcomes of the procedure.^8–10 Somerson and colleagues¹¹ investigated fellowship training among surgeons performing TSA in 2012 and found that only 28% had completed a shoulder and elbow fellowship. In addition to prior analyses2, ¹², Somerson and colleagues confirmed a persistent geographic variation in utilization of TSA.¹¹ In conjunction with the evolution of shoulder arthroplasty, dedicated shoulder and elbow fellowship training has expanded. With a shift toward specialization in care, nearly 90% of orthopedic surgery residents plan to pursue shoulder and elbow fellowships, comprising 4.6% of (42/897) of available positions.¹³  

What remains unknown is the specialization of surgeons performing TSA, the regularity of their arthroplasty volume, and trends in TSA specialization over time. Therefore, this study aims to (a) identify surgeons performing shoulder arthroplasty and cohort changes over time, (b) determine the case profile of surgeons consistently performing shoulder arthroplasty, and (c) establish the characteristics of shoulder arthroplasty surgeons with a specific focus on fellowship training. 

METHODS

Prior to collecting surgeon-specific data, we identified surgeons performing TSA through the Centers for Medicare and Medicaid Services’ public release of “Medicare Provider Utilization and Payment Data: Physician and Other Supplier.”¹⁴ Datasets from 2012, 2013, and 2014 were used to identify all surgeons performing >10 TSAs (Current Procedural Terminology [CPT] Code 23472) during at least 1 of those years. This dataset provides the name, identification number, address, and all billing (by volume) for each unique CPT code submitted ≥10 times in a calendar year.

Once the cohort of surgeons had been generated, the number of surgeons consistently performing TSA year-over-year was determined. This allowed for an analysis of the consistency with which surgeons are performing moderate- to high-volume TSA. To form a case profile of surgeons performing TSA and observe how this shifted over time, a count and a description of each CPT code submitted by each surgeon was identified. To maintain patient privacy, only those claims made >10 times are reported for a provider (both physicians and physician-extenders are included in this dataset). First, all CPT codes were reviewed and tagged as surgical or non-surgical events. Then, every procedural CPT code identified was reviewed and categorized based upon anatomic location and procedure (eg, total knee arthroplasty [TKA]). It is important to note that all claims in this dataset are limited to those patients participating in Medicare’s fee-for-service program. 

Specialization was defined as the number of categorized procedures as a percentage of all procedures performed on Medicare patients. The trends for national, regional, and individual specialization of TSA, arthroplasty (major joint), and shoulder procedures were determined.

Continue to: To investigate the characteristics of surgeons...

To investigate the characteristics of surgeons consistently performing TSA, all surgeons performing a minimum of 11 TSA in Medicare fee-for-service beneficiaries in all years between 2012 and 2014 were identified. Such surgeons were defined as consistent TSA surgeons. Investigation of this cohort included a web-based search of their self-reported post-graduate fellowship training and year of graduation from medical school. Using these data, the percentage of surgeons performing TSA who underwent formal shoulder and elbow training was determined. In addition, the impact of fellowship training on shoulder specialization and practice location was determined. Surgeons who had completed multiple fellowships were categorized under all of them. As such, there may be some duplication of surgeons in the comparisons. In addition, other potential characteristics of shoulder and elbow fellowship-trained surgeons were investigated: number of regional shoulder surgeons, urban area, total number of Medicare beneficiaries, average reimbursement for TSA, ethnicity of Medicare beneficiaries, and percentage of Medicare patients eligible for Medicaid. Geographic regions were defined by the Dartmouth Atlas and assigned by hospital referral region.¹⁵ These defined regions were used to assess the beneficiaries (number and characteristics) that individual surgeons were likely serving. The United States Census Bureau characterization of zip code-based regions as urban areas (population >50,000), urban clusters (2500 to 50,000), and rural region (<2500) was used to categorize practice location.¹⁶ 

Descriptive statistics were used initially to report these findings. To analyze predictors of utilization and specialization, comparative statistics were undertaken. For comparison of binomial variables between groups, a χ2analysis was utilized. For continuous variables, data normality was assessed. A skewness and kurtosis <2 and 12, respectively, was considered to represent parametric data. For parametric data, the mean was reported; conversely, the median is reported for non-parametric data. To assess continuous variables between groups, a t test or a Wilcoxon rank-sum test was used for parametric and non-parametric distributions, respectively. 

RESULTS

Between 2012 and 2014, 1374 surgeons (39 females [2.8%]) performed >10 TSA in Medicare patients in at least 1 year, for a combined total of 71,973 TSAs (Table 1). In 2012, only 834 surgeons (13 females [1.6%]) performed a minimum of 10 TSA in Medicare patients (21,137 arthroplasties; 25.3 per surgeon). This increased to 1078 surgeons (33 females [3.1%]; P = .04) performing 26,865 TSA (24.92 per surgeon) in 2014. Utilization of non-physician assistants in TSA also increased significantly over this period, with 307 assisting in 6885 TSAs (22.4 per provider) in 2012 and 465 assisting 10,433 TSA (22.4 per provider) in 2014. When all procedures were considered, including those performed at outpatient visits, 1319 physicians (95.9% of cohort) were active in 2012—providing either surgical procedures or outpatient consults to the Medicare population. Yet, only 63.2% performed >10 TSA in Medicare patients. The number of active surgeons performing TSA increased to 79.6% (1078/1353) in 2014 (P < .001). 

Table 1. Trends in the Number of Providers Performing TSA between 2012 and 2014*

* Included are the number of arthroplasties and total procedures over time among this cohort. The number of active providers, determined by billing Medicare for office or surgical procedures within that year, is reported.

Abbreviation: TSA, total shoulder arthroplasty.  

In addition to TSA, this cohort of surgeons submitted 240 unique CPT codes with case volumes >10 annually over the 3-year study period. Of these, 80.2% (1102/1374) of surgeons performed non-arthroplasty shoulder procedures on Medicare patients, for a combined total of 202,335 procedures over the 3-year study period (Table 2). A significant proportion of these procedures were arthroscopic debridement (60,014 procedures performed by 908 surgeons) and arthroscopic rotator cuff repair (47,089 procedures performed by 809 surgeons). Just under half (49.1%; 674/1374) of surgeons performing TSA also performed TKA during this period (77,873 arthroplasties). Fewer surgeons (27.8%; 382/1374) performed total hip arthroplasty during this period (27,322 arthroplasties). Other procedure types that this group of surgeons routinely performed on Medicare patients were repairs of traumatic injuries (29.8%), non-arthroplasty knee surgeries (27.2%), elbow procedures (19.6%), and hand surgery (15.4%). By case load, non-arthroplasty shoulder procedures consisted of 43% of Medicare volume over the study period (Figure 1). Between 2012 and 2014, the average proportion of Medicare cases that were shoulder arthroplasties increased from 13.8% (21,137/152,862) to 16.7% (26,865/160,851; P = .001). Shoulder arthroplasty constituted 100% of the Medicare surgical case volume for 67 (4.9%; 67/1374) of the surgeons.    

Table 2. Case Volumes over Time with All Procedures Categorized by Anatomic Region and Arthroplasty vs Non-arthroplasty*

* Procedures of interest with high case volumes are reported individually.

Only 44.3% (609/1374) of surgeons performed TSA in a minimum of 11 Medicare patients in all 3 years of the study period (consistent providers of TSA), providing a total of 55,538 TSA (77.2%; 55,538/71,973). When fellowship training was evaluated, 191 (31.4%; 191/609) of these surgeons were shoulder and elbow fellowship trained (21,444 TSA; 38.6%; Table 3). More than one-third (36.6%; 223/609) had completed a sports surgery fellowship (18,899 TSA; 34.0%). Surgeons trained in hand surgery (12.5%; 76/609) and adult reconstruction (5.3%; 32/610) also made contributions to meeting the TSA demand with 6971 (12.6%) and 2485 (4.5%) TSA, respectively. One-fifth of this cohort (18.1%; 110/609) had unknown fellowship training: they either reported no fellowship (13.6%; 83/609) or did not specify the type of training (4.4%; 27/609). Shoulder and elbow fellowship-trained surgeons performed more TSA (median: 89.0 TSA per surgeon between 2012 and 2014) than surgeons without shoulder and elbow fellowship training (median: 67.0 TSA per surgeon; P < 0.001). More than one-third (37%) of shoulder and elbow fellowship-trained surgeons’ surgical case volume was comprised of TSA, with an additional 35% from non-arthroplasty shoulder procedures (Figure 2). In order for the current supply of shoulder and elbow fellowship-trained surgeons to meet the Medicare TSA demand, each fellowship graduate would have to perform 140.6 TSA in Medicare patients annually. Shoulder and elbow fellowship-trained surgeons were more likely to practice in referral regions with an increased Medicare population (P < .001), an increased number of surgeons performing TSA (P < .001), and a higher proportion of Medicaid-eligible patients (P = .01; Table 4). Shoulder and elbow fellowship-trained surgeons (18.7 years post-medical school graduation) were also earlier in their careers than other consistent TSA surgeons (23.1 years post-graduation; P < .001). 

Table 3. A Representation of Fellowships Among TSA Surgeons and Their Shoulder Arthroplasty Case Load*

* Not all fellowships (eg, oncology) included due to small numbers. Also, many surgeons performed multiple fellowships.

Abbreviations: SA, shoulder arthroplasty; TSA, total shoulder arthroplasty.

Table 4. Breakdown of Geographic Characteristics of Orthopedic Surgeons Consistently

Performing TSA between 2012 and 2014 Stratified by Fellowship Training

Abbreviations: HRR, hospital referral region; TSA, total shoulder arthroplasty.

Continue to: DISCUSSION...

DISCUSSION

Utilization of TSA has continued to rise; however, access to this cost-effective procedure was recently demonstrated to be limited.¹¹ In a separate analysis, we established the continued rise in use of TSA in the Medicare population, coupled with an increase in the number of surgeons routinely performing TSA.² Multiple analyses have demonstrated the importance of high-volume surgeons and hospitals familiar with the intricacies of shoulder arthroplasty concepts in minimizing complications, improving the quality and decreasing the cost of TSA.^6,10,17 Specifically, Singh and colleagues18 demonstrated from a multi-center registry that surgeons and hospitals with greater shoulder arthroplasty volumes had decreased intra-operative blood loss, operative time, and hospital length of stay. As the demand for TSA, both anatomic and reverse, continues to rise, it is imperative that the healthcare delivery system is optimized to provide the best possible care. Before we can determine whether specialized training in shoulder arthroplasty influences surgical outcomes, characteristics and training of surgeons performing TSA should be described. 

The number of surgeons performing >10 TSA in the Medicare population rose significantly between 2012 and 2014 (29.3%). However, the number of TSAs per surgeon over this time period remained consistent (approximately 25 per surgeon). Furthermore, the increase in the number of surgeons performing a reportable volume of TSA by 2014 was from the addition of already active surgeons (ie, the growth in TSA was not from the addition of newly trained arthroplasty surgeons but originated from the existing orthopedic surgeon workforce). In a recently published analysis, Somerson and colleagues, ¹¹ using this same dataset, demonstrated persistent limitations in access to high-volume TSA surgeons. In a more recent analysis, we showed that while still lacking for some patients, access to a high-volume TSA surgeon has improved significantly over the past 3 years, with 96.9% of the United States population residing within 200 kilometers of a high-volume TSA surgeon (>20 Medicare cases).² This analysis validates those findings, with the caveat that the average annual volume per surgeon is not increasing. What remains unknown, due to limitations of this dataset, is how many surgeons are not identified because they are performing ≤10 TSA each year or are performing TSA in non-Medicare patients. 

With the specialization of healthcare delivery, specifically in orthopedics, it is imperative that mechanisms for providing specialty-focused care be established. However, the proportion of their practice that surgeons dedicate to TSA was unknown. This study demonstrates that this proportion is increasing. Including non-arthroplasty procedures, more than half (58%) of the procedures performed by this surgeon cohort were shoulder-specific. Furthermore, this analysis demonstrates that surgeons performing TSA have significant case diversity, including nearly half of the cohort performing TKA. Repeated evidence has demonstrated the effect of case volume on improved outcomes following orthopedic procedures.^8,19–21 The pre-existing location-based model for delivering orthopedic care supports case diversity; however, this model continues to be challenged with high-volume centers of excellence and patient travel.^22–24 Hip and knee arthroplasty experienced a similar surge in demand, with a subsequent shift in care to high-volume surgeons and centers.²⁵ Shoulder and elbow fellowship-trained surgeons would need to nearly quadruple their current Medicare TSA volume to meet the entire current demand for TSA in the Medicare population (and this does not account for TSA performed by very low-volume surgeons not included in this cohort). With increased utilization of TSA, policymakers and the orthopedic community must determine the structure of delivery (centers of excellence or medium-volume disseminated throughout the country) that is optimal. 

For those surgeons consistently performing TSA over the study period, fellowship training was diverse. While the current focus in orthopedics is on case volume, research in other specialties, namely general surgery, has provided repeated evidence that surgical specialization (more so than high case volume) provides improved outcomes.^26–29Furthermore, Leopold and colleagues³⁰ demonstrated an inverse relationship between competency in performing a procedure and confidence in one’s ability to do so. In their study, educational intervention provided improved competency in the procedure. Less than one-third (29.8%) of TSA in this cohort were performed by a shoulder and elbow fellowship-trained surgeon consistently performing this procedure. Approximately another quarter (26.2%) were performed by consistent TSA surgeons trained in sports surgery. Meanwhile, 34.6% of TSA in this study cohort were performed by a surgeon who did not consistently meet the minimum threshold in all study years (16,435 TSA; 22.8%) or by a surgeon performing TSA without fellowship training (8,489 TSA; 11.8%). There has been a trend toward orthopedic subspecialty training with an increased demand for fellowship-trained surgeons.³¹ Despite this and the complexities of TSA, many continue to be performed by surgeons with an inconsistent volume and those without arthroplasty-specific fellowship training. The available evidence supports a push toward the fellowship-trained, high-volume TSA surgeon in providing reproducible high-quality shoulder arthroplasty care. For now, that surgeon is more likely to be earlier in his/her career and reside in large, referral-based centers surrounded by other surgeons performing TSA.

These findings must be considered in the light of the study limitations. First, this is a large publicly available database. While this type of database provides a unique opportunity to assess the geographic distributions and characteristics of orthopedic surgeons, specifically those performing TSA, it completely prevents any assessment of the relationship between these findings and quality. As such, while the reader may generate hypotheses regarding the implications of our findings on the quality of TSA delivery, the true effects cannot be determined. In the same vein, for the purpose of privacy, surgeons performing ≤10 TSA were not included in this dataset. This limitation prevents the identification of low-volume TSA surgeons. Also, it is likely that the observed increase in surgeons over time is likely a reflection of small increases in volume for surgeons already performing TSA. Lastly, a web-based search was undertaken to identify surgeons’ self-reported fellowship training. The results of this web-based search could not be validated, and it is possible that fellowship training, or the lack thereof, was mischaracterized and simply not obtainable through a web-based search. Furthermore, it is not possible to fully assess the extent of high-quality TSA training in these various fellowships.

CONCLUSION

In just the past decade, the utilization of TSA in the Medicare population has increased significantly. However, this increase was not achieved by the addition of highly specialized, high-volume surgeons but by the addition of many surgeons performing lower numbers of TSA surgeries. Furthermore, for those performing this cost-effective procedure, TSA constitutes a relatively small proportion of the surgeries they perform. Shoulder and elbow fellowship-trained surgeons currently account for a low percentage of the overall number of surgeons performing TSA. The implications of these findings must be considered and investigated.

Prior reviews have demonstrated a strong link between surgeon and hospital TSA volume and outcomes of the procedure.^8–10 Somerson and colleagues¹¹ investigated fellowship training among surgeons performing TSA in 2012 and found that only 28% had completed a shoulder and elbow fellowship. In addition to prior analyses2, ¹², Somerson and colleagues confirmed a persistent geographic variation in utilization of TSA.¹¹ In conjunction with the evolution of shoulder arthroplasty, dedicated shoulder and elbow fellowship training has expanded. With a shift toward specialization in care, nearly 90% of orthopedic surgery residents plan to pursue shoulder and elbow fellowships, comprising 4.6% of (42/897) of available positions.¹³  

What remains unknown is the specialization of surgeons performing TSA, the regularity of their arthroplasty volume, and trends in TSA specialization over time. Therefore, this study aims to (a) identify surgeons performing shoulder arthroplasty and cohort changes over time, (b) determine the case profile of surgeons consistently performing shoulder arthroplasty, and (c) establish the characteristics of shoulder arthroplasty surgeons with a specific focus on fellowship training. 

METHODS

Prior to collecting surgeon-specific data, we identified surgeons performing TSA through the Centers for Medicare and Medicaid Services’ public release of “Medicare Provider Utilization and Payment Data: Physician and Other Supplier.”¹⁴ Datasets from 2012, 2013, and 2014 were used to identify all surgeons performing >10 TSAs (Current Procedural Terminology [CPT] Code 23472) during at least 1 of those years. This dataset provides the name, identification number, address, and all billing (by volume) for each unique CPT code submitted ≥10 times in a calendar year.

Once the cohort of surgeons had been generated, the number of surgeons consistently performing TSA year-over-year was determined. This allowed for an analysis of the consistency with which surgeons are performing moderate- to high-volume TSA. To form a case profile of surgeons performing TSA and observe how this shifted over time, a count and a description of each CPT code submitted by each surgeon was identified. To maintain patient privacy, only those claims made >10 times are reported for a provider (both physicians and physician-extenders are included in this dataset). First, all CPT codes were reviewed and tagged as surgical or non-surgical events. Then, every procedural CPT code identified was reviewed and categorized based upon anatomic location and procedure (eg, total knee arthroplasty [TKA]). It is important to note that all claims in this dataset are limited to those patients participating in Medicare’s fee-for-service program. 

Specialization was defined as the number of categorized procedures as a percentage of all procedures performed on Medicare patients. The trends for national, regional, and individual specialization of TSA, arthroplasty (major joint), and shoulder procedures were determined.

Continue to: To investigate the characteristics of surgeons...

To investigate the characteristics of surgeons consistently performing TSA, all surgeons performing a minimum of 11 TSA in Medicare fee-for-service beneficiaries in all years between 2012 and 2014 were identified. Such surgeons were defined as consistent TSA surgeons. Investigation of this cohort included a web-based search of their self-reported post-graduate fellowship training and year of graduation from medical school. Using these data, the percentage of surgeons performing TSA who underwent formal shoulder and elbow training was determined. In addition, the impact of fellowship training on shoulder specialization and practice location was determined. Surgeons who had completed multiple fellowships were categorized under all of them. As such, there may be some duplication of surgeons in the comparisons. In addition, other potential characteristics of shoulder and elbow fellowship-trained surgeons were investigated: number of regional shoulder surgeons, urban area, total number of Medicare beneficiaries, average reimbursement for TSA, ethnicity of Medicare beneficiaries, and percentage of Medicare patients eligible for Medicaid. Geographic regions were defined by the Dartmouth Atlas and assigned by hospital referral region.¹⁵ These defined regions were used to assess the beneficiaries (number and characteristics) that individual surgeons were likely serving. The United States Census Bureau characterization of zip code-based regions as urban areas (population >50,000), urban clusters (2500 to 50,000), and rural region (<2500) was used to categorize practice location.¹⁶ 

Descriptive statistics were used initially to report these findings. To analyze predictors of utilization and specialization, comparative statistics were undertaken. For comparison of binomial variables between groups, a χ2analysis was utilized. For continuous variables, data normality was assessed. A skewness and kurtosis <2 and 12, respectively, was considered to represent parametric data. For parametric data, the mean was reported; conversely, the median is reported for non-parametric data. To assess continuous variables between groups, a t test or a Wilcoxon rank-sum test was used for parametric and non-parametric distributions, respectively. 

RESULTS

Between 2012 and 2014, 1374 surgeons (39 females [2.8%]) performed >10 TSA in Medicare patients in at least 1 year, for a combined total of 71,973 TSAs (Table 1). In 2012, only 834 surgeons (13 females [1.6%]) performed a minimum of 10 TSA in Medicare patients (21,137 arthroplasties; 25.3 per surgeon). This increased to 1078 surgeons (33 females [3.1%]; P = .04) performing 26,865 TSA (24.92 per surgeon) in 2014. Utilization of non-physician assistants in TSA also increased significantly over this period, with 307 assisting in 6885 TSAs (22.4 per provider) in 2012 and 465 assisting 10,433 TSA (22.4 per provider) in 2014. When all procedures were considered, including those performed at outpatient visits, 1319 physicians (95.9% of cohort) were active in 2012—providing either surgical procedures or outpatient consults to the Medicare population. Yet, only 63.2% performed >10 TSA in Medicare patients. The number of active surgeons performing TSA increased to 79.6% (1078/1353) in 2014 (P < .001). 

Table 1. Trends in the Number of Providers Performing TSA between 2012 and 2014*

* Included are the number of arthroplasties and total procedures over time among this cohort. The number of active providers, determined by billing Medicare for office or surgical procedures within that year, is reported.

Abbreviation: TSA, total shoulder arthroplasty.  

In addition to TSA, this cohort of surgeons submitted 240 unique CPT codes with case volumes >10 annually over the 3-year study period. Of these, 80.2% (1102/1374) of surgeons performed non-arthroplasty shoulder procedures on Medicare patients, for a combined total of 202,335 procedures over the 3-year study period (Table 2). A significant proportion of these procedures were arthroscopic debridement (60,014 procedures performed by 908 surgeons) and arthroscopic rotator cuff repair (47,089 procedures performed by 809 surgeons). Just under half (49.1%; 674/1374) of surgeons performing TSA also performed TKA during this period (77,873 arthroplasties). Fewer surgeons (27.8%; 382/1374) performed total hip arthroplasty during this period (27,322 arthroplasties). Other procedure types that this group of surgeons routinely performed on Medicare patients were repairs of traumatic injuries (29.8%), non-arthroplasty knee surgeries (27.2%), elbow procedures (19.6%), and hand surgery (15.4%). By case load, non-arthroplasty shoulder procedures consisted of 43% of Medicare volume over the study period (Figure 1). Between 2012 and 2014, the average proportion of Medicare cases that were shoulder arthroplasties increased from 13.8% (21,137/152,862) to 16.7% (26,865/160,851; P = .001). Shoulder arthroplasty constituted 100% of the Medicare surgical case volume for 67 (4.9%; 67/1374) of the surgeons.    

Table 2. Case Volumes over Time with All Procedures Categorized by Anatomic Region and Arthroplasty vs Non-arthroplasty*

* Procedures of interest with high case volumes are reported individually.

Only 44.3% (609/1374) of surgeons performed TSA in a minimum of 11 Medicare patients in all 3 years of the study period (consistent providers of TSA), providing a total of 55,538 TSA (77.2%; 55,538/71,973). When fellowship training was evaluated, 191 (31.4%; 191/609) of these surgeons were shoulder and elbow fellowship trained (21,444 TSA; 38.6%; Table 3). More than one-third (36.6%; 223/609) had completed a sports surgery fellowship (18,899 TSA; 34.0%). Surgeons trained in hand surgery (12.5%; 76/609) and adult reconstruction (5.3%; 32/610) also made contributions to meeting the TSA demand with 6971 (12.6%) and 2485 (4.5%) TSA, respectively. One-fifth of this cohort (18.1%; 110/609) had unknown fellowship training: they either reported no fellowship (13.6%; 83/609) or did not specify the type of training (4.4%; 27/609). Shoulder and elbow fellowship-trained surgeons performed more TSA (median: 89.0 TSA per surgeon between 2012 and 2014) than surgeons without shoulder and elbow fellowship training (median: 67.0 TSA per surgeon; P < 0.001). More than one-third (37%) of shoulder and elbow fellowship-trained surgeons’ surgical case volume was comprised of TSA, with an additional 35% from non-arthroplasty shoulder procedures (Figure 2). In order for the current supply of shoulder and elbow fellowship-trained surgeons to meet the Medicare TSA demand, each fellowship graduate would have to perform 140.6 TSA in Medicare patients annually. Shoulder and elbow fellowship-trained surgeons were more likely to practice in referral regions with an increased Medicare population (P < .001), an increased number of surgeons performing TSA (P < .001), and a higher proportion of Medicaid-eligible patients (P = .01; Table 4). Shoulder and elbow fellowship-trained surgeons (18.7 years post-medical school graduation) were also earlier in their careers than other consistent TSA surgeons (23.1 years post-graduation; P < .001). 

Table 3. A Representation of Fellowships Among TSA Surgeons and Their Shoulder Arthroplasty Case Load*

* Not all fellowships (eg, oncology) included due to small numbers. Also, many surgeons performed multiple fellowships.

Abbreviations: SA, shoulder arthroplasty; TSA, total shoulder arthroplasty.

Table 4. Breakdown of Geographic Characteristics of Orthopedic Surgeons Consistently

Performing TSA between 2012 and 2014 Stratified by Fellowship Training

Abbreviations: HRR, hospital referral region; TSA, total shoulder arthroplasty.

Continue to: DISCUSSION...

DISCUSSION

Utilization of TSA has continued to rise; however, access to this cost-effective procedure was recently demonstrated to be limited.¹¹ In a separate analysis, we established the continued rise in use of TSA in the Medicare population, coupled with an increase in the number of surgeons routinely performing TSA.² Multiple analyses have demonstrated the importance of high-volume surgeons and hospitals familiar with the intricacies of shoulder arthroplasty concepts in minimizing complications, improving the quality and decreasing the cost of TSA.^6,10,17 Specifically, Singh and colleagues18 demonstrated from a multi-center registry that surgeons and hospitals with greater shoulder arthroplasty volumes had decreased intra-operative blood loss, operative time, and hospital length of stay. As the demand for TSA, both anatomic and reverse, continues to rise, it is imperative that the healthcare delivery system is optimized to provide the best possible care. Before we can determine whether specialized training in shoulder arthroplasty influences surgical outcomes, characteristics and training of surgeons performing TSA should be described. 

The number of surgeons performing >10 TSA in the Medicare population rose significantly between 2012 and 2014 (29.3%). However, the number of TSAs per surgeon over this time period remained consistent (approximately 25 per surgeon). Furthermore, the increase in the number of surgeons performing a reportable volume of TSA by 2014 was from the addition of already active surgeons (ie, the growth in TSA was not from the addition of newly trained arthroplasty surgeons but originated from the existing orthopedic surgeon workforce). In a recently published analysis, Somerson and colleagues, ¹¹ using this same dataset, demonstrated persistent limitations in access to high-volume TSA surgeons. In a more recent analysis, we showed that while still lacking for some patients, access to a high-volume TSA surgeon has improved significantly over the past 3 years, with 96.9% of the United States population residing within 200 kilometers of a high-volume TSA surgeon (>20 Medicare cases).² This analysis validates those findings, with the caveat that the average annual volume per surgeon is not increasing. What remains unknown, due to limitations of this dataset, is how many surgeons are not identified because they are performing ≤10 TSA each year or are performing TSA in non-Medicare patients. 

With the specialization of healthcare delivery, specifically in orthopedics, it is imperative that mechanisms for providing specialty-focused care be established. However, the proportion of their practice that surgeons dedicate to TSA was unknown. This study demonstrates that this proportion is increasing. Including non-arthroplasty procedures, more than half (58%) of the procedures performed by this surgeon cohort were shoulder-specific. Furthermore, this analysis demonstrates that surgeons performing TSA have significant case diversity, including nearly half of the cohort performing TKA. Repeated evidence has demonstrated the effect of case volume on improved outcomes following orthopedic procedures.^8,19–21 The pre-existing location-based model for delivering orthopedic care supports case diversity; however, this model continues to be challenged with high-volume centers of excellence and patient travel.^22–24 Hip and knee arthroplasty experienced a similar surge in demand, with a subsequent shift in care to high-volume surgeons and centers.²⁵ Shoulder and elbow fellowship-trained surgeons would need to nearly quadruple their current Medicare TSA volume to meet the entire current demand for TSA in the Medicare population (and this does not account for TSA performed by very low-volume surgeons not included in this cohort). With increased utilization of TSA, policymakers and the orthopedic community must determine the structure of delivery (centers of excellence or medium-volume disseminated throughout the country) that is optimal. 

For those surgeons consistently performing TSA over the study period, fellowship training was diverse. While the current focus in orthopedics is on case volume, research in other specialties, namely general surgery, has provided repeated evidence that surgical specialization (more so than high case volume) provides improved outcomes.^26–29Furthermore, Leopold and colleagues³⁰ demonstrated an inverse relationship between competency in performing a procedure and confidence in one’s ability to do so. In their study, educational intervention provided improved competency in the procedure. Less than one-third (29.8%) of TSA in this cohort were performed by a shoulder and elbow fellowship-trained surgeon consistently performing this procedure. Approximately another quarter (26.2%) were performed by consistent TSA surgeons trained in sports surgery. Meanwhile, 34.6% of TSA in this study cohort were performed by a surgeon who did not consistently meet the minimum threshold in all study years (16,435 TSA; 22.8%) or by a surgeon performing TSA without fellowship training (8,489 TSA; 11.8%). There has been a trend toward orthopedic subspecialty training with an increased demand for fellowship-trained surgeons.³¹ Despite this and the complexities of TSA, many continue to be performed by surgeons with an inconsistent volume and those without arthroplasty-specific fellowship training. The available evidence supports a push toward the fellowship-trained, high-volume TSA surgeon in providing reproducible high-quality shoulder arthroplasty care. For now, that surgeon is more likely to be earlier in his/her career and reside in large, referral-based centers surrounded by other surgeons performing TSA.

These findings must be considered in the light of the study limitations. First, this is a large publicly available database. While this type of database provides a unique opportunity to assess the geographic distributions and characteristics of orthopedic surgeons, specifically those performing TSA, it completely prevents any assessment of the relationship between these findings and quality. As such, while the reader may generate hypotheses regarding the implications of our findings on the quality of TSA delivery, the true effects cannot be determined. In the same vein, for the purpose of privacy, surgeons performing ≤10 TSA were not included in this dataset. This limitation prevents the identification of low-volume TSA surgeons. Also, it is likely that the observed increase in surgeons over time is likely a reflection of small increases in volume for surgeons already performing TSA. Lastly, a web-based search was undertaken to identify surgeons’ self-reported fellowship training. The results of this web-based search could not be validated, and it is possible that fellowship training, or the lack thereof, was mischaracterized and simply not obtainable through a web-based search. Furthermore, it is not possible to fully assess the extent of high-quality TSA training in these various fellowships.

CONCLUSION

In just the past decade, the utilization of TSA in the Medicare population has increased significantly. However, this increase was not achieved by the addition of highly specialized, high-volume surgeons but by the addition of many surgeons performing lower numbers of TSA surgeries. Furthermore, for those performing this cost-effective procedure, TSA constitutes a relatively small proportion of the surgeries they perform. Shoulder and elbow fellowship-trained surgeons currently account for a low percentage of the overall number of surgeons performing TSA. The implications of these findings must be considered and investigated.

The dilemma for the treating surgeon is deciding which patients to treat with RCR and who to treat with rTSA. Predicting which surgical procedure will provide a better functional result is difficult and controversial.⁷ The RCR method is a bone-conserving procedure with relatively low surgical risk and allows the option for rTSA to be performed as a salvage surgery should repair fail. It also may be less costly in the appropriate population.⁸ However, large rotator cuff tears in elderly patients have low healing potential, and the prospect of participating in a lengthy rehabilitation after an operation that may not prove successful can be deterring.^9,10 In the elderly population, rTSA may be a reliable option, as tendon healing of the cuff is not necessary to restore function. However, rTSA does not conserve bone, provides a non-anatomic solution, and has had historically high complication rates.^4,5

In an effort to aid in the decision-making process when considering these 2 surgical options, we compared RCR and rTSA performed at a single institution for MCT in patients >65 years. Our aim was to identify preoperative patient variables that influence a surgeon’s decision to proceed with 1 of the 2 procedures. Moreover, we evaluated clinical outcomes in these 2 patient populations. We hypothesized that (1) patients selected for rTSA would have worse preoperative function, less range of motion, more comorbidities, more evidence of radiographic subluxation, and a higher likelihood of having undergone previous RCR than those selected for RCR, and (2) both RCR and rTSA would be successful and result in improved clinical outcomes with high patient satisfaction.

MATERIALS AND METHODS

PATIENT SELECTION

We performed a retrospective chart review using our practice database of all patients undergoing arthroscopic RCR and rTSA for any indication by the senior author (M.A.F.) between January 2004 and April 2015. A total of 1503 RCRs and 1973 rTSAs were conducted during the study period. Patient medical records were reviewed, and those meeting the following criteria were included in the study: >65 years at the time of surgery, MCT, no preoperative glenohumeral arthritis, minimum follow-up of 12 months, functional deltoid muscle on physical examination, and no prior shoulder surgery except for RCR or diagnostic arthroscopy. A total of 92 patients who underwent arthroscopic RCR and 89 patients who underwent rTSA met the inclusion criteria. For patients with bilateral shoulder surgery, we measured each shoulder independently. Three patients underwent bilateral rTSA, and 3 patients underwent bilateral RCR, leaving 95 shoulders in the RCR group and 92 in the rTSA group. The Western Institutional Review Board determined this study to be exempt from review.

RADIOGRAPHIC EVALUATION

All patient charts included a radiology report and documented interpretation of the images by the treating surgeon prior to surgery. Radiographs were assessed to assure the absence of preoperative glenohumeral osteoarthritis. The images were also graded based on the Hamada classification.¹¹ Stage 1 is associated with minimal radiographic change with an acromiohumeral interval (AHI) >6 mm; stage 2 is characterized by narrowing of the AHI <6 mm; and Stage 3 is defined by narrowing of the AHI with radiographic changes of the acromion. Stages 4 and higher include arthritic changes to the glenohumeral joint, and they were not included in the study population. The AHI measurements and the presence or absence of glenohumeral subluxation were documented.

Continue to: MASSIVE CUFF TEAR DETERMINATION...

We defined MCT on the basis of previously described criteria of tears involving ≥2 tendons or tears measuring ≥5 cm in greatest dimension.^12,13 Patient charts were screened, and those whose clinical notes or radiology reports indicated an absence of MCT were excluded. Preoperative imaging of the remaining patients was then evaluated by 3 fellowship-trained shoulder surgeons to confirm MCT in all patients with a clinically documented MCT, as well as to assess those who had insufficient documentation of tear size in the notes.

Advanced imaging was evaluated for fatty atrophy of the rotator cuff musculature, and Goutallier classification was assigned.^14,15 Length of retraction was measured from the tendon end to the medial aspect of the footprint on coronal imaging, and the subscapularis and teres minor were assessed and documented as torn or intact.^16,17

DATA COLLECTION

We reviewed clinical charts and patient questionnaire forms from both the preoperative and follow-up visits. Clinical data collected included gender, age at surgery, active range of motion (forward elevation, abduction, external rotation, and internal rotation), comorbidities, smoking status, BMI, history of shoulder surgery, and any postoperative complications or need for secondary surgery. All patients completed patient-centered questionnaires regarding shoulder pain and dysfunction at each visit or via telephone communication with clinic staff. Outcome measurements used for analysis included American Shoulder and Elbow Surgeons (ASES) Score, simple shoulder test (SST), visual analog score (VAS) pain scale, and patient-reported satisfaction (Graded 1-10; 1 = poor outcome; 4 = satisfactory outcome; 7 = good outcome; 10 = excellent outcome).

DATA ANALYSIS AND STATISTICAL METHODS

Statistical tests were selected based on the result of Shapiro–Wilk test for normality. Continuous variables were evaluated with either independent t test or Mann–Whitney U test. Dependent t test was used to evaluate outcome variables. For categorical variables, either Pearson’s χ²or Fisher’s exact test was performed depending on the sample size. Alpha was set at P =.05.

Continue to: RESULTS...

PREOPERATIVE CHARACTERISTICS

Of the 187 shoulders in the study group, 95 had RCR and 92 had rTSA. Demographic information and preoperative variables for both groups are summarized in Table 1 and Table 2. Patients in the RCR group had greater preoperative forward elevation, abduction, and external rotation and higher preoperative functional scores than those in the rTSA group. Patients in the rTSA group were older and more likely to be female than those in the RCR group. More patients in the rTSA group had undergone prior RCR compared with those in the RCR group. Each of these differences was statistically significant. Subjective pain scores, BMI, and comorbidities were similar between the 2 groups.

Table 1. Patient demographics

Abbreviations: BMI, body mass index; RCR, rotator cuff repair; rTSA, reverse total shoulder arthroplasty.

Table 2. Preoperative variables

Abbreviations: AB, abduction; ASES, American Shoulder and Elbow Society score; ER, external rotation; FE, forward elevation; IR, internal rotation; preop, preoperative; SST, simple shoulder test.

Radiographically, patients selected to undergo rTSA had a smaller AHI (4.8 vs 8.7, P < .0001) and more evidence of superior subluxation (50.6% vs 14.1%, P < .0001) than those in the RCR group. Average Hamada grade was 1.4 ± 0.7 and 2.2 ± 0.7 for the RCR and rTSA groups, respectively (P < .0001). Average Goutallier grade was similar between the groups (2.2 ± 0.6 for RCR vs 2.2 ± 0.8 for rTSA, P =.227), and 25.5% of the RCR group had Grade 3 or 4 atrophy compared with 28.1% of the rTSA group.

POSTOPERATIVE OUTCOMES

The average follow-up time was 44 months for RCR and 47 months for rTSA. Patients in the RCR and rTSA groups were highly satisfied with the surgery (8.5 ± 2.6 vs 8.2 ± 2.6, P = .461) and had significantly increased range of motion in all planes and improved functional scores (Table 3). The rTSA group had greater net improvement in forward elevation, abduction, and external rotation than the RCR group. Both groups demonstrated similar improvement in ASES, SST, and VAS pain scores.

Table 3. Postoperative outcomes

Abbreviations: AB, abduction; ASES, American Shoulder and Elbow Society score; ER, external rotation; FE, forward elevation; IR, internal rotation; SST, simple shoulder test; VAS – visual analog score.

In the RCR group, 5 patients (5.3%) required reoperation: 3 patients underwent conversion to rTSA, 1 patient underwent biceps tenotomy with subacromial decompression, and 1 patient underwent arthroscopic irrigation and debridement for a postoperative Propionibacterium acnes infection. In the rTSA group, 2 patients (2.2%) required reoperation: 1 patient underwent open reduction internal fixation for a scapula fracture that failed conservative management, and 1 patient had an open irrigation and debridement with polyethylene exchange for an acute postoperative infection of unknown source.

DISCUSSION

Massive, retracted rotator cuff tears are a common and difficult problem.¹ The treatment options are numerous and depend on a variety of preoperative factors including patient-specific characteristics and factors specific to the tear. For certain patients, nonoperative management may be a reasonable first step, as an MCT does not necessarily preclude painless, functional shoulder motion. Elderly, lower demand individuals have been shown to do well with physical rehabilitation.¹⁸ Similarly, for the same category of elderly patients who do not respond to conservative measures, arthroscopic tendon debridement with or without subacromial decompression and/or biceps tenotomy may be effective.^1,19 This technique has been described as “limited goals surgery;” despite some mixed results in the literature, multiple studies have reported symptomatic and functional improvement after simple debridement.^2,19–21The consensus among several authors has been that this procedure continues to play a role for elderly, low-demand patients whose functional goals are limited and whose primary complaint is pain.^1,2,20

For the majority of patients with MCT who desire pain relief and a restoration of shoulder function, RCR remains the gold standard of treatment and should be the primary aim if feasible. Complete RCR has consistently outperformed both partial repair and debridement in multiple studies in terms of pain relief and functional improvement.^10,21,22However, elderly patients with chronic, massive tears, particularly in the setting of muscle atrophy, are at high risk of failure with attempted cuff repair.^9,23 Novel techniques such as superior capsular reconstruction and subacromial balloon spacer implantation may offer a minimally invasive method of re-centering the humeral head and stabilizing the glenohumeral joint; however, these new treatment options lack any long-term data in the literature to support their widespread use.^24–26 Alternatively, rTSA has been shown to be a reliable option to restore shoulder function in the setting of a massive irreparable rotator cuff tear, even in the absence of arthritis.^5,27-31

Continue to: The decision-making process...

The decision-making process for selecting RCR or rTSA in the setting of MCT without arthritis in the older population (age >65 years) remains challenging. We attempted to quantify the data of a high-volume surgeon and identify the differences and similarities between those patients selected for either procedure. At our institution, we generally performed rTSA on patients with low preoperative range of motion, poor function based on SST and ASES scores, small AHI, and strong evidence of superior subluxation. We were also more likely to perform rTSA if the patient had a history of rotator cuff surgery. There was a predilection for older age and female gender in those who underwent rTSA.

For our study, we elected to focus on patients >65 years. In our experience, the choice of which surgical procedure to perform is generally easier in younger patients. Most surgeons appropriately opt for an arthroscopic procedure or tendon transfer to preserve bone and maintain the option of rTSA as a salvage procedure if necessary in the future. Studies have reported that <60 years is a predictor of poor outcome with rTSA, and patients <65 years who undergo rTSA have been shown to have high complication rates.^30-32 Furthermore, the longevity of the implant in young patients is a significant concern, and revision surgery after rTSA is technically demanding and known to result in poor functional outcomes.^32,33

Although the indications for rTSA are expanding, attempts at RCR in the setting of MCT remain largely appropriate. Preserved preoperative anterior elevation >90° has been associated with loss of motion after rTSA and poor satisfaction, and one should exercise caution when considering rTSA in this setting.³ The current study confirmed that even older patients with MCT may be very satisfied with arthroscopic RCR (Figure 1). Both range of motion and function significantly improved, and patients were largely satisfied with the procedure with an average self-reported outcome of good to excellent. At the time of final follow-up for this study, only 3 shoulders in the RCR group had undergone conversion to rTSA. This number may be expected to rise with long follow-up periods, and we feel that prolonging the time before arthroplasty is generally in the best interest of the patient.

Our results were consistent with several reported studies in which RCR has been shown to be successful in the setting of MCT.^34–37 Henry and colleagues³⁶ performed a systematic review that evaluated 954 patients who underwent partial or complete anatomic RCR for MCT. Although the average age was 63 years (range, 37–87), functional outcome scores, VAS pain score, and overall range of motion consistently and significantly improved.

rTSA may be a “more reliable” option than RCR in treating MCT in the older population because it does not rely on tendon healing. However, the relationship between tendon healing and clinical outcomes after RCR is unclear. The aforementioned systematic review reported re-tear rates to be as high as 79%, but several studies have reported high satisfaction even in the setting of retear.³⁶ Yoo and colleagues³⁸ and Chung and colleagues⁹ reported re-tear rates of 45.5% and 39.8%, respectively, but both studies noted that there was no difference in outcome measures between those patients with and without re-tears. In particular, for patients who have had no prior rotator cuff surgery, an attempt at arthroscopic repair may be a prudent option with relatively low risk.

Although certain patients may clinically improve despite suffering a re-tear (or inability to heal in the first place), others continue to experience pain and dysfunction that negatively affect their quality of life.^39–41 These patients are more often appropriate candidates for rTSA. Indeed, several studies have demonstrated a higher re-tear rate in patients with a history of surgery than in those without.^23,31,38,42 Shamsudin and colleagues⁴³ found revision arthroscopic RCR, even in a younger age group with tears of all sizes, to be twice as likely to re-tear. Notably, re-tear after revision repair may be more likely to be symptomatic, as these re-tears are routinely associated with pain, stiffness, and loss of function. Even in the hands of experienced surgeons in a younger population, revision repair has only been able to reverse pseudoparalysis in 43% of patients, leading to only 39% return to sport or full activity.⁴⁴ In examining our data, we were much less likely to perform an RCR in patients who had a history of cuff repair surgery than in those without this history.

Continue to: Overall, those patients selected for rTSA...

Overall, those patients selected for rTSA in our study population performed well postoperatively (Figure 2 and Figure 3). Vast improvements were noted in range of motion, function, and pain scores at final follow up. Moreover, no patients in the study group required revision arthroplasty during the follow-up period. Although the average follow-up period was only 47 months, these results suggested that elderly patients with MCT without arthritis may be particularly ideal candidates for rTSA with regard to implant survival and anticipated revision rate when chosen appropriately.

Several weaknesses were noted within this paper. First, the study was retrospective, precluding randomization of treatment groups and standardization of data collection and follow-up. The outcomes of RCR and rTSA could not be compared directly due to the inherent selection bias. The groups clearly differed in many respects, and these preoperative factors likely played a role in postoperative outcomes. However, the primary goal of this study was not to compare outcomes of the treatment groups but to analyze the patterns of patient selection by an experienced treating surgeon and contribute to published data that each surgery can be successful in this patient population when chosen appropriately.

Second, our data were based on a single surgeon’s decisions, and results may not be generalizable. Furthermore, the senior author has had a longstanding interest in reverse shoulder arthroplasty and has published data illustrating successful outcomes for rTSA in patients with MCT. For this reason, one could presume that there may have been some bias toward treating patients with rTSA. However, we feel that the senior author’s unique and longstanding experience in treating MCT allows for a thorough evaluation and comparison of preoperative variables and outcomes declared within this study. Indeed, many patients included in this study were referred from outside institutions specifically for rTSA but instead were deemed more appropriate candidates for RCR and underwent successful arthroscopic repair, a common scenario which served as an impetus for this study.

CONCLUSION

RCR and rTSA are both viable options for patients >65 years with MCT without arthritis. Treatment must be individualized for each patient with careful consideration of a number of preoperative variables and patient characteristics. At our institution, patients with previous RCR, decreased range of motion, poor function, and strong radiographic evidence of subluxation are more likely to undergo rTSA. When chosen appropriately, both RCR and rTSA can result in improved range of motion, function, and high patient satisfaction in this patient population.

The dilemma for the treating surgeon is deciding which patients to treat with RCR and who to treat with rTSA. Predicting which surgical procedure will provide a better functional result is difficult and controversial.⁷ The RCR method is a bone-conserving procedure with relatively low surgical risk and allows the option for rTSA to be performed as a salvage surgery should repair fail. It also may be less costly in the appropriate population.⁸ However, large rotator cuff tears in elderly patients have low healing potential, and the prospect of participating in a lengthy rehabilitation after an operation that may not prove successful can be deterring.^9,10 In the elderly population, rTSA may be a reliable option, as tendon healing of the cuff is not necessary to restore function. However, rTSA does not conserve bone, provides a non-anatomic solution, and has had historically high complication rates.^4,5

In an effort to aid in the decision-making process when considering these 2 surgical options, we compared RCR and rTSA performed at a single institution for MCT in patients >65 years. Our aim was to identify preoperative patient variables that influence a surgeon’s decision to proceed with 1 of the 2 procedures. Moreover, we evaluated clinical outcomes in these 2 patient populations. We hypothesized that (1) patients selected for rTSA would have worse preoperative function, less range of motion, more comorbidities, more evidence of radiographic subluxation, and a higher likelihood of having undergone previous RCR than those selected for RCR, and (2) both RCR and rTSA would be successful and result in improved clinical outcomes with high patient satisfaction.

MATERIALS AND METHODS

PATIENT SELECTION

We performed a retrospective chart review using our practice database of all patients undergoing arthroscopic RCR and rTSA for any indication by the senior author (M.A.F.) between January 2004 and April 2015. A total of 1503 RCRs and 1973 rTSAs were conducted during the study period. Patient medical records were reviewed, and those meeting the following criteria were included in the study: >65 years at the time of surgery, MCT, no preoperative glenohumeral arthritis, minimum follow-up of 12 months, functional deltoid muscle on physical examination, and no prior shoulder surgery except for RCR or diagnostic arthroscopy. A total of 92 patients who underwent arthroscopic RCR and 89 patients who underwent rTSA met the inclusion criteria. For patients with bilateral shoulder surgery, we measured each shoulder independently. Three patients underwent bilateral rTSA, and 3 patients underwent bilateral RCR, leaving 95 shoulders in the RCR group and 92 in the rTSA group. The Western Institutional Review Board determined this study to be exempt from review.

RADIOGRAPHIC EVALUATION

All patient charts included a radiology report and documented interpretation of the images by the treating surgeon prior to surgery. Radiographs were assessed to assure the absence of preoperative glenohumeral osteoarthritis. The images were also graded based on the Hamada classification.¹¹ Stage 1 is associated with minimal radiographic change with an acromiohumeral interval (AHI) >6 mm; stage 2 is characterized by narrowing of the AHI <6 mm; and Stage 3 is defined by narrowing of the AHI with radiographic changes of the acromion. Stages 4 and higher include arthritic changes to the glenohumeral joint, and they were not included in the study population. The AHI measurements and the presence or absence of glenohumeral subluxation were documented.

Continue to: MASSIVE CUFF TEAR DETERMINATION...

We defined MCT on the basis of previously described criteria of tears involving ≥2 tendons or tears measuring ≥5 cm in greatest dimension.^12,13 Patient charts were screened, and those whose clinical notes or radiology reports indicated an absence of MCT were excluded. Preoperative imaging of the remaining patients was then evaluated by 3 fellowship-trained shoulder surgeons to confirm MCT in all patients with a clinically documented MCT, as well as to assess those who had insufficient documentation of tear size in the notes.

Advanced imaging was evaluated for fatty atrophy of the rotator cuff musculature, and Goutallier classification was assigned.^14,15 Length of retraction was measured from the tendon end to the medial aspect of the footprint on coronal imaging, and the subscapularis and teres minor were assessed and documented as torn or intact.^16,17

DATA COLLECTION

We reviewed clinical charts and patient questionnaire forms from both the preoperative and follow-up visits. Clinical data collected included gender, age at surgery, active range of motion (forward elevation, abduction, external rotation, and internal rotation), comorbidities, smoking status, BMI, history of shoulder surgery, and any postoperative complications or need for secondary surgery. All patients completed patient-centered questionnaires regarding shoulder pain and dysfunction at each visit or via telephone communication with clinic staff. Outcome measurements used for analysis included American Shoulder and Elbow Surgeons (ASES) Score, simple shoulder test (SST), visual analog score (VAS) pain scale, and patient-reported satisfaction (Graded 1-10; 1 = poor outcome; 4 = satisfactory outcome; 7 = good outcome; 10 = excellent outcome).

DATA ANALYSIS AND STATISTICAL METHODS

Statistical tests were selected based on the result of Shapiro–Wilk test for normality. Continuous variables were evaluated with either independent t test or Mann–Whitney U test. Dependent t test was used to evaluate outcome variables. For categorical variables, either Pearson’s χ²or Fisher’s exact test was performed depending on the sample size. Alpha was set at P =.05.

Continue to: RESULTS...

PREOPERATIVE CHARACTERISTICS

Of the 187 shoulders in the study group, 95 had RCR and 92 had rTSA. Demographic information and preoperative variables for both groups are summarized in Table 1 and Table 2. Patients in the RCR group had greater preoperative forward elevation, abduction, and external rotation and higher preoperative functional scores than those in the rTSA group. Patients in the rTSA group were older and more likely to be female than those in the RCR group. More patients in the rTSA group had undergone prior RCR compared with those in the RCR group. Each of these differences was statistically significant. Subjective pain scores, BMI, and comorbidities were similar between the 2 groups.

Table 1. Patient demographics

Abbreviations: BMI, body mass index; RCR, rotator cuff repair; rTSA, reverse total shoulder arthroplasty.

Table 2. Preoperative variables

Abbreviations: AB, abduction; ASES, American Shoulder and Elbow Society score; ER, external rotation; FE, forward elevation; IR, internal rotation; preop, preoperative; SST, simple shoulder test.

Radiographically, patients selected to undergo rTSA had a smaller AHI (4.8 vs 8.7, P < .0001) and more evidence of superior subluxation (50.6% vs 14.1%, P < .0001) than those in the RCR group. Average Hamada grade was 1.4 ± 0.7 and 2.2 ± 0.7 for the RCR and rTSA groups, respectively (P < .0001). Average Goutallier grade was similar between the groups (2.2 ± 0.6 for RCR vs 2.2 ± 0.8 for rTSA, P =.227), and 25.5% of the RCR group had Grade 3 or 4 atrophy compared with 28.1% of the rTSA group.

POSTOPERATIVE OUTCOMES

The average follow-up time was 44 months for RCR and 47 months for rTSA. Patients in the RCR and rTSA groups were highly satisfied with the surgery (8.5 ± 2.6 vs 8.2 ± 2.6, P = .461) and had significantly increased range of motion in all planes and improved functional scores (Table 3). The rTSA group had greater net improvement in forward elevation, abduction, and external rotation than the RCR group. Both groups demonstrated similar improvement in ASES, SST, and VAS pain scores.

Table 3. Postoperative outcomes

Abbreviations: AB, abduction; ASES, American Shoulder and Elbow Society score; ER, external rotation; FE, forward elevation; IR, internal rotation; SST, simple shoulder test; VAS – visual analog score.

In the RCR group, 5 patients (5.3%) required reoperation: 3 patients underwent conversion to rTSA, 1 patient underwent biceps tenotomy with subacromial decompression, and 1 patient underwent arthroscopic irrigation and debridement for a postoperative Propionibacterium acnes infection. In the rTSA group, 2 patients (2.2%) required reoperation: 1 patient underwent open reduction internal fixation for a scapula fracture that failed conservative management, and 1 patient had an open irrigation and debridement with polyethylene exchange for an acute postoperative infection of unknown source.

DISCUSSION

Massive, retracted rotator cuff tears are a common and difficult problem.¹ The treatment options are numerous and depend on a variety of preoperative factors including patient-specific characteristics and factors specific to the tear. For certain patients, nonoperative management may be a reasonable first step, as an MCT does not necessarily preclude painless, functional shoulder motion. Elderly, lower demand individuals have been shown to do well with physical rehabilitation.¹⁸ Similarly, for the same category of elderly patients who do not respond to conservative measures, arthroscopic tendon debridement with or without subacromial decompression and/or biceps tenotomy may be effective.^1,19 This technique has been described as “limited goals surgery;” despite some mixed results in the literature, multiple studies have reported symptomatic and functional improvement after simple debridement.^2,19–21The consensus among several authors has been that this procedure continues to play a role for elderly, low-demand patients whose functional goals are limited and whose primary complaint is pain.^1,2,20

For the majority of patients with MCT who desire pain relief and a restoration of shoulder function, RCR remains the gold standard of treatment and should be the primary aim if feasible. Complete RCR has consistently outperformed both partial repair and debridement in multiple studies in terms of pain relief and functional improvement.^10,21,22However, elderly patients with chronic, massive tears, particularly in the setting of muscle atrophy, are at high risk of failure with attempted cuff repair.^9,23 Novel techniques such as superior capsular reconstruction and subacromial balloon spacer implantation may offer a minimally invasive method of re-centering the humeral head and stabilizing the glenohumeral joint; however, these new treatment options lack any long-term data in the literature to support their widespread use.^24–26 Alternatively, rTSA has been shown to be a reliable option to restore shoulder function in the setting of a massive irreparable rotator cuff tear, even in the absence of arthritis.^5,27-31

Continue to: The decision-making process...

The decision-making process for selecting RCR or rTSA in the setting of MCT without arthritis in the older population (age >65 years) remains challenging. We attempted to quantify the data of a high-volume surgeon and identify the differences and similarities between those patients selected for either procedure. At our institution, we generally performed rTSA on patients with low preoperative range of motion, poor function based on SST and ASES scores, small AHI, and strong evidence of superior subluxation. We were also more likely to perform rTSA if the patient had a history of rotator cuff surgery. There was a predilection for older age and female gender in those who underwent rTSA.

For our study, we elected to focus on patients >65 years. In our experience, the choice of which surgical procedure to perform is generally easier in younger patients. Most surgeons appropriately opt for an arthroscopic procedure or tendon transfer to preserve bone and maintain the option of rTSA as a salvage procedure if necessary in the future. Studies have reported that <60 years is a predictor of poor outcome with rTSA, and patients <65 years who undergo rTSA have been shown to have high complication rates.^30-32 Furthermore, the longevity of the implant in young patients is a significant concern, and revision surgery after rTSA is technically demanding and known to result in poor functional outcomes.^32,33

Although the indications for rTSA are expanding, attempts at RCR in the setting of MCT remain largely appropriate. Preserved preoperative anterior elevation >90° has been associated with loss of motion after rTSA and poor satisfaction, and one should exercise caution when considering rTSA in this setting.³ The current study confirmed that even older patients with MCT may be very satisfied with arthroscopic RCR (Figure 1). Both range of motion and function significantly improved, and patients were largely satisfied with the procedure with an average self-reported outcome of good to excellent. At the time of final follow-up for this study, only 3 shoulders in the RCR group had undergone conversion to rTSA. This number may be expected to rise with long follow-up periods, and we feel that prolonging the time before arthroplasty is generally in the best interest of the patient.

Our results were consistent with several reported studies in which RCR has been shown to be successful in the setting of MCT.^34–37 Henry and colleagues³⁶ performed a systematic review that evaluated 954 patients who underwent partial or complete anatomic RCR for MCT. Although the average age was 63 years (range, 37–87), functional outcome scores, VAS pain score, and overall range of motion consistently and significantly improved.

rTSA may be a “more reliable” option than RCR in treating MCT in the older population because it does not rely on tendon healing. However, the relationship between tendon healing and clinical outcomes after RCR is unclear. The aforementioned systematic review reported re-tear rates to be as high as 79%, but several studies have reported high satisfaction even in the setting of retear.³⁶ Yoo and colleagues³⁸ and Chung and colleagues⁹ reported re-tear rates of 45.5% and 39.8%, respectively, but both studies noted that there was no difference in outcome measures between those patients with and without re-tears. In particular, for patients who have had no prior rotator cuff surgery, an attempt at arthroscopic repair may be a prudent option with relatively low risk.

Although certain patients may clinically improve despite suffering a re-tear (or inability to heal in the first place), others continue to experience pain and dysfunction that negatively affect their quality of life.^39–41 These patients are more often appropriate candidates for rTSA. Indeed, several studies have demonstrated a higher re-tear rate in patients with a history of surgery than in those without.^23,31,38,42 Shamsudin and colleagues⁴³ found revision arthroscopic RCR, even in a younger age group with tears of all sizes, to be twice as likely to re-tear. Notably, re-tear after revision repair may be more likely to be symptomatic, as these re-tears are routinely associated with pain, stiffness, and loss of function. Even in the hands of experienced surgeons in a younger population, revision repair has only been able to reverse pseudoparalysis in 43% of patients, leading to only 39% return to sport or full activity.⁴⁴ In examining our data, we were much less likely to perform an RCR in patients who had a history of cuff repair surgery than in those without this history.

Continue to: Overall, those patients selected for rTSA...

Overall, those patients selected for rTSA in our study population performed well postoperatively (Figure 2 and Figure 3). Vast improvements were noted in range of motion, function, and pain scores at final follow up. Moreover, no patients in the study group required revision arthroplasty during the follow-up period. Although the average follow-up period was only 47 months, these results suggested that elderly patients with MCT without arthritis may be particularly ideal candidates for rTSA with regard to implant survival and anticipated revision rate when chosen appropriately.

Several weaknesses were noted within this paper. First, the study was retrospective, precluding randomization of treatment groups and standardization of data collection and follow-up. The outcomes of RCR and rTSA could not be compared directly due to the inherent selection bias. The groups clearly differed in many respects, and these preoperative factors likely played a role in postoperative outcomes. However, the primary goal of this study was not to compare outcomes of the treatment groups but to analyze the patterns of patient selection by an experienced treating surgeon and contribute to published data that each surgery can be successful in this patient population when chosen appropriately.

Second, our data were based on a single surgeon’s decisions, and results may not be generalizable. Furthermore, the senior author has had a longstanding interest in reverse shoulder arthroplasty and has published data illustrating successful outcomes for rTSA in patients with MCT. For this reason, one could presume that there may have been some bias toward treating patients with rTSA. However, we feel that the senior author’s unique and longstanding experience in treating MCT allows for a thorough evaluation and comparison of preoperative variables and outcomes declared within this study. Indeed, many patients included in this study were referred from outside institutions specifically for rTSA but instead were deemed more appropriate candidates for RCR and underwent successful arthroscopic repair, a common scenario which served as an impetus for this study.

CONCLUSION

RCR and rTSA are both viable options for patients >65 years with MCT without arthritis. Treatment must be individualized for each patient with careful consideration of a number of preoperative variables and patient characteristics. At our institution, patients with previous RCR, decreased range of motion, poor function, and strong radiographic evidence of subluxation are more likely to undergo rTSA. When chosen appropriately, both RCR and rTSA can result in improved range of motion, function, and high patient satisfaction in this patient population.

The dilemma for the treating surgeon is deciding which patients to treat with RCR and who to treat with rTSA. Predicting which surgical procedure will provide a better functional result is difficult and controversial.⁷ The RCR method is a bone-conserving procedure with relatively low surgical risk and allows the option for rTSA to be performed as a salvage surgery should repair fail. It also may be less costly in the appropriate population.⁸ However, large rotator cuff tears in elderly patients have low healing potential, and the prospect of participating in a lengthy rehabilitation after an operation that may not prove successful can be deterring.^9,10 In the elderly population, rTSA may be a reliable option, as tendon healing of the cuff is not necessary to restore function. However, rTSA does not conserve bone, provides a non-anatomic solution, and has had historically high complication rates.^4,5

In an effort to aid in the decision-making process when considering these 2 surgical options, we compared RCR and rTSA performed at a single institution for MCT in patients >65 years. Our aim was to identify preoperative patient variables that influence a surgeon’s decision to proceed with 1 of the 2 procedures. Moreover, we evaluated clinical outcomes in these 2 patient populations. We hypothesized that (1) patients selected for rTSA would have worse preoperative function, less range of motion, more comorbidities, more evidence of radiographic subluxation, and a higher likelihood of having undergone previous RCR than those selected for RCR, and (2) both RCR and rTSA would be successful and result in improved clinical outcomes with high patient satisfaction.

MATERIALS AND METHODS

PATIENT SELECTION

We performed a retrospective chart review using our practice database of all patients undergoing arthroscopic RCR and rTSA for any indication by the senior author (M.A.F.) between January 2004 and April 2015. A total of 1503 RCRs and 1973 rTSAs were conducted during the study period. Patient medical records were reviewed, and those meeting the following criteria were included in the study: >65 years at the time of surgery, MCT, no preoperative glenohumeral arthritis, minimum follow-up of 12 months, functional deltoid muscle on physical examination, and no prior shoulder surgery except for RCR or diagnostic arthroscopy. A total of 92 patients who underwent arthroscopic RCR and 89 patients who underwent rTSA met the inclusion criteria. For patients with bilateral shoulder surgery, we measured each shoulder independently. Three patients underwent bilateral rTSA, and 3 patients underwent bilateral RCR, leaving 95 shoulders in the RCR group and 92 in the rTSA group. The Western Institutional Review Board determined this study to be exempt from review.

RADIOGRAPHIC EVALUATION

All patient charts included a radiology report and documented interpretation of the images by the treating surgeon prior to surgery. Radiographs were assessed to assure the absence of preoperative glenohumeral osteoarthritis. The images were also graded based on the Hamada classification.¹¹ Stage 1 is associated with minimal radiographic change with an acromiohumeral interval (AHI) >6 mm; stage 2 is characterized by narrowing of the AHI <6 mm; and Stage 3 is defined by narrowing of the AHI with radiographic changes of the acromion. Stages 4 and higher include arthritic changes to the glenohumeral joint, and they were not included in the study population. The AHI measurements and the presence or absence of glenohumeral subluxation were documented.

Continue to: MASSIVE CUFF TEAR DETERMINATION...

We defined MCT on the basis of previously described criteria of tears involving ≥2 tendons or tears measuring ≥5 cm in greatest dimension.^12,13 Patient charts were screened, and those whose clinical notes or radiology reports indicated an absence of MCT were excluded. Preoperative imaging of the remaining patients was then evaluated by 3 fellowship-trained shoulder surgeons to confirm MCT in all patients with a clinically documented MCT, as well as to assess those who had insufficient documentation of tear size in the notes.

Advanced imaging was evaluated for fatty atrophy of the rotator cuff musculature, and Goutallier classification was assigned.^14,15 Length of retraction was measured from the tendon end to the medial aspect of the footprint on coronal imaging, and the subscapularis and teres minor were assessed and documented as torn or intact.^16,17

DATA COLLECTION

We reviewed clinical charts and patient questionnaire forms from both the preoperative and follow-up visits. Clinical data collected included gender, age at surgery, active range of motion (forward elevation, abduction, external rotation, and internal rotation), comorbidities, smoking status, BMI, history of shoulder surgery, and any postoperative complications or need for secondary surgery. All patients completed patient-centered questionnaires regarding shoulder pain and dysfunction at each visit or via telephone communication with clinic staff. Outcome measurements used for analysis included American Shoulder and Elbow Surgeons (ASES) Score, simple shoulder test (SST), visual analog score (VAS) pain scale, and patient-reported satisfaction (Graded 1-10; 1 = poor outcome; 4 = satisfactory outcome; 7 = good outcome; 10 = excellent outcome).

DATA ANALYSIS AND STATISTICAL METHODS

Statistical tests were selected based on the result of Shapiro–Wilk test for normality. Continuous variables were evaluated with either independent t test or Mann–Whitney U test. Dependent t test was used to evaluate outcome variables. For categorical variables, either Pearson’s χ²or Fisher’s exact test was performed depending on the sample size. Alpha was set at P =.05.

Continue to: RESULTS...

PREOPERATIVE CHARACTERISTICS

Of the 187 shoulders in the study group, 95 had RCR and 92 had rTSA. Demographic information and preoperative variables for both groups are summarized in Table 1 and Table 2. Patients in the RCR group had greater preoperative forward elevation, abduction, and external rotation and higher preoperative functional scores than those in the rTSA group. Patients in the rTSA group were older and more likely to be female than those in the RCR group. More patients in the rTSA group had undergone prior RCR compared with those in the RCR group. Each of these differences was statistically significant. Subjective pain scores, BMI, and comorbidities were similar between the 2 groups.

Table 1. Patient demographics

Abbreviations: BMI, body mass index; RCR, rotator cuff repair; rTSA, reverse total shoulder arthroplasty.

Table 2. Preoperative variables

Abbreviations: AB, abduction; ASES, American Shoulder and Elbow Society score; ER, external rotation; FE, forward elevation; IR, internal rotation; preop, preoperative; SST, simple shoulder test.

Radiographically, patients selected to undergo rTSA had a smaller AHI (4.8 vs 8.7, P < .0001) and more evidence of superior subluxation (50.6% vs 14.1%, P < .0001) than those in the RCR group. Average Hamada grade was 1.4 ± 0.7 and 2.2 ± 0.7 for the RCR and rTSA groups, respectively (P < .0001). Average Goutallier grade was similar between the groups (2.2 ± 0.6 for RCR vs 2.2 ± 0.8 for rTSA, P =.227), and 25.5% of the RCR group had Grade 3 or 4 atrophy compared with 28.1% of the rTSA group.

POSTOPERATIVE OUTCOMES

The average follow-up time was 44 months for RCR and 47 months for rTSA. Patients in the RCR and rTSA groups were highly satisfied with the surgery (8.5 ± 2.6 vs 8.2 ± 2.6, P = .461) and had significantly increased range of motion in all planes and improved functional scores (Table 3). The rTSA group had greater net improvement in forward elevation, abduction, and external rotation than the RCR group. Both groups demonstrated similar improvement in ASES, SST, and VAS pain scores.

Table 3. Postoperative outcomes

Abbreviations: AB, abduction; ASES, American Shoulder and Elbow Society score; ER, external rotation; FE, forward elevation; IR, internal rotation; SST, simple shoulder test; VAS – visual analog score.

In the RCR group, 5 patients (5.3%) required reoperation: 3 patients underwent conversion to rTSA, 1 patient underwent biceps tenotomy with subacromial decompression, and 1 patient underwent arthroscopic irrigation and debridement for a postoperative Propionibacterium acnes infection. In the rTSA group, 2 patients (2.2%) required reoperation: 1 patient underwent open reduction internal fixation for a scapula fracture that failed conservative management, and 1 patient had an open irrigation and debridement with polyethylene exchange for an acute postoperative infection of unknown source.

DISCUSSION

Massive, retracted rotator cuff tears are a common and difficult problem.¹ The treatment options are numerous and depend on a variety of preoperative factors including patient-specific characteristics and factors specific to the tear. For certain patients, nonoperative management may be a reasonable first step, as an MCT does not necessarily preclude painless, functional shoulder motion. Elderly, lower demand individuals have been shown to do well with physical rehabilitation.¹⁸ Similarly, for the same category of elderly patients who do not respond to conservative measures, arthroscopic tendon debridement with or without subacromial decompression and/or biceps tenotomy may be effective.^1,19 This technique has been described as “limited goals surgery;” despite some mixed results in the literature, multiple studies have reported symptomatic and functional improvement after simple debridement.^2,19–21The consensus among several authors has been that this procedure continues to play a role for elderly, low-demand patients whose functional goals are limited and whose primary complaint is pain.^1,2,20

For the majority of patients with MCT who desire pain relief and a restoration of shoulder function, RCR remains the gold standard of treatment and should be the primary aim if feasible. Complete RCR has consistently outperformed both partial repair and debridement in multiple studies in terms of pain relief and functional improvement.^10,21,22However, elderly patients with chronic, massive tears, particularly in the setting of muscle atrophy, are at high risk of failure with attempted cuff repair.^9,23 Novel techniques such as superior capsular reconstruction and subacromial balloon spacer implantation may offer a minimally invasive method of re-centering the humeral head and stabilizing the glenohumeral joint; however, these new treatment options lack any long-term data in the literature to support their widespread use.^24–26 Alternatively, rTSA has been shown to be a reliable option to restore shoulder function in the setting of a massive irreparable rotator cuff tear, even in the absence of arthritis.^5,27-31

Continue to: The decision-making process...

The decision-making process for selecting RCR or rTSA in the setting of MCT without arthritis in the older population (age >65 years) remains challenging. We attempted to quantify the data of a high-volume surgeon and identify the differences and similarities between those patients selected for either procedure. At our institution, we generally performed rTSA on patients with low preoperative range of motion, poor function based on SST and ASES scores, small AHI, and strong evidence of superior subluxation. We were also more likely to perform rTSA if the patient had a history of rotator cuff surgery. There was a predilection for older age and female gender in those who underwent rTSA.

For our study, we elected to focus on patients >65 years. In our experience, the choice of which surgical procedure to perform is generally easier in younger patients. Most surgeons appropriately opt for an arthroscopic procedure or tendon transfer to preserve bone and maintain the option of rTSA as a salvage procedure if necessary in the future. Studies have reported that <60 years is a predictor of poor outcome with rTSA, and patients <65 years who undergo rTSA have been shown to have high complication rates.^30-32 Furthermore, the longevity of the implant in young patients is a significant concern, and revision surgery after rTSA is technically demanding and known to result in poor functional outcomes.^32,33

Although the indications for rTSA are expanding, attempts at RCR in the setting of MCT remain largely appropriate. Preserved preoperative anterior elevation >90° has been associated with loss of motion after rTSA and poor satisfaction, and one should exercise caution when considering rTSA in this setting.³ The current study confirmed that even older patients with MCT may be very satisfied with arthroscopic RCR (Figure 1). Both range of motion and function significantly improved, and patients were largely satisfied with the procedure with an average self-reported outcome of good to excellent. At the time of final follow-up for this study, only 3 shoulders in the RCR group had undergone conversion to rTSA. This number may be expected to rise with long follow-up periods, and we feel that prolonging the time before arthroplasty is generally in the best interest of the patient.

Our results were consistent with several reported studies in which RCR has been shown to be successful in the setting of MCT.^34–37 Henry and colleagues³⁶ performed a systematic review that evaluated 954 patients who underwent partial or complete anatomic RCR for MCT. Although the average age was 63 years (range, 37–87), functional outcome scores, VAS pain score, and overall range of motion consistently and significantly improved.

rTSA may be a “more reliable” option than RCR in treating MCT in the older population because it does not rely on tendon healing. However, the relationship between tendon healing and clinical outcomes after RCR is unclear. The aforementioned systematic review reported re-tear rates to be as high as 79%, but several studies have reported high satisfaction even in the setting of retear.³⁶ Yoo and colleagues³⁸ and Chung and colleagues⁹ reported re-tear rates of 45.5% and 39.8%, respectively, but both studies noted that there was no difference in outcome measures between those patients with and without re-tears. In particular, for patients who have had no prior rotator cuff surgery, an attempt at arthroscopic repair may be a prudent option with relatively low risk.

Although certain patients may clinically improve despite suffering a re-tear (or inability to heal in the first place), others continue to experience pain and dysfunction that negatively affect their quality of life.^39–41 These patients are more often appropriate candidates for rTSA. Indeed, several studies have demonstrated a higher re-tear rate in patients with a history of surgery than in those without.^23,31,38,42 Shamsudin and colleagues⁴³ found revision arthroscopic RCR, even in a younger age group with tears of all sizes, to be twice as likely to re-tear. Notably, re-tear after revision repair may be more likely to be symptomatic, as these re-tears are routinely associated with pain, stiffness, and loss of function. Even in the hands of experienced surgeons in a younger population, revision repair has only been able to reverse pseudoparalysis in 43% of patients, leading to only 39% return to sport or full activity.⁴⁴ In examining our data, we were much less likely to perform an RCR in patients who had a history of cuff repair surgery than in those without this history.

Continue to: Overall, those patients selected for rTSA...

Overall, those patients selected for rTSA in our study population performed well postoperatively (Figure 2 and Figure 3). Vast improvements were noted in range of motion, function, and pain scores at final follow up. Moreover, no patients in the study group required revision arthroplasty during the follow-up period. Although the average follow-up period was only 47 months, these results suggested that elderly patients with MCT without arthritis may be particularly ideal candidates for rTSA with regard to implant survival and anticipated revision rate when chosen appropriately.

Several weaknesses were noted within this paper. First, the study was retrospective, precluding randomization of treatment groups and standardization of data collection and follow-up. The outcomes of RCR and rTSA could not be compared directly due to the inherent selection bias. The groups clearly differed in many respects, and these preoperative factors likely played a role in postoperative outcomes. However, the primary goal of this study was not to compare outcomes of the treatment groups but to analyze the patterns of patient selection by an experienced treating surgeon and contribute to published data that each surgery can be successful in this patient population when chosen appropriately.

Second, our data were based on a single surgeon’s decisions, and results may not be generalizable. Furthermore, the senior author has had a longstanding interest in reverse shoulder arthroplasty and has published data illustrating successful outcomes for rTSA in patients with MCT. For this reason, one could presume that there may have been some bias toward treating patients with rTSA. However, we feel that the senior author’s unique and longstanding experience in treating MCT allows for a thorough evaluation and comparison of preoperative variables and outcomes declared within this study. Indeed, many patients included in this study were referred from outside institutions specifically for rTSA but instead were deemed more appropriate candidates for RCR and underwent successful arthroscopic repair, a common scenario which served as an impetus for this study.

CONCLUSION

RCR and rTSA are both viable options for patients >65 years with MCT without arthritis. Treatment must be individualized for each patient with careful consideration of a number of preoperative variables and patient characteristics. At our institution, patients with previous RCR, decreased range of motion, poor function, and strong radiographic evidence of subluxation are more likely to undergo rTSA. When chosen appropriately, both RCR and rTSA can result in improved range of motion, function, and high patient satisfaction in this patient population.

Anatomic factors that contribute to PFI include ligament laxity, trochlear dysplasia, patella alta, excessive femoral anteversion or tibial torsion, genu valgum, and increased tibial tubercle-to-trochlear groove distance.^1,3,8-11 When considering surgical treatment for PFI, all anatomic factors should be considered, with an emphasis on, and understanding of, the role of residual growth and development. One must also consider balancing static soft tissue, dynamic soft tissue, and osseous constraints of the patellofemoral joint to optimize the overall health and balance of the patella (Figure 1).

CHALLENGES IN TREATING PEDIATRIC PFI

A primary challenge in the treatment of PFI in pediatric patients is accounting for the impact of anatomic changes occurring secondary to growth and maturity. From birth to adulthood, the collagen composition of soft tissue changes from an elastic type III collagen to a stiffer type I collagen.¹² These physiologic changes may influence the rigidity of the soft tissue restraints around the patellofemoral joint during periods of rapid growth. Longitudinal growth and rotational changes can also occur at the distal femoral and proximal tibial physes. The position of the tibial tubercle and the magnitude of femoral anteversion may also change after growth in adolescents.

Developing effective technical analogs of surgical procedures performed in mature patients with PFI for use in skeletally immature patients has been a second challenge. For example, a varus-producing distal femoral osteotomy to address genu valgum and PFI¹³ is contraindicated in immature patients, when a hemiepiphysiodesis for guided growth should be considered.¹⁴ Similarly, a periosteal sleeve medialization of the insertion of the patellar tendon may be used instead of a tibial tubercle transfer.¹⁵

If a medial patellofemoral ligament (MPFL) reconstruction is considered in immature patients, careful consideration of the position of the MPFL in relation to the distal femoral physis is paramount. Shea and colleagues¹⁶ originally described the position of the MPFL, based on Schottle’s point¹⁷ on lateral radiographs, to be proximal to the distal femoral physis (Figure 2). However, due to the undulation of the physis, the lateral projection may falsely demonstrate Schottle’s point to be on or just proximal to the physis. Other researchers have evaluated the position of the distal femoral physis to be proximal to the origin of the MPFL by a range of 2.9 mm to 8.5 mm on AP radiograph and MRI, respectively.^18,19 More recent cadaveric studies have demonstrated the origin of the MPFL in pediatric specimens in relation to the physis to be variable.²⁰ Although we believe that a relative anatomic femoral position of the MPFL origin can be achieved without disrupting the physis using Schottle’s point, some have indicated concerns that this may produce a nonisometric position, which has not been our experience.²¹

Continue to: TREATMENT OPTIONS...

The evaluation of a skeletally immature patient may differ from that of an adult patient.  (Table 1) represents a standard evaluation of a skeletally immature patient with PFI. The injury, physical examination findings, activity level, presence of an osteochondral fracture, and severity of bony dysplasia may indicate surgical treatment and influence selection of the surgical technique. Indications for surgical treatment include recurrent patellar instability, symptomatic patellar instability, the presence of a chondral or an osteochondral fracture, and a primary patella dislocation in patients with a high risk for recurrent dislocations. Table 2 represents a list of the possible surgical treatment options for the skeletally immature patient with PFI.^21,22 Variable outcomes and recurrence rates are noted with each procedure.^21,22

MPFL REPAIR VS RECONSTRUCTION

Although there is increased consensus for restoring normal anatomy, continued controversy exists regarding the utility of an MPFL repair versus reconstruction. An MPFL repair to the patella has been reported in a large series and was noted to have a 72% success rate.²³ Although a perceived benefit of the procedure is the absence of risk to the physis, concerns exist regarding the unacceptable rate of continued patellar instability after a repair of native tissue.

Several studies have demonstrated that reconstruction of the MPFL yields lower rates of recurrent dislocation, improved knee function, and an ability to return to prior levels of activity with little to no functional impairment.^1,2,24-26 Studies have also shown that MPFL reconstruction can be performed safely in skeletally immature patients, with little risk of growth impairment, good to excellent results, and low re-rupture rates.^27,28

We propose that MPFL reconstruction for the skeletally immature patient with PFI should be the primary surgical treatment. Any patient with atypical or severe bony dysplasia may warrant additional intervention. Additional correction of valgus, in the form of guided growth, should be considered in conjunction with an MPFL reconstruction if the patient demonstrates a valgus of grade >2 or a lateral distal femoral angle of <84°.

GRAFT OPTIONS

When an MPFL reconstruction is indicated, graft options include autologous hamstring, quadriceps, adductor, or patellar tendon grafts. Allograft tendon may be an acceptable choice, and use of synthetic graft has also been described. A recent systematic review²⁹ concluded that there was no difference in recurrence rates or outcomes based on graft type. However, in this study, the overall complication rates were higher in patients who had autologous graft than in those who had allograft. Although the use of allograft has not been specifically reported in a pediatric cohort, allografts have been successfully used in this age group. Obviating graft harvest eliminates an additional or extended incision, limits postoperative weakness, and may speed early recovery.

Continue to: PREFERRED SURGICAL TECHNIQUE FOR MPFL RECONSTRUCTION WITH ALLOGRAFT...

Surgery should be scheduled and performed as an outpatient procedure. Preoperative instructions may include crutch training, when indicated, and postoperative education regarding expected early therapy and pain management strategy may be employed. Chlorhexidine wipes are provided to patients preoperatively with instruction to use daily starting 3 days prior to surgery. A home exercise program with a focus on quadriceps activation and range of motion is provided and requested to be performed postoperatively.

A standard operating room setup for knee arthroscopy is employed (preference list noted in Table 3; items bolded may be unique to an MPFL reconstruction). Regional, single-shot anesthesia is employed using a sensory-only adductor canal block. A universal surgical timeout is performed before beginning the surgery to include verification of prophylactic antibiotics (Ancef or clindamycin in penicillin-allergic patients) and consideration of the use of a sequential compression device in children aged >10 years when ≥1 venous thromboembolic risk factors are detected in preoperative screening.

Nonirradiated allografts (semitendinosus or tibialis anterior), preferably from a donor <30 years, are our preferred graft choice. The minimum length of the graft should be 240 mm and the doubled thickness should be 5 to 6 mm. After thawing, the graft is lavaged with a mixture of antibiotic saline consisting of 50,000 U of bacitracin in 1 l of normal saline. Tension is then placed on the graft using the graft preparation board, and a whip stitch is placed on each side of the graft using a #2 FiberWire (Arthrex). The graft is sized with care to ensure that it is not thicker than 6.5 mm with an optimal goal of 6 mm.

A standard knee arthroscopy is performed with an emphasis on evaluating the patellofemoral anatomy. Although insufflation can cause distortion and, often, lateralize the patella, the surgeon should consider an estimate of when the patella engages the trochlea during knee flexion. This position of knee flexion will determine the appropriate position of the knee during graft fixation. A bipolar device (90°Arthrocare wand) may be used to debride synovial folds (ie, plica) when affecting the patellar tracking. To maximize visualization of the patella within the knee, we recommend switching the arthroscope from the anterolateral portal to the anteromedial portal. This allows for improved visualization of the lateral femoral condyle and the inferomedial patellar facet, common locations for chondral and osteochondral damage, as well as optimal visualization of patellar engagement within the trochlea. During an arthroscopic dynamic exam of the patella tracking, a lateralized patella may be observed. If the patella tilts upon manual medial translation toward the trochlear groove during the dynamic exam, the lateral retinaculum may be tight or constraining the patella laterally. If this occurs, a partial or complete lateral release may be indicated with a bipolar wand. Visualization of the posteromedial and the posterolateral compartments is required as loose chondral bodies may be present in these locations.     

Any osteochondral or chondral injuries are addressed during the arthroscopy. Large osteochondral fractures of the lateral femoral condyle or the patella are repaired using headless compression screws (Acutrak Screws, Acumed) or headed low-profile screws (1.5-mm screws, DePuy Synthes). Headless screws can be buried 2 to 3 mm below the chondral surface when the fragment has an appropriate thickness, whereas thin fragments may necessitate headed screws for adequate fixation and subsequent implant removal in 8 to 10 weeks. Defects in the patella most often require an open arthrotomy to evert the patella 90° to allow adequate exposure for treatment. Chronic chondral fragmentation may be debrided using the motorized chondrotome or removed and indicated for microfracture (<2 cm² in surface area and <4 mm in depth) or other types of chondral replacement. Chondral lesions over the inferior medial facet are typically not symptomatic and often require minimal treatment.

Continue to: There are 3 methods to identify...

There are 3 methods to identify the insertion of the MPFL into the patella. During the diagnostic arthroscopy, an 18-gauge needle can be used to mark the insertion of the MPFL as visualization of the ligament arthroscopically is often possible (Figure 3). Another useful technique is to follow the inferior aspect of the distal insertion of the vastus medialis oblique (VMO) into the patella. The typical insertion point of the MPFL is immediately distal to the insertion of the inferior aspect of the VMO (Figure 4). It is also helpful to note that the center point of the insertion of the MPFL is at the junction of the proximal one-third and distal two-thirds of the palpable osseous patella. The MPFL origin has been noted to be at the exact midpoint of the chondral surface of the patella or 5 mm proximal (41% of the length of the patella) to the midpoint of the osseous patella.³⁰

Following arthroscopic examination and treatment, a linear incision is made at the superior two-thirds of the patella, 1-finger breadth medial to the patella (Figure 5). During the subcutaneous dissection, the goal is to visualize the fascia overlying the VMO. Once this is identified, the dissection, in this layer, is carried over to expose the anterior and central surface of the patella. Army/navy retractors are used to retract the skin, and the assistant will place manual pressure on the patella to stabilize it for preparation of the patellar surface.

A bovie cautery or a knife is used to mark the insertion of the MPFL, which is immediately distal to the inferior border of the VMO. An incision over the medial surface of the patella creates a T-incision with elevation of the subsequent flaps proximal and distal. This allows exposure of the superficial and medial surface of the patella. During medial exposure of the patella, care is taken to avoid an arthrotomy by leaving the synovial lining attachment. A rongeur is used to decorticate the medial patella and the superficial surface (Figure 6). If an MPFL avulsion is present, it is often embedded within the soft tissues adjacent to the medial patellar. The MPFL avulsion can be exposed and removed if small. During an MPFL reconstruction, a repair of the avulsion is typically not performed. A double-loaded suture anchor is inserted at the site identified as the insertion of the native MPFL (3.0 Biocomposite SutureTak, Arthrex). A single suture anchor is used instead of an interosseous tunnel or double tunnels to avoid creating a large defect that may increase the risk of fracture in a small, skeletally immature patella.³¹

A hemostat is used to identify the layer between the medial retinaculum and the synovium over the medial soft tissue of the knee. The MPFL is a well-defined thickening of the medial retinaculum, and the ideal placement of the reconstruction is immediately inferior to this layer. Once this layer has been identified, a blunt hemostat is inserted to mark the end of a blind pouch that is apparent in this layer. This blind pouch is marked on the surface of the skin as the origin of the MPFL.

Fluoroscopy is now used to identify the origin of the MPFL on the medial femoral condyle. A perfect lateral is obtained by lining up the posterior condyle. Often, existent trochlear dysplasia will modify the normal appearance of the anterior structures of the condyles. Schottle’s point¹⁷ is USED to locate the origin of the MPFL. This is defined by drawing a line from the posterior cortex distally through the condyles (Figure 7). Two perpendicular lines are drawn, one to the extension of the posterior cortex at the level of the posterior extent of Blumensaat’s line and a second at the metaphyseal flare. Schottle’s point is located just anterior to the posterior cortex and in between these perpendicular lines. In a skeletally immature patient, this point appears to be at the level of the physis, on the lateral projection.

Continue to: An anteroposterior and/or a notch view projection...

An anteroposterior and/or a notch view projection is then performed to confirm the location of Schottle’s point, distal to the physis. Because the origin of the MPFL is posterior to the condyle, the notch view should be used to visualize Schottle’s point in relation to the physis in the coronal plane. This location has been confirmed to approximate the origin of the MPFL by published anatomic studies. If the graft measures 6 mm, an ideal distance of 5 to 8 mm distal to the physis will ensure a tunnel location with a proximal margin 2 to 4 mm distal to the physis. In our experience, Schottle’s point is approximately 5 to 7 mm distal to the physis, and rarely do we adjust our position. The guide wire is placed on the medial femoral condyle at this point and then angled 30° proximal to distal to allow tunnel trajectory distal to the undulating physis (Figure 8).

A 6-mm reamer is used over the guide wire to a depth of 20 mm. The newly created tunnel is now exposed and visualized. The graft is folded over at the center aspect and marked 20 mm from the tip of a Bio-tenodesis screwdriver (Arthrex). A 6.25 × 15 mm Bio-tenodesis screw is used. The graft and the screw are inserted into the condyle, and the appropriate graft fixation is confirmed by longitudinal tension placed upon the graft.

The hemostat or the passing suture previously placed into the blind pouch below the native MPFL is used to pass the graft immediately superficial to the synovial lining, in an extra-articular location. The ends of the graft are now exposed through the incision adjacent to the patella. The knee is confirmed to be flexed to approximately 45° on the surgical positioning triangleor at the flexion position determined during diagnostic arthroscopy at which the patella engages the trochlea. The graft is set to length alongside the medial surface of the patella at the position of the suture anchor, and a single stitch is placed through both ends of the graft at this adjacent position using a free needle (Davis Tonsil ½ Circle Taper Point; Vessel Medical). A hemostat is placed on the single suture throw to provisionally secure the graft in this trial position. This allows the surgeon to trial and examine the MPFL graft in extension, 45° of flexion, and terminal flexion. The goal is to provide a check rein to lateral translation of the patella and provide a firm endpoint to avoid further dislocation without overtension of the graft that may lead to increased contact pressures in the patellofemoral joint.

Once appropriate graft position is confirmed, using the double-loaded suture, we prefer to secure each limb separately to the decorticated medial patella. One end of the suture is threaded through the graft once and will act as a post. The other end of the suture is threaded through the graft 3 times in a modified Mason-Allen stitch. With the knee in that same position of flexion, the knot is tied and the graft is secured to the medial side of the patella. For secondary fixation, the remaining ends of the graft are passed below the periosteum on the anterior surface of the patella. Using a 0 Vicryl (Ethicon US), the periosteal flaps are sutured on top of the graft, incorporating the residual graft. The ends are cut and a repeat dynamic examination is performed to confirm the position of the patella and the patellar tracking and to ensure that overtensioning did not occur. Following irrigation, a standard closure is performed. The patient is placed in a hinged knee brace and cryotherapy is applied (Polar Care, Breg Inc).

POSTOPERATIVE PROTOCOL

For the initial or the acute postoperative phases, an emphasis on edema control, early quadriceps activation, and range of motion is recommended. We recommend weight-bearing as tolerated with the leg locked in extension until adequate quadriceps control is achieved. The patient must be able to perform 10 straight leg raises without an extension loss to be cleared to weight-bear as tolerated without motion restriction in the brace. Full motion is allowed immediately.

OUTCOMES

In our experience using isolated MPFL reconstruction in the skeletally immature patient, we have had no evidence of physeal arrest, leg-length inequality, or angular deformity, and only11.4% of patients have had recurrent instability. The mean Kujala score in this cohort was 90.4, with a mean Tegner activity of 7, after the procedure. All failures in our cohort had had severe trochlear dysplasia.

Anatomic factors that contribute to PFI include ligament laxity, trochlear dysplasia, patella alta, excessive femoral anteversion or tibial torsion, genu valgum, and increased tibial tubercle-to-trochlear groove distance.^1,3,8-11 When considering surgical treatment for PFI, all anatomic factors should be considered, with an emphasis on, and understanding of, the role of residual growth and development. One must also consider balancing static soft tissue, dynamic soft tissue, and osseous constraints of the patellofemoral joint to optimize the overall health and balance of the patella (Figure 1).

CHALLENGES IN TREATING PEDIATRIC PFI

A primary challenge in the treatment of PFI in pediatric patients is accounting for the impact of anatomic changes occurring secondary to growth and maturity. From birth to adulthood, the collagen composition of soft tissue changes from an elastic type III collagen to a stiffer type I collagen.¹² These physiologic changes may influence the rigidity of the soft tissue restraints around the patellofemoral joint during periods of rapid growth. Longitudinal growth and rotational changes can also occur at the distal femoral and proximal tibial physes. The position of the tibial tubercle and the magnitude of femoral anteversion may also change after growth in adolescents.

Developing effective technical analogs of surgical procedures performed in mature patients with PFI for use in skeletally immature patients has been a second challenge. For example, a varus-producing distal femoral osteotomy to address genu valgum and PFI¹³ is contraindicated in immature patients, when a hemiepiphysiodesis for guided growth should be considered.¹⁴ Similarly, a periosteal sleeve medialization of the insertion of the patellar tendon may be used instead of a tibial tubercle transfer.¹⁵

If a medial patellofemoral ligament (MPFL) reconstruction is considered in immature patients, careful consideration of the position of the MPFL in relation to the distal femoral physis is paramount. Shea and colleagues¹⁶ originally described the position of the MPFL, based on Schottle’s point¹⁷ on lateral radiographs, to be proximal to the distal femoral physis (Figure 2). However, due to the undulation of the physis, the lateral projection may falsely demonstrate Schottle’s point to be on or just proximal to the physis. Other researchers have evaluated the position of the distal femoral physis to be proximal to the origin of the MPFL by a range of 2.9 mm to 8.5 mm on AP radiograph and MRI, respectively.^18,19 More recent cadaveric studies have demonstrated the origin of the MPFL in pediatric specimens in relation to the physis to be variable.²⁰ Although we believe that a relative anatomic femoral position of the MPFL origin can be achieved without disrupting the physis using Schottle’s point, some have indicated concerns that this may produce a nonisometric position, which has not been our experience.²¹