Original Research

The electronic health record (EHR) has revolutionized the practice of medicine. As part of the economic stimulus package in 2009, Congress enacted the Health Information Technology for Economic and Clinical Health Act, which included incentives for physicians and hospitals to adopt an EHR by 2015. In the setting of more limited duty hours and demands for increased clinical productivity, EHRs have functions that may improve the quality and efficiency of clinical documentation.[1, 2, 3, 4, 5]

The process of note writing and the use of notes for clinical care have changed substantially with EHR implementation. Use of efficiency tools (ie, copy forward functions and autopopulation of data) may increase the speed of documentation.[5] Notes in an EHR are more legible and accessible and may be able to organize data to improve clinical care.[6]

Yet, many have commented on the negative consequences of documentation in an EHR. In a New England Journal of Medicine Perspective article, Drs. Hartzband and Groopman wrote, we have observed the electronic medical record become a powerful vehicle for perpetuating erroneous information, leading to diagnostic errors that gain momentum when passed on electronically.[7] As a result, the copy forward and autopopulation functions have come under significant scrutiny.[8, 9, 10] A survey conducted at 2 academic institutions found that 71% of residents and attendings believed that the copy forward function led to inconsistencies and outdated information.[11] Autopopulation has been criticized for creating lengthy notes full of trivial or redundant data, a phenomenon termed note bloat. Bloated notes may be less effective as a communication tool.[12] Additionally, the process of composing a note often stimulates critical thinking and may lead to changes in care. The act of copying forward a previous note and autopopulating data bypasses that process and in effect may suppress critical thinking.[13] Previous studies have raised numerous concerns regarding copy forward and autopopulation functionality in the EHR. Many have described the duplication of outdated data and the possibility of the introduction and perpetuation of errors.[14, 15, 16] The Veterans Affairs (VA) Puget Sound Health system evaluated 6322 copy events and found that 1 in 10 electronic patient charts contained an instance of high‐risk copying.[17] In a survey of faculty and residents at a single academic medical center, the majority of users of copy and paste functionality recognized the hazards; they responded that their notes may contain more outdated (66%) and more inconsistent information (69%). Yet, most felt copy forwarding improved the documentation of the entire hospital course (87%), overall physician documentation (69%), and should definitely be continued (91%).[11] Others have complained about the impact of copy forward on the expression of clinical reasoning.[7, 9, 18]

Previous discussions on the topic of overall note quality following EHR implementation have been limited to perspectives or opinion pieces of individual attending providers.[18] We conducted a survey across 4 academic institutions to analyze both housestaff and attendings perceptions of the quality of notes since the implementation of an EHR to better inform the discussion of the impact of an EHR on note quality.

METHODS

RESULTS

Overall Assessment of Note Quality

Participants were asked to rate the quality of progress notes on a 5‐point scale (poor, fair, good, very good, excellent). Half of interns and residents rated their own progress notes as very good or excellent. A total of 44% percent of interns and 24% of residents rated their peers notes as very good or excellent, whereas only 15% of attending physicians rated housestaff notes as very good or excellent.

When asked to rate the change in progress note quality since their hospital had adopted the EHR, the majority of residents answered unchanged or better, and the majority of attendings answered unchanged or worse (Figure 1).

Figure 1

PDQI‐9 Framework

Participants answered each PDQI‐9 question on a 5‐point Likert scale ranging from not at all (1) to extremely (5). In 8 of the 9 PDQI‐9 domains, there were no significant differences between interns and residents. Across each domain, attending perceptions of housestaff notes were significantly lower than housestaff perceptions of their own notes (P<0.001) (Figure 2). Both housestaff and attendings gave the highest ratings to thorough, up to date, and synthesized and the lowest rating to succinct.

Figure 2

Copy Forward and Autopopulation

Overall, the effect of copy forward and autopopulation on critical thinking, note accuracy, and prioritizing the problem list was thought to be neutral or somewhat positive by interns, neutral by residents, and neutral or somewhat negative by attendings (P<0.001) (Figure 3). In all, 16% of interns, 22% of residents, and 55% of attendings reported that copy forward had a somewhat negative or very negative impact on critical thinking (P<0.001). In all, 16% of interns, 29% of residents and 39% of attendings thought that autopopulation had a somewhat negative or very negative impact on critical thinking (P<0.001).

Figure 3

DISCUSSION

This is the first large multicenter analysis of both attendings and housestaff perceptions of note quality in the EHR era. The findings provide insight into important differences and similarities in the perceptions of the 2 groups. Most striking is the difference in opinion of overall note quality, with only a small minority of faculty rating current housestaff notes as very good or excellent, whereas a much larger proportion of housestaff rated their own notes and those of their peers to be of high quality. Though participants were not specifically asked why note quality in general was suboptimal, housestaff and faculty rankings of specific domains from the PDQI‐9 may yield an important clue. Specifically, all groups expressed that the weakest attribute of current progress notes is succinct. This finding is consistent with the note bloat phenomenon, which has been maligned as a consequence of EHR implementation.[7, 14, 18, 21, 22]

One interesting finding was that only 5% of interns rated the notes of other housestaff as fair or poor. One possible explanation for this may be the tendency for an individual to enhance or augment the status or performance of the group to which he or she belongs as a mechanism to increase self‐image, known as the social identity theory.[23] Thus, housestaff may not criticize their peers to allow for identification with a group that is not deficient in note writing.

The more positive assessment of overall note quality among housestaff could be related to the different roles of housestaff and attendings on a teaching service. On a teaching service, housestaff are typically the writer, whereas attendings are almost exclusively the reader of progress notes. Housestaff may reap benefits, including efficiency, beyond the finished product. A perception of higher quality may reflect the process of note writing, data gathering, and critical thinking required to build an assessment and plan. The scores on the PDQI‐9 support this notion, as housestaff rated all 9 domains significantly higher than attendings.

Housestaff and attendings held greater differences of opinion with respect to the EHR's impact on note quality. Generally, housestaff perceived the EHR to have improved progress note quality, whereas attendings perceived the opposite. One explanation could be that these results reflect changing stages of development of physicians well described through the RIME framework (reporter, interpreter, manager, educator). Attendings may expect notes to reflect synthesis and analysis, whereas trainees may be satisfied with the data gathering that an EHR facilitates. In our survey, the trend of answers from intern to resident to attending suggests an evolving process of attitudes toward note quality.

The above reasons may also explain why housestaff were generally more positive than attendings about the effect of copy forward and autopopulation functions on critical thinking. Perhaps, as these functions can potentially increase efficiency and decrease time spent at the computer, although data are mixed on this finding, housestaff may have more time to spend with patients or develop a thorough plan and thus rate these functions positively.

Notably, housestaff and attendings had excellent agreement on the purposes of a progress note. They agreed that the 2 most important purposes were communication with other providers and documenting important events and the plan for the day. These are the 2 listed purposes that are most directly related to patient care. If future interventions to improve note quality require housestaff and attendings to significantly change their behavior, a focus on the impact on patient care might yield the best results.

There were several limitations in our study. Any study based on self‐assessment is subject to bias. A previous meta‐analysis and review described poor to moderate correlations between self‐assessed and external measures of performance.[24, 25] The survey data were aggregated from 4 institutions despite somewhat different, though relatively high, response rates between the institutions. There could be a response bias; those who did not respond may have systematically different perceptions of note quality. It should be noted that the general demographics of the respondents reflected those of the housestaff and attendings at 4 academic centers. All 4 of the participating institutions adopted the Epic EHR within the last several years of the survey being administered, and perceptions of note quality may be biased depending on the prior system used (ie, change from handwritten to electronic vs electronic to other electronic system). In addition, the survey results reflect experience with only 1 EHR, and our results may not apply to other EHR vendors or institutions like the VA, which have a long‐standing system in place. Last, we did not explore the impact of perceived note quality on the measured or perceived quality of care. One previous study found no direct correlation between note quality and clinical quality.[26]

There are several future directions for research based on our findings. First, potential differences between housestaff and attending perceptions of note quality could be further teased apart by studying the perceptions of attendings on a nonteaching service who write their own daily progress notes. Second, housestaff perceptions on why copy forward and autopopulation may increase critical thinking could be explored further with more direct questioning. Finally, although our study captured only perceptions of note quality, validated tools could be used to objectively measure note quality; these measurements could then be compared to perception of note quality as well as clinical outcomes.

Given the prevalence and the apparent belief that the benefits of an EHR outweigh the hazards, institutions should embrace these innovations but take steps to mitigate the potential errors and problems associated with copy forward and autopopulation. The results of our study should help inform future interventions.

The electronic health record (EHR) has revolutionized the practice of medicine. As part of the economic stimulus package in 2009, Congress enacted the Health Information Technology for Economic and Clinical Health Act, which included incentives for physicians and hospitals to adopt an EHR by 2015. In the setting of more limited duty hours and demands for increased clinical productivity, EHRs have functions that may improve the quality and efficiency of clinical documentation.[1, 2, 3, 4, 5]

The process of note writing and the use of notes for clinical care have changed substantially with EHR implementation. Use of efficiency tools (ie, copy forward functions and autopopulation of data) may increase the speed of documentation.[5] Notes in an EHR are more legible and accessible and may be able to organize data to improve clinical care.[6]

Yet, many have commented on the negative consequences of documentation in an EHR. In a New England Journal of Medicine Perspective article, Drs. Hartzband and Groopman wrote, we have observed the electronic medical record become a powerful vehicle for perpetuating erroneous information, leading to diagnostic errors that gain momentum when passed on electronically.[7] As a result, the copy forward and autopopulation functions have come under significant scrutiny.[8, 9, 10] A survey conducted at 2 academic institutions found that 71% of residents and attendings believed that the copy forward function led to inconsistencies and outdated information.[11] Autopopulation has been criticized for creating lengthy notes full of trivial or redundant data, a phenomenon termed note bloat. Bloated notes may be less effective as a communication tool.[12] Additionally, the process of composing a note often stimulates critical thinking and may lead to changes in care. The act of copying forward a previous note and autopopulating data bypasses that process and in effect may suppress critical thinking.[13] Previous studies have raised numerous concerns regarding copy forward and autopopulation functionality in the EHR. Many have described the duplication of outdated data and the possibility of the introduction and perpetuation of errors.[14, 15, 16] The Veterans Affairs (VA) Puget Sound Health system evaluated 6322 copy events and found that 1 in 10 electronic patient charts contained an instance of high‐risk copying.[17] In a survey of faculty and residents at a single academic medical center, the majority of users of copy and paste functionality recognized the hazards; they responded that their notes may contain more outdated (66%) and more inconsistent information (69%). Yet, most felt copy forwarding improved the documentation of the entire hospital course (87%), overall physician documentation (69%), and should definitely be continued (91%).[11] Others have complained about the impact of copy forward on the expression of clinical reasoning.[7, 9, 18]

Previous discussions on the topic of overall note quality following EHR implementation have been limited to perspectives or opinion pieces of individual attending providers.[18] We conducted a survey across 4 academic institutions to analyze both housestaff and attendings perceptions of the quality of notes since the implementation of an EHR to better inform the discussion of the impact of an EHR on note quality.

METHODS

RESULTS

Overall Assessment of Note Quality

Participants were asked to rate the quality of progress notes on a 5‐point scale (poor, fair, good, very good, excellent). Half of interns and residents rated their own progress notes as very good or excellent. A total of 44% percent of interns and 24% of residents rated their peers notes as very good or excellent, whereas only 15% of attending physicians rated housestaff notes as very good or excellent.

When asked to rate the change in progress note quality since their hospital had adopted the EHR, the majority of residents answered unchanged or better, and the majority of attendings answered unchanged or worse (Figure 1).

Figure 1

PDQI‐9 Framework

Participants answered each PDQI‐9 question on a 5‐point Likert scale ranging from not at all (1) to extremely (5). In 8 of the 9 PDQI‐9 domains, there were no significant differences between interns and residents. Across each domain, attending perceptions of housestaff notes were significantly lower than housestaff perceptions of their own notes (P<0.001) (Figure 2). Both housestaff and attendings gave the highest ratings to thorough, up to date, and synthesized and the lowest rating to succinct.

Figure 2

Copy Forward and Autopopulation

Overall, the effect of copy forward and autopopulation on critical thinking, note accuracy, and prioritizing the problem list was thought to be neutral or somewhat positive by interns, neutral by residents, and neutral or somewhat negative by attendings (P<0.001) (Figure 3). In all, 16% of interns, 22% of residents, and 55% of attendings reported that copy forward had a somewhat negative or very negative impact on critical thinking (P<0.001). In all, 16% of interns, 29% of residents and 39% of attendings thought that autopopulation had a somewhat negative or very negative impact on critical thinking (P<0.001).

Figure 3

DISCUSSION

This is the first large multicenter analysis of both attendings and housestaff perceptions of note quality in the EHR era. The findings provide insight into important differences and similarities in the perceptions of the 2 groups. Most striking is the difference in opinion of overall note quality, with only a small minority of faculty rating current housestaff notes as very good or excellent, whereas a much larger proportion of housestaff rated their own notes and those of their peers to be of high quality. Though participants were not specifically asked why note quality in general was suboptimal, housestaff and faculty rankings of specific domains from the PDQI‐9 may yield an important clue. Specifically, all groups expressed that the weakest attribute of current progress notes is succinct. This finding is consistent with the note bloat phenomenon, which has been maligned as a consequence of EHR implementation.[7, 14, 18, 21, 22]

One interesting finding was that only 5% of interns rated the notes of other housestaff as fair or poor. One possible explanation for this may be the tendency for an individual to enhance or augment the status or performance of the group to which he or she belongs as a mechanism to increase self‐image, known as the social identity theory.[23] Thus, housestaff may not criticize their peers to allow for identification with a group that is not deficient in note writing.

The more positive assessment of overall note quality among housestaff could be related to the different roles of housestaff and attendings on a teaching service. On a teaching service, housestaff are typically the writer, whereas attendings are almost exclusively the reader of progress notes. Housestaff may reap benefits, including efficiency, beyond the finished product. A perception of higher quality may reflect the process of note writing, data gathering, and critical thinking required to build an assessment and plan. The scores on the PDQI‐9 support this notion, as housestaff rated all 9 domains significantly higher than attendings.

Housestaff and attendings held greater differences of opinion with respect to the EHR's impact on note quality. Generally, housestaff perceived the EHR to have improved progress note quality, whereas attendings perceived the opposite. One explanation could be that these results reflect changing stages of development of physicians well described through the RIME framework (reporter, interpreter, manager, educator). Attendings may expect notes to reflect synthesis and analysis, whereas trainees may be satisfied with the data gathering that an EHR facilitates. In our survey, the trend of answers from intern to resident to attending suggests an evolving process of attitudes toward note quality.

The above reasons may also explain why housestaff were generally more positive than attendings about the effect of copy forward and autopopulation functions on critical thinking. Perhaps, as these functions can potentially increase efficiency and decrease time spent at the computer, although data are mixed on this finding, housestaff may have more time to spend with patients or develop a thorough plan and thus rate these functions positively.

Notably, housestaff and attendings had excellent agreement on the purposes of a progress note. They agreed that the 2 most important purposes were communication with other providers and documenting important events and the plan for the day. These are the 2 listed purposes that are most directly related to patient care. If future interventions to improve note quality require housestaff and attendings to significantly change their behavior, a focus on the impact on patient care might yield the best results.

There were several limitations in our study. Any study based on self‐assessment is subject to bias. A previous meta‐analysis and review described poor to moderate correlations between self‐assessed and external measures of performance.[24, 25] The survey data were aggregated from 4 institutions despite somewhat different, though relatively high, response rates between the institutions. There could be a response bias; those who did not respond may have systematically different perceptions of note quality. It should be noted that the general demographics of the respondents reflected those of the housestaff and attendings at 4 academic centers. All 4 of the participating institutions adopted the Epic EHR within the last several years of the survey being administered, and perceptions of note quality may be biased depending on the prior system used (ie, change from handwritten to electronic vs electronic to other electronic system). In addition, the survey results reflect experience with only 1 EHR, and our results may not apply to other EHR vendors or institutions like the VA, which have a long‐standing system in place. Last, we did not explore the impact of perceived note quality on the measured or perceived quality of care. One previous study found no direct correlation between note quality and clinical quality.[26]

There are several future directions for research based on our findings. First, potential differences between housestaff and attending perceptions of note quality could be further teased apart by studying the perceptions of attendings on a nonteaching service who write their own daily progress notes. Second, housestaff perceptions on why copy forward and autopopulation may increase critical thinking could be explored further with more direct questioning. Finally, although our study captured only perceptions of note quality, validated tools could be used to objectively measure note quality; these measurements could then be compared to perception of note quality as well as clinical outcomes.

Given the prevalence and the apparent belief that the benefits of an EHR outweigh the hazards, institutions should embrace these innovations but take steps to mitigate the potential errors and problems associated with copy forward and autopopulation. The results of our study should help inform future interventions.

The electronic health record (EHR) has revolutionized the practice of medicine. As part of the economic stimulus package in 2009, Congress enacted the Health Information Technology for Economic and Clinical Health Act, which included incentives for physicians and hospitals to adopt an EHR by 2015. In the setting of more limited duty hours and demands for increased clinical productivity, EHRs have functions that may improve the quality and efficiency of clinical documentation.[1, 2, 3, 4, 5]

The process of note writing and the use of notes for clinical care have changed substantially with EHR implementation. Use of efficiency tools (ie, copy forward functions and autopopulation of data) may increase the speed of documentation.[5] Notes in an EHR are more legible and accessible and may be able to organize data to improve clinical care.[6]

Yet, many have commented on the negative consequences of documentation in an EHR. In a New England Journal of Medicine Perspective article, Drs. Hartzband and Groopman wrote, we have observed the electronic medical record become a powerful vehicle for perpetuating erroneous information, leading to diagnostic errors that gain momentum when passed on electronically.[7] As a result, the copy forward and autopopulation functions have come under significant scrutiny.[8, 9, 10] A survey conducted at 2 academic institutions found that 71% of residents and attendings believed that the copy forward function led to inconsistencies and outdated information.[11] Autopopulation has been criticized for creating lengthy notes full of trivial or redundant data, a phenomenon termed note bloat. Bloated notes may be less effective as a communication tool.[12] Additionally, the process of composing a note often stimulates critical thinking and may lead to changes in care. The act of copying forward a previous note and autopopulating data bypasses that process and in effect may suppress critical thinking.[13] Previous studies have raised numerous concerns regarding copy forward and autopopulation functionality in the EHR. Many have described the duplication of outdated data and the possibility of the introduction and perpetuation of errors.[14, 15, 16] The Veterans Affairs (VA) Puget Sound Health system evaluated 6322 copy events and found that 1 in 10 electronic patient charts contained an instance of high‐risk copying.[17] In a survey of faculty and residents at a single academic medical center, the majority of users of copy and paste functionality recognized the hazards; they responded that their notes may contain more outdated (66%) and more inconsistent information (69%). Yet, most felt copy forwarding improved the documentation of the entire hospital course (87%), overall physician documentation (69%), and should definitely be continued (91%).[11] Others have complained about the impact of copy forward on the expression of clinical reasoning.[7, 9, 18]

Previous discussions on the topic of overall note quality following EHR implementation have been limited to perspectives or opinion pieces of individual attending providers.[18] We conducted a survey across 4 academic institutions to analyze both housestaff and attendings perceptions of the quality of notes since the implementation of an EHR to better inform the discussion of the impact of an EHR on note quality.

METHODS

RESULTS

Overall Assessment of Note Quality

Participants were asked to rate the quality of progress notes on a 5‐point scale (poor, fair, good, very good, excellent). Half of interns and residents rated their own progress notes as very good or excellent. A total of 44% percent of interns and 24% of residents rated their peers notes as very good or excellent, whereas only 15% of attending physicians rated housestaff notes as very good or excellent.

When asked to rate the change in progress note quality since their hospital had adopted the EHR, the majority of residents answered unchanged or better, and the majority of attendings answered unchanged or worse (Figure 1).

Figure 1

PDQI‐9 Framework

Participants answered each PDQI‐9 question on a 5‐point Likert scale ranging from not at all (1) to extremely (5). In 8 of the 9 PDQI‐9 domains, there were no significant differences between interns and residents. Across each domain, attending perceptions of housestaff notes were significantly lower than housestaff perceptions of their own notes (P<0.001) (Figure 2). Both housestaff and attendings gave the highest ratings to thorough, up to date, and synthesized and the lowest rating to succinct.

Figure 2

Copy Forward and Autopopulation

Overall, the effect of copy forward and autopopulation on critical thinking, note accuracy, and prioritizing the problem list was thought to be neutral or somewhat positive by interns, neutral by residents, and neutral or somewhat negative by attendings (P<0.001) (Figure 3). In all, 16% of interns, 22% of residents, and 55% of attendings reported that copy forward had a somewhat negative or very negative impact on critical thinking (P<0.001). In all, 16% of interns, 29% of residents and 39% of attendings thought that autopopulation had a somewhat negative or very negative impact on critical thinking (P<0.001).

Figure 3

DISCUSSION

This is the first large multicenter analysis of both attendings and housestaff perceptions of note quality in the EHR era. The findings provide insight into important differences and similarities in the perceptions of the 2 groups. Most striking is the difference in opinion of overall note quality, with only a small minority of faculty rating current housestaff notes as very good or excellent, whereas a much larger proportion of housestaff rated their own notes and those of their peers to be of high quality. Though participants were not specifically asked why note quality in general was suboptimal, housestaff and faculty rankings of specific domains from the PDQI‐9 may yield an important clue. Specifically, all groups expressed that the weakest attribute of current progress notes is succinct. This finding is consistent with the note bloat phenomenon, which has been maligned as a consequence of EHR implementation.[7, 14, 18, 21, 22]

One interesting finding was that only 5% of interns rated the notes of other housestaff as fair or poor. One possible explanation for this may be the tendency for an individual to enhance or augment the status or performance of the group to which he or she belongs as a mechanism to increase self‐image, known as the social identity theory.[23] Thus, housestaff may not criticize their peers to allow for identification with a group that is not deficient in note writing.

The more positive assessment of overall note quality among housestaff could be related to the different roles of housestaff and attendings on a teaching service. On a teaching service, housestaff are typically the writer, whereas attendings are almost exclusively the reader of progress notes. Housestaff may reap benefits, including efficiency, beyond the finished product. A perception of higher quality may reflect the process of note writing, data gathering, and critical thinking required to build an assessment and plan. The scores on the PDQI‐9 support this notion, as housestaff rated all 9 domains significantly higher than attendings.

Housestaff and attendings held greater differences of opinion with respect to the EHR's impact on note quality. Generally, housestaff perceived the EHR to have improved progress note quality, whereas attendings perceived the opposite. One explanation could be that these results reflect changing stages of development of physicians well described through the RIME framework (reporter, interpreter, manager, educator). Attendings may expect notes to reflect synthesis and analysis, whereas trainees may be satisfied with the data gathering that an EHR facilitates. In our survey, the trend of answers from intern to resident to attending suggests an evolving process of attitudes toward note quality.

The above reasons may also explain why housestaff were generally more positive than attendings about the effect of copy forward and autopopulation functions on critical thinking. Perhaps, as these functions can potentially increase efficiency and decrease time spent at the computer, although data are mixed on this finding, housestaff may have more time to spend with patients or develop a thorough plan and thus rate these functions positively.

Notably, housestaff and attendings had excellent agreement on the purposes of a progress note. They agreed that the 2 most important purposes were communication with other providers and documenting important events and the plan for the day. These are the 2 listed purposes that are most directly related to patient care. If future interventions to improve note quality require housestaff and attendings to significantly change their behavior, a focus on the impact on patient care might yield the best results.

There were several limitations in our study. Any study based on self‐assessment is subject to bias. A previous meta‐analysis and review described poor to moderate correlations between self‐assessed and external measures of performance.[24, 25] The survey data were aggregated from 4 institutions despite somewhat different, though relatively high, response rates between the institutions. There could be a response bias; those who did not respond may have systematically different perceptions of note quality. It should be noted that the general demographics of the respondents reflected those of the housestaff and attendings at 4 academic centers. All 4 of the participating institutions adopted the Epic EHR within the last several years of the survey being administered, and perceptions of note quality may be biased depending on the prior system used (ie, change from handwritten to electronic vs electronic to other electronic system). In addition, the survey results reflect experience with only 1 EHR, and our results may not apply to other EHR vendors or institutions like the VA, which have a long‐standing system in place. Last, we did not explore the impact of perceived note quality on the measured or perceived quality of care. One previous study found no direct correlation between note quality and clinical quality.[26]

There are several future directions for research based on our findings. First, potential differences between housestaff and attending perceptions of note quality could be further teased apart by studying the perceptions of attendings on a nonteaching service who write their own daily progress notes. Second, housestaff perceptions on why copy forward and autopopulation may increase critical thinking could be explored further with more direct questioning. Finally, although our study captured only perceptions of note quality, validated tools could be used to objectively measure note quality; these measurements could then be compared to perception of note quality as well as clinical outcomes.

Given the prevalence and the apparent belief that the benefits of an EHR outweigh the hazards, institutions should embrace these innovations but take steps to mitigate the potential errors and problems associated with copy forward and autopopulation. The results of our study should help inform future interventions.

Successfully treated human immunodeficiency virus (HIV)‐infected individuals in the United States currently have life expectancy and mortality rates that are similar to the general population.[1, 2, 3, 4] A large multinational study found that the excess mortality rate among HIV‐positive individuals decreased from 40.8 to 6.1 per 1000 person‐years from pre‐1995 to 2006.¹ This is largely due to improved access to comprehensive HIV care, in particular widespread antiretroviral (ART) use. However, the proportion of deaths that are not classically considered acquired immunodeficiency syndrome (AIDS)‐related such as liver disease, cardiovascular disease, and non‐AIDS malignancy has increased,[1, 5, 6, 7] particularly among patients with higher CD4 T‐cell counts.[5, 8] Additionally, despite overall decline in mortality, there is evidence of racial and gender differences, with increased mortality risk associated with female gender and black race.[9, 10]

In the current ART era, HIV care has shifted focus from inpatient to outpatient care, with more emphasis on chronic disease management. However, hospitalization rates among HIV‐positive persons remain higher than that of the general population.[11, 12] A cross‐sectional study of HIV‐infected persons in the United States estimated a hospitalization rate of 26.6 per 100 persons in 2009,[13] compared to a rate of 11.9 for the general population during the same year.[14] Possible reasons for higher hospitalization rates include complications of aging or other chronic comorbidities, and consequences of behavioral risk factors such tobacco use and substance abuse.

Characterizing deaths among inpatient HIV‐infected individuals in the ART era is important to developing targeted interventions to further reduce mortality. Prior studies examining in‐hospital deaths of HIV‐positive patients evaluated more limited time periods,[15, 16, 17, 18] and thus did not necessarily assess the full spectrum of changes in mortality that have occurred with the introduction of ART. Furthermore, these studies described causes of death, but did not consistently identify factors associated with non‐AIDS deaths. We examined the trends in in‐hospital deaths among HIV‐infected patients from 1995 to 2011 and identified contributing factors to mortality. As the HIV population is aging, we hypothesized that HIV‐infected patients are more likely to die from nonAIDS‐related death in the late ART era due to factors related to cardiovascular and liver disease, compared to the early ART era.

METHODS

The study was performed at YaleNew Haven Hospital, an urban tertiary care academic teaching hospital with 1008 beds and the state of Connecticut's largest ambulatory HIV clinic. Connecticut ranks seventh nationally (10/100,000) in HIV prevalence; New Haven is second among Connecticut cities in the number of people living with HIV/AIDS.[19] We reviewed all patients with an International Classification of Diseases, Ninth Revision (ICD‐9) code of HIV or AIDS (ICD‐9 codes V08 and 042) who died during hospitalization between January 1, 1995 to December 31, 2011. The Yale Human Investigation Committee granted ethical approval to conduct the study.

A standardized data collection tool was used to abstract demographic characteristics (ie, age, gender, and race), medical comorbidities (ie, diabetes, chronic kidney disease, chronic hepatitis B or C, liver cirrhosis, hypertension, coronary artery disease, congestive heart failure, chronic obstructive lung disease, alcohol and substance abuse), ART use (yes or no), HIV viral load (VL), CD4 cell count, and causes of death. Comorbidities were defined using the Coding of Death in HIV Project protocol, a multinational endeavor to standardize data collection in studies of HIV‐positive patients.[20] Chronic kidney disease included individuals with National Kidney Foundation stage I to V disease. Chronic hepatitis B or C infection was identified in patients who had serologic testing indicative of prior infection. Alcohol and substance abuse were identified when source documents mentioned any history of current alcohol or illicit drug abuse or dependence. ART use was defined as documentation of ART on admission or prescription during hospitalization. This included individuals who were on 2 or more ART agents. The last HIV VL and CD4 cell count available within 1 year and closest to death were recorded. HIV VL suppression was defined as <400 copies/mL.

Two clinicians independently classified the cause of death as AIDS related or non‐AIDS related in accordance with published definitions.[21, 22] Cause of death was determined by review of the medical record, discharge diagnosis, and autopsy report when available. Official death certificates were not available for review. There was discordance in assigning 23 of the 400 causes of death. In these cases, the medical record was reviewed and determined by consensus between the 2 clinicians.

AIDS‐related deaths were categorized as nonspecified AIDS, AIDS infection, and AIDS malignancy. AIDS‐related deaths were defined as those caused by conditions meeting the Centers for Disease Control and Prevention AIDS case definition.[22] Non‐specified AIDS deaths were those occurring in patients with a CD4 count 50 cells/mm³ or with an AIDS‐defining illness, who died from a condition that was not clearly AIDS related. This included septic shock of unclear etiology, first known episode of pneumonia, a gastrointestinal bleed of unclear etiology, and altered mental status of unclear etiology when cerebrospinal fluid analysis or imaging of the brain was not available.

Non‐AIDS deaths included non‐AIDS infection in patients with a CD4 count >50 cells/mm³, cardiovascular disease, liver disease, non‐AIDS malignancy, and renal disease (Table 1). Deaths classified as other incorporated the deaths that did not fall into these categories. Chronic obstructive pulmonary disease (COPD) exacerbation and status asthmaticus were included in this category, because there was only 1 death from each of these causes.

The early ART era was defined as 1995 to 2001 and the late ART era from 2002 to 2011. During the early period, combination ART was introduced and significantly impacted overall mortality.[23, 24] The late ART era better reflected current in‐hospital deaths and was compared to the early era to evaluate trends over time.[15, 25]

² analysis and parametric (t test and analysis of variance) methods compared categorical and continuous variables, respectively. Bivariate analysis was used to determine associations with AIDS versus non‐AIDS deaths in the entire study cohort. Multivariable logistic regression was used to identify correlates of non‐AIDS deaths in the (1) complete 17‐year period and (2) late ART era. For all analyses, a P value <0.05 was considered statistically significant. All statistical analysis was performed using SAS 9.3 (SAS Institute, Cary, NC).

RESULTS

Among 12,183 hospital discharges of HIV‐infected patients from 1995 to 2011, 406 (3.3%) died. Six medical records were missing or incomplete; 400 were available for review. The proportion of hospitalized HIV‐infected patients who died declined from 6.2% in 1995 to 1.5% in 2011 (P<0.0001).

Table 2 summarizes all 400 patients' demographic and clinical characteristics, and cause of death. The majority were male (65.5%), nonwhite (73.3%), and taking ART (65.9%), though only one‐third achieved a VL <400 copies/mL on the most recent measurement available in the year prior to death. The majority (56.3%) died due to nonAIDS‐related causes.

Among all AIDS‐related deaths from 1995 to 2011 (Table 2), AIDS‐defining infection was the most common cause (21.3%), followed by nonspecified AIDS (18.3%), and AIDS malignancy (4.3%). The proportion of nonAIDS‐related deaths increased significantly over time (Figure 1). The most common cause of nonAIDS‐related deaths was non‐AIDS infection (20.3%), followed by cardiovascular disease (11.3%), liver disease (8.5%), malignancy (7.8%), and renal failure (4.5%). The most common non‐AIDS infection was sepsis in 43 patients (60.6%), followed by nonrecurrent bacterial pneumonia in 24 patients (33.8%) and Clostridium difficile infection in 4 patients (5.6%). NonAIDS‐related malignancy was the only category to significantly increase from the early ART to late ART era (P=0.01).

Figure 1

Compared to those dying of AIDS‐related causes over the 17‐year period (Table 3), patients dying of nonAIDS‐related causes were older (P<0.0001), less likely to have a CD4 count 200 cells/mm³ (P<0.0001), and more likely to be on ART and virologically suppressed (P<0.0001). Patients who died from nonAIDS‐related causes were also more likely to have diabetes mellitus (P=0.01), chronic kidney disease (P<0.0001), hepatitis C (P<0.0001), liver cirrhosis (P<0.0001), hypertension (P=0.0002), coronary artery disease (P=0.004), and COPD (P=0.04). Of note, there was no statistically significant difference in gender, race, or substance abuse between AIDS‐related and nonAIDS‐related deaths.

DISCUSSION

Our study demonstrated changes in the causes of death among HIV‐infected hospitalized patients from 1995 to 2011. To our knowledge, this is the longest duration retrospective analysis of in‐hospital deaths among HIV‐infected patients during the ART era. Knowledge of the changes in comorbidities and causes of death among hospitalized HIV‐infected patients during the ART era could help inpatient providers focus diagnostic and therapeutic efforts and improve overall care. Our findings emphasize that HIV‐infected patients remain at high risk for complications from non‐AIDS infections, even when their immune system has been restored as measured by the CD4 cell count, and at increased risk of cardiovascular and liver disease, which highlights the need to carefully monitor HIV‐positive patients admitted with these conditions.

Comparison of AIDS‐related and nonAIDS‐related deaths in 2 time periods has revealed important findings. First, inpatient deaths of HIV‐infected patients have decreased dramatically (from 6.2% to 1.5%, P<0.0001), and the mortality due to nonAIDS‐related causes has increased significantly over time. Second, we defined demographic and clinical characteristics independently associated with HIV‐infected inpatient mortality. Third, a substantial proportion of in‐hospital deaths were caused by potentially preventable non‐AIDS as well as AIDS‐related diseases.

The striking decline in hospital deaths over time is likely the result of expanded ART use resulting in improved immunologic profiles. NonAIDS‐related causes were responsible for almost three‐quarters of deaths in this large inpatient HIV‐positive population during the late ART era. Similar findings have been reported from other settings in industrialized countries.[5, 7, 16, 17, 18, 26, 27] In our urban population, although cardiovascular disease, liver disease, renal failure, and malignancy were frequent causes of non‐AIDS death, the most common cause was non‐AIDS infection. Further, the proportion of deaths due to non‐AIDS infections did not decrease significantly over time.

A similar study of HIV‐positive inpatients in New York City also found that the majority of non‐AIDS deaths were due to non‐AIDS infections in the ART era.[17] The most common causes of non‐AIDS infection identified in the study were identical to ours: unspecified sepsis followed by nonrecurrent bacterial pneumonia and Clostridium difficile infection. Evidence suggests that individuals with HIV infection have multiple immunological defects that not only lead to increased susceptibility to bacterial infection but also to an unregulated inflammatory response, even in patients who are on ART and virologically suppressed.[28, 29] This highlights the need for hospital physicians to evaluate an HIV‐infected patient's risk for more routine infections that are not commonly considered AIDS related in addition to traditional opportunistic infections. It also implies that inpatient providers should carefully monitor HIV‐positive patients admitted for bacterial infections, as they remain at higher risk for the development of septic shock.

Cardiovascular and liver disease represented the next most common causes of death, which is similar to the New York City study and is consistent with other studies from the ART era.[15, 16, 17, 18] Although deaths due directly to cardiovascular and liver disease did not significantly change over time, these represented the major comorbidities associated with non‐AIDS mortality and, along with renal disease, increased significantly over the study period. There are accumulating studies indicating that HIV infection is associated with accelerated coronary artery disease due to the immune and inflammatory response to the viral replication.[30] Additionally, ART side effects such as hyperlipidemia, metabolic syndrome, and insulin resistance contribute to an increased cardiovascular risk profile.[31] Our findings emphasize the importance of assessing comorbidities not classically considered HIV related. For example, acute coronary syndrome should be in the differential diagnosis for HIV‐infected patients admitted with chest pain regardless of age. Furthermore, HIV‐infected patients are at increased risk for hepatitis B and C coinfection due to related behavioral risk, and coinfection is associated with rapid progression to liver cirrhosis[32, 33, 34] and increased risk for oncogenesis over time rapidly expanding therapeutic options will benefit patients with chronic liver disease.[35, 36, 37]

Although the numbers are relatively small, non‐AIDS malignancy deaths more than quadrupled from the early to the late ART eras. This finding likely underestimates the proportion of overall hospital deaths due to non‐AIDS malignancies given the increased use of hospice facilities and community‐based care,[38] though it is consistent with increasing trends noted in other studies.[39] Doubling of malignancy as a cause of death among AIDS patients from 2000 to 2010 was reported in a French study, as well as in a large multicohort study from 1999 to 2011, consistent with our findings.[16, 40] Developing and implementing screening guidelines for non‐AIDS malignancy among those with HIV at the primary care level may potentially reduce this upward trend.[41] Inpatient providers need to be aware of this trend and consider undiagnosed non‐AIDS malignancy as part of their differential diagnosis when evaluating HIV‐positive patients.

Although emphasis has been placed on non‐AIDS causes, nearly one‐half of all deaths for the entire period, and almost one‐third of deaths in the late ART era were still due to AIDS‐related causes. This is similar to a study of 40,000 patients in Europe and North America from 1996 to 2006, where AIDS deaths comprised almost half of all deaths,[7] as well as a French national study,[16] and remains characteristic of resource‐limited settings.[42] This indicates the need for continued vigilance toward earlier HIV case detection and retention in care to prevent disease progression and AIDS‐related mortality. Primary care and hospital physicians should assess risk for HIV infection in all patients and institute universal HIV testing in both the inpatient and outpatient settings.

Although the majority of our sample was nonwhite and male, there was sufficient demographic diversity to determine that race and gender differences were not statistically significant contributors to mortality. In contrast, hospital‐based and population‐based studies reporting racial and gender disparities in HIV‐associated mortality have attributed this to poor access to health care.[9, 17, 43, 44, 45, 46] Compared to the New York City study, patients in our study had comparable median age and CD4 cell count, but also had greater ART use and better virologic control.[17] We speculate that in our smaller urban area, characterized by strong community and clinical HIV programs, patients may have had improved access to care without regard to race and gender.

Our study strengths include a large sample size, a diverse population with a relatively high proportion of women, and varied age and race, as well as data acquired in a standardized fashion over a prolonged period of ART availability. Further, 2 clinicians classified causes of death independently, utilizing validated definitions to minimize bias. Our late ART era evaluation is consistent with other HIV cohort studies,[25] though we utilized multivariate analysis to uncover independent correlates of mortality, a feature not employed in other studies.[16, 17]

We also recognize several limitations in our study. Our study design was associated with the recognized limitations of retrospective research, including missing data. We examined in‐hospital deaths at a single urban hospital in the Northeastern United States only, affecting the generalizability of our findings. The study did not include a control group of hospitalized HIV‐infected patients who survived or hospitalized HIV‐negative patients who died, which might have further strengthened our findings. Despite these limitations, this study provides important observations that can inform strategies to impact HIV‐associated mortality in the inpatient setting.

In conclusion, the mortality profile of hospitalized HIV‐infected patients has evolved with the epidemic. Caring for the hospitalized HIV‐infected patient has become increasingly complex because patients are more likely to suffer from multiple comorbidities, especially cardiovascular and liver diseases, and to die from non‐AIDS causes. Inpatient providers need to understand the changing trends in chronic HIV disease management as patients are living longer with antiretroviral therapy and are increasingly likely to succumb to nonAIDS‐related causes of death. Clinicians can no longer remain focused on AIDS‐defining opportunistic infections and need to recognize the emerging importance of chronic comorbidities when developing a differential diagnosis, and the higher risk of death due to non‐AIDS infectious causes. Physicians caring for hospitalized patients should appreciate the current trends in the HIV epidemic to provide comprehensive and appropriate interventions that can reduce mortality for HIV‐infected inpatients.

Disclosures: This research was supported by the National Institute of Allergy and Infectious Diseases (S.S.; 1K23AI089260). The authors report no conflicts of interest.

Successfully treated human immunodeficiency virus (HIV)‐infected individuals in the United States currently have life expectancy and mortality rates that are similar to the general population.[1, 2, 3, 4] A large multinational study found that the excess mortality rate among HIV‐positive individuals decreased from 40.8 to 6.1 per 1000 person‐years from pre‐1995 to 2006.¹ This is largely due to improved access to comprehensive HIV care, in particular widespread antiretroviral (ART) use. However, the proportion of deaths that are not classically considered acquired immunodeficiency syndrome (AIDS)‐related such as liver disease, cardiovascular disease, and non‐AIDS malignancy has increased,[1, 5, 6, 7] particularly among patients with higher CD4 T‐cell counts.[5, 8] Additionally, despite overall decline in mortality, there is evidence of racial and gender differences, with increased mortality risk associated with female gender and black race.[9, 10]

In the current ART era, HIV care has shifted focus from inpatient to outpatient care, with more emphasis on chronic disease management. However, hospitalization rates among HIV‐positive persons remain higher than that of the general population.[11, 12] A cross‐sectional study of HIV‐infected persons in the United States estimated a hospitalization rate of 26.6 per 100 persons in 2009,[13] compared to a rate of 11.9 for the general population during the same year.[14] Possible reasons for higher hospitalization rates include complications of aging or other chronic comorbidities, and consequences of behavioral risk factors such tobacco use and substance abuse.

Characterizing deaths among inpatient HIV‐infected individuals in the ART era is important to developing targeted interventions to further reduce mortality. Prior studies examining in‐hospital deaths of HIV‐positive patients evaluated more limited time periods,[15, 16, 17, 18] and thus did not necessarily assess the full spectrum of changes in mortality that have occurred with the introduction of ART. Furthermore, these studies described causes of death, but did not consistently identify factors associated with non‐AIDS deaths. We examined the trends in in‐hospital deaths among HIV‐infected patients from 1995 to 2011 and identified contributing factors to mortality. As the HIV population is aging, we hypothesized that HIV‐infected patients are more likely to die from nonAIDS‐related death in the late ART era due to factors related to cardiovascular and liver disease, compared to the early ART era.

METHODS

The study was performed at YaleNew Haven Hospital, an urban tertiary care academic teaching hospital with 1008 beds and the state of Connecticut's largest ambulatory HIV clinic. Connecticut ranks seventh nationally (10/100,000) in HIV prevalence; New Haven is second among Connecticut cities in the number of people living with HIV/AIDS.[19] We reviewed all patients with an International Classification of Diseases, Ninth Revision (ICD‐9) code of HIV or AIDS (ICD‐9 codes V08 and 042) who died during hospitalization between January 1, 1995 to December 31, 2011. The Yale Human Investigation Committee granted ethical approval to conduct the study.

A standardized data collection tool was used to abstract demographic characteristics (ie, age, gender, and race), medical comorbidities (ie, diabetes, chronic kidney disease, chronic hepatitis B or C, liver cirrhosis, hypertension, coronary artery disease, congestive heart failure, chronic obstructive lung disease, alcohol and substance abuse), ART use (yes or no), HIV viral load (VL), CD4 cell count, and causes of death. Comorbidities were defined using the Coding of Death in HIV Project protocol, a multinational endeavor to standardize data collection in studies of HIV‐positive patients.[20] Chronic kidney disease included individuals with National Kidney Foundation stage I to V disease. Chronic hepatitis B or C infection was identified in patients who had serologic testing indicative of prior infection. Alcohol and substance abuse were identified when source documents mentioned any history of current alcohol or illicit drug abuse or dependence. ART use was defined as documentation of ART on admission or prescription during hospitalization. This included individuals who were on 2 or more ART agents. The last HIV VL and CD4 cell count available within 1 year and closest to death were recorded. HIV VL suppression was defined as <400 copies/mL.

Two clinicians independently classified the cause of death as AIDS related or non‐AIDS related in accordance with published definitions.[21, 22] Cause of death was determined by review of the medical record, discharge diagnosis, and autopsy report when available. Official death certificates were not available for review. There was discordance in assigning 23 of the 400 causes of death. In these cases, the medical record was reviewed and determined by consensus between the 2 clinicians.

AIDS‐related deaths were categorized as nonspecified AIDS, AIDS infection, and AIDS malignancy. AIDS‐related deaths were defined as those caused by conditions meeting the Centers for Disease Control and Prevention AIDS case definition.[22] Non‐specified AIDS deaths were those occurring in patients with a CD4 count 50 cells/mm³ or with an AIDS‐defining illness, who died from a condition that was not clearly AIDS related. This included septic shock of unclear etiology, first known episode of pneumonia, a gastrointestinal bleed of unclear etiology, and altered mental status of unclear etiology when cerebrospinal fluid analysis or imaging of the brain was not available.

Non‐AIDS deaths included non‐AIDS infection in patients with a CD4 count >50 cells/mm³, cardiovascular disease, liver disease, non‐AIDS malignancy, and renal disease (Table 1). Deaths classified as other incorporated the deaths that did not fall into these categories. Chronic obstructive pulmonary disease (COPD) exacerbation and status asthmaticus were included in this category, because there was only 1 death from each of these causes.

The early ART era was defined as 1995 to 2001 and the late ART era from 2002 to 2011. During the early period, combination ART was introduced and significantly impacted overall mortality.[23, 24] The late ART era better reflected current in‐hospital deaths and was compared to the early era to evaluate trends over time.[15, 25]

² analysis and parametric (t test and analysis of variance) methods compared categorical and continuous variables, respectively. Bivariate analysis was used to determine associations with AIDS versus non‐AIDS deaths in the entire study cohort. Multivariable logistic regression was used to identify correlates of non‐AIDS deaths in the (1) complete 17‐year period and (2) late ART era. For all analyses, a P value <0.05 was considered statistically significant. All statistical analysis was performed using SAS 9.3 (SAS Institute, Cary, NC).

RESULTS

Among 12,183 hospital discharges of HIV‐infected patients from 1995 to 2011, 406 (3.3%) died. Six medical records were missing or incomplete; 400 were available for review. The proportion of hospitalized HIV‐infected patients who died declined from 6.2% in 1995 to 1.5% in 2011 (P<0.0001).

Table 2 summarizes all 400 patients' demographic and clinical characteristics, and cause of death. The majority were male (65.5%), nonwhite (73.3%), and taking ART (65.9%), though only one‐third achieved a VL <400 copies/mL on the most recent measurement available in the year prior to death. The majority (56.3%) died due to nonAIDS‐related causes.

Among all AIDS‐related deaths from 1995 to 2011 (Table 2), AIDS‐defining infection was the most common cause (21.3%), followed by nonspecified AIDS (18.3%), and AIDS malignancy (4.3%). The proportion of nonAIDS‐related deaths increased significantly over time (Figure 1). The most common cause of nonAIDS‐related deaths was non‐AIDS infection (20.3%), followed by cardiovascular disease (11.3%), liver disease (8.5%), malignancy (7.8%), and renal failure (4.5%). The most common non‐AIDS infection was sepsis in 43 patients (60.6%), followed by nonrecurrent bacterial pneumonia in 24 patients (33.8%) and Clostridium difficile infection in 4 patients (5.6%). NonAIDS‐related malignancy was the only category to significantly increase from the early ART to late ART era (P=0.01).

Figure 1

Compared to those dying of AIDS‐related causes over the 17‐year period (Table 3), patients dying of nonAIDS‐related causes were older (P<0.0001), less likely to have a CD4 count 200 cells/mm³ (P<0.0001), and more likely to be on ART and virologically suppressed (P<0.0001). Patients who died from nonAIDS‐related causes were also more likely to have diabetes mellitus (P=0.01), chronic kidney disease (P<0.0001), hepatitis C (P<0.0001), liver cirrhosis (P<0.0001), hypertension (P=0.0002), coronary artery disease (P=0.004), and COPD (P=0.04). Of note, there was no statistically significant difference in gender, race, or substance abuse between AIDS‐related and nonAIDS‐related deaths.

DISCUSSION

Our study demonstrated changes in the causes of death among HIV‐infected hospitalized patients from 1995 to 2011. To our knowledge, this is the longest duration retrospective analysis of in‐hospital deaths among HIV‐infected patients during the ART era. Knowledge of the changes in comorbidities and causes of death among hospitalized HIV‐infected patients during the ART era could help inpatient providers focus diagnostic and therapeutic efforts and improve overall care. Our findings emphasize that HIV‐infected patients remain at high risk for complications from non‐AIDS infections, even when their immune system has been restored as measured by the CD4 cell count, and at increased risk of cardiovascular and liver disease, which highlights the need to carefully monitor HIV‐positive patients admitted with these conditions.

Comparison of AIDS‐related and nonAIDS‐related deaths in 2 time periods has revealed important findings. First, inpatient deaths of HIV‐infected patients have decreased dramatically (from 6.2% to 1.5%, P<0.0001), and the mortality due to nonAIDS‐related causes has increased significantly over time. Second, we defined demographic and clinical characteristics independently associated with HIV‐infected inpatient mortality. Third, a substantial proportion of in‐hospital deaths were caused by potentially preventable non‐AIDS as well as AIDS‐related diseases.

The striking decline in hospital deaths over time is likely the result of expanded ART use resulting in improved immunologic profiles. NonAIDS‐related causes were responsible for almost three‐quarters of deaths in this large inpatient HIV‐positive population during the late ART era. Similar findings have been reported from other settings in industrialized countries.[5, 7, 16, 17, 18, 26, 27] In our urban population, although cardiovascular disease, liver disease, renal failure, and malignancy were frequent causes of non‐AIDS death, the most common cause was non‐AIDS infection. Further, the proportion of deaths due to non‐AIDS infections did not decrease significantly over time.

A similar study of HIV‐positive inpatients in New York City also found that the majority of non‐AIDS deaths were due to non‐AIDS infections in the ART era.[17] The most common causes of non‐AIDS infection identified in the study were identical to ours: unspecified sepsis followed by nonrecurrent bacterial pneumonia and Clostridium difficile infection. Evidence suggests that individuals with HIV infection have multiple immunological defects that not only lead to increased susceptibility to bacterial infection but also to an unregulated inflammatory response, even in patients who are on ART and virologically suppressed.[28, 29] This highlights the need for hospital physicians to evaluate an HIV‐infected patient's risk for more routine infections that are not commonly considered AIDS related in addition to traditional opportunistic infections. It also implies that inpatient providers should carefully monitor HIV‐positive patients admitted for bacterial infections, as they remain at higher risk for the development of septic shock.

Cardiovascular and liver disease represented the next most common causes of death, which is similar to the New York City study and is consistent with other studies from the ART era.[15, 16, 17, 18] Although deaths due directly to cardiovascular and liver disease did not significantly change over time, these represented the major comorbidities associated with non‐AIDS mortality and, along with renal disease, increased significantly over the study period. There are accumulating studies indicating that HIV infection is associated with accelerated coronary artery disease due to the immune and inflammatory response to the viral replication.[30] Additionally, ART side effects such as hyperlipidemia, metabolic syndrome, and insulin resistance contribute to an increased cardiovascular risk profile.[31] Our findings emphasize the importance of assessing comorbidities not classically considered HIV related. For example, acute coronary syndrome should be in the differential diagnosis for HIV‐infected patients admitted with chest pain regardless of age. Furthermore, HIV‐infected patients are at increased risk for hepatitis B and C coinfection due to related behavioral risk, and coinfection is associated with rapid progression to liver cirrhosis[32, 33, 34] and increased risk for oncogenesis over time rapidly expanding therapeutic options will benefit patients with chronic liver disease.[35, 36, 37]

Although the numbers are relatively small, non‐AIDS malignancy deaths more than quadrupled from the early to the late ART eras. This finding likely underestimates the proportion of overall hospital deaths due to non‐AIDS malignancies given the increased use of hospice facilities and community‐based care,[38] though it is consistent with increasing trends noted in other studies.[39] Doubling of malignancy as a cause of death among AIDS patients from 2000 to 2010 was reported in a French study, as well as in a large multicohort study from 1999 to 2011, consistent with our findings.[16, 40] Developing and implementing screening guidelines for non‐AIDS malignancy among those with HIV at the primary care level may potentially reduce this upward trend.[41] Inpatient providers need to be aware of this trend and consider undiagnosed non‐AIDS malignancy as part of their differential diagnosis when evaluating HIV‐positive patients.

Although emphasis has been placed on non‐AIDS causes, nearly one‐half of all deaths for the entire period, and almost one‐third of deaths in the late ART era were still due to AIDS‐related causes. This is similar to a study of 40,000 patients in Europe and North America from 1996 to 2006, where AIDS deaths comprised almost half of all deaths,[7] as well as a French national study,[16] and remains characteristic of resource‐limited settings.[42] This indicates the need for continued vigilance toward earlier HIV case detection and retention in care to prevent disease progression and AIDS‐related mortality. Primary care and hospital physicians should assess risk for HIV infection in all patients and institute universal HIV testing in both the inpatient and outpatient settings.

Although the majority of our sample was nonwhite and male, there was sufficient demographic diversity to determine that race and gender differences were not statistically significant contributors to mortality. In contrast, hospital‐based and population‐based studies reporting racial and gender disparities in HIV‐associated mortality have attributed this to poor access to health care.[9, 17, 43, 44, 45, 46] Compared to the New York City study, patients in our study had comparable median age and CD4 cell count, but also had greater ART use and better virologic control.[17] We speculate that in our smaller urban area, characterized by strong community and clinical HIV programs, patients may have had improved access to care without regard to race and gender.

Our study strengths include a large sample size, a diverse population with a relatively high proportion of women, and varied age and race, as well as data acquired in a standardized fashion over a prolonged period of ART availability. Further, 2 clinicians classified causes of death independently, utilizing validated definitions to minimize bias. Our late ART era evaluation is consistent with other HIV cohort studies,[25] though we utilized multivariate analysis to uncover independent correlates of mortality, a feature not employed in other studies.[16, 17]

We also recognize several limitations in our study. Our study design was associated with the recognized limitations of retrospective research, including missing data. We examined in‐hospital deaths at a single urban hospital in the Northeastern United States only, affecting the generalizability of our findings. The study did not include a control group of hospitalized HIV‐infected patients who survived or hospitalized HIV‐negative patients who died, which might have further strengthened our findings. Despite these limitations, this study provides important observations that can inform strategies to impact HIV‐associated mortality in the inpatient setting.

In conclusion, the mortality profile of hospitalized HIV‐infected patients has evolved with the epidemic. Caring for the hospitalized HIV‐infected patient has become increasingly complex because patients are more likely to suffer from multiple comorbidities, especially cardiovascular and liver diseases, and to die from non‐AIDS causes. Inpatient providers need to understand the changing trends in chronic HIV disease management as patients are living longer with antiretroviral therapy and are increasingly likely to succumb to nonAIDS‐related causes of death. Clinicians can no longer remain focused on AIDS‐defining opportunistic infections and need to recognize the emerging importance of chronic comorbidities when developing a differential diagnosis, and the higher risk of death due to non‐AIDS infectious causes. Physicians caring for hospitalized patients should appreciate the current trends in the HIV epidemic to provide comprehensive and appropriate interventions that can reduce mortality for HIV‐infected inpatients.

Disclosures: This research was supported by the National Institute of Allergy and Infectious Diseases (S.S.; 1K23AI089260). The authors report no conflicts of interest.

Successfully treated human immunodeficiency virus (HIV)‐infected individuals in the United States currently have life expectancy and mortality rates that are similar to the general population.[1, 2, 3, 4] A large multinational study found that the excess mortality rate among HIV‐positive individuals decreased from 40.8 to 6.1 per 1000 person‐years from pre‐1995 to 2006.¹ This is largely due to improved access to comprehensive HIV care, in particular widespread antiretroviral (ART) use. However, the proportion of deaths that are not classically considered acquired immunodeficiency syndrome (AIDS)‐related such as liver disease, cardiovascular disease, and non‐AIDS malignancy has increased,[1, 5, 6, 7] particularly among patients with higher CD4 T‐cell counts.[5, 8] Additionally, despite overall decline in mortality, there is evidence of racial and gender differences, with increased mortality risk associated with female gender and black race.[9, 10]

In the current ART era, HIV care has shifted focus from inpatient to outpatient care, with more emphasis on chronic disease management. However, hospitalization rates among HIV‐positive persons remain higher than that of the general population.[11, 12] A cross‐sectional study of HIV‐infected persons in the United States estimated a hospitalization rate of 26.6 per 100 persons in 2009,[13] compared to a rate of 11.9 for the general population during the same year.[14] Possible reasons for higher hospitalization rates include complications of aging or other chronic comorbidities, and consequences of behavioral risk factors such tobacco use and substance abuse.

Characterizing deaths among inpatient HIV‐infected individuals in the ART era is important to developing targeted interventions to further reduce mortality. Prior studies examining in‐hospital deaths of HIV‐positive patients evaluated more limited time periods,[15, 16, 17, 18] and thus did not necessarily assess the full spectrum of changes in mortality that have occurred with the introduction of ART. Furthermore, these studies described causes of death, but did not consistently identify factors associated with non‐AIDS deaths. We examined the trends in in‐hospital deaths among HIV‐infected patients from 1995 to 2011 and identified contributing factors to mortality. As the HIV population is aging, we hypothesized that HIV‐infected patients are more likely to die from nonAIDS‐related death in the late ART era due to factors related to cardiovascular and liver disease, compared to the early ART era.

METHODS

The study was performed at YaleNew Haven Hospital, an urban tertiary care academic teaching hospital with 1008 beds and the state of Connecticut's largest ambulatory HIV clinic. Connecticut ranks seventh nationally (10/100,000) in HIV prevalence; New Haven is second among Connecticut cities in the number of people living with HIV/AIDS.[19] We reviewed all patients with an International Classification of Diseases, Ninth Revision (ICD‐9) code of HIV or AIDS (ICD‐9 codes V08 and 042) who died during hospitalization between January 1, 1995 to December 31, 2011. The Yale Human Investigation Committee granted ethical approval to conduct the study.

A standardized data collection tool was used to abstract demographic characteristics (ie, age, gender, and race), medical comorbidities (ie, diabetes, chronic kidney disease, chronic hepatitis B or C, liver cirrhosis, hypertension, coronary artery disease, congestive heart failure, chronic obstructive lung disease, alcohol and substance abuse), ART use (yes or no), HIV viral load (VL), CD4 cell count, and causes of death. Comorbidities were defined using the Coding of Death in HIV Project protocol, a multinational endeavor to standardize data collection in studies of HIV‐positive patients.[20] Chronic kidney disease included individuals with National Kidney Foundation stage I to V disease. Chronic hepatitis B or C infection was identified in patients who had serologic testing indicative of prior infection. Alcohol and substance abuse were identified when source documents mentioned any history of current alcohol or illicit drug abuse or dependence. ART use was defined as documentation of ART on admission or prescription during hospitalization. This included individuals who were on 2 or more ART agents. The last HIV VL and CD4 cell count available within 1 year and closest to death were recorded. HIV VL suppression was defined as <400 copies/mL.

Two clinicians independently classified the cause of death as AIDS related or non‐AIDS related in accordance with published definitions.[21, 22] Cause of death was determined by review of the medical record, discharge diagnosis, and autopsy report when available. Official death certificates were not available for review. There was discordance in assigning 23 of the 400 causes of death. In these cases, the medical record was reviewed and determined by consensus between the 2 clinicians.

AIDS‐related deaths were categorized as nonspecified AIDS, AIDS infection, and AIDS malignancy. AIDS‐related deaths were defined as those caused by conditions meeting the Centers for Disease Control and Prevention AIDS case definition.[22] Non‐specified AIDS deaths were those occurring in patients with a CD4 count 50 cells/mm³ or with an AIDS‐defining illness, who died from a condition that was not clearly AIDS related. This included septic shock of unclear etiology, first known episode of pneumonia, a gastrointestinal bleed of unclear etiology, and altered mental status of unclear etiology when cerebrospinal fluid analysis or imaging of the brain was not available.

Non‐AIDS deaths included non‐AIDS infection in patients with a CD4 count >50 cells/mm³, cardiovascular disease, liver disease, non‐AIDS malignancy, and renal disease (Table 1). Deaths classified as other incorporated the deaths that did not fall into these categories. Chronic obstructive pulmonary disease (COPD) exacerbation and status asthmaticus were included in this category, because there was only 1 death from each of these causes.

The early ART era was defined as 1995 to 2001 and the late ART era from 2002 to 2011. During the early period, combination ART was introduced and significantly impacted overall mortality.[23, 24] The late ART era better reflected current in‐hospital deaths and was compared to the early era to evaluate trends over time.[15, 25]

² analysis and parametric (t test and analysis of variance) methods compared categorical and continuous variables, respectively. Bivariate analysis was used to determine associations with AIDS versus non‐AIDS deaths in the entire study cohort. Multivariable logistic regression was used to identify correlates of non‐AIDS deaths in the (1) complete 17‐year period and (2) late ART era. For all analyses, a P value <0.05 was considered statistically significant. All statistical analysis was performed using SAS 9.3 (SAS Institute, Cary, NC).

RESULTS

Among 12,183 hospital discharges of HIV‐infected patients from 1995 to 2011, 406 (3.3%) died. Six medical records were missing or incomplete; 400 were available for review. The proportion of hospitalized HIV‐infected patients who died declined from 6.2% in 1995 to 1.5% in 2011 (P<0.0001).

Table 2 summarizes all 400 patients' demographic and clinical characteristics, and cause of death. The majority were male (65.5%), nonwhite (73.3%), and taking ART (65.9%), though only one‐third achieved a VL <400 copies/mL on the most recent measurement available in the year prior to death. The majority (56.3%) died due to nonAIDS‐related causes.

Among all AIDS‐related deaths from 1995 to 2011 (Table 2), AIDS‐defining infection was the most common cause (21.3%), followed by nonspecified AIDS (18.3%), and AIDS malignancy (4.3%). The proportion of nonAIDS‐related deaths increased significantly over time (Figure 1). The most common cause of nonAIDS‐related deaths was non‐AIDS infection (20.3%), followed by cardiovascular disease (11.3%), liver disease (8.5%), malignancy (7.8%), and renal failure (4.5%). The most common non‐AIDS infection was sepsis in 43 patients (60.6%), followed by nonrecurrent bacterial pneumonia in 24 patients (33.8%) and Clostridium difficile infection in 4 patients (5.6%). NonAIDS‐related malignancy was the only category to significantly increase from the early ART to late ART era (P=0.01).

Figure 1

Compared to those dying of AIDS‐related causes over the 17‐year period (Table 3), patients dying of nonAIDS‐related causes were older (P<0.0001), less likely to have a CD4 count 200 cells/mm³ (P<0.0001), and more likely to be on ART and virologically suppressed (P<0.0001). Patients who died from nonAIDS‐related causes were also more likely to have diabetes mellitus (P=0.01), chronic kidney disease (P<0.0001), hepatitis C (P<0.0001), liver cirrhosis (P<0.0001), hypertension (P=0.0002), coronary artery disease (P=0.004), and COPD (P=0.04). Of note, there was no statistically significant difference in gender, race, or substance abuse between AIDS‐related and nonAIDS‐related deaths.

DISCUSSION

Our study demonstrated changes in the causes of death among HIV‐infected hospitalized patients from 1995 to 2011. To our knowledge, this is the longest duration retrospective analysis of in‐hospital deaths among HIV‐infected patients during the ART era. Knowledge of the changes in comorbidities and causes of death among hospitalized HIV‐infected patients during the ART era could help inpatient providers focus diagnostic and therapeutic efforts and improve overall care. Our findings emphasize that HIV‐infected patients remain at high risk for complications from non‐AIDS infections, even when their immune system has been restored as measured by the CD4 cell count, and at increased risk of cardiovascular and liver disease, which highlights the need to carefully monitor HIV‐positive patients admitted with these conditions.

Comparison of AIDS‐related and nonAIDS‐related deaths in 2 time periods has revealed important findings. First, inpatient deaths of HIV‐infected patients have decreased dramatically (from 6.2% to 1.5%, P<0.0001), and the mortality due to nonAIDS‐related causes has increased significantly over time. Second, we defined demographic and clinical characteristics independently associated with HIV‐infected inpatient mortality. Third, a substantial proportion of in‐hospital deaths were caused by potentially preventable non‐AIDS as well as AIDS‐related diseases.

The striking decline in hospital deaths over time is likely the result of expanded ART use resulting in improved immunologic profiles. NonAIDS‐related causes were responsible for almost three‐quarters of deaths in this large inpatient HIV‐positive population during the late ART era. Similar findings have been reported from other settings in industrialized countries.[5, 7, 16, 17, 18, 26, 27] In our urban population, although cardiovascular disease, liver disease, renal failure, and malignancy were frequent causes of non‐AIDS death, the most common cause was non‐AIDS infection. Further, the proportion of deaths due to non‐AIDS infections did not decrease significantly over time.

A similar study of HIV‐positive inpatients in New York City also found that the majority of non‐AIDS deaths were due to non‐AIDS infections in the ART era.[17] The most common causes of non‐AIDS infection identified in the study were identical to ours: unspecified sepsis followed by nonrecurrent bacterial pneumonia and Clostridium difficile infection. Evidence suggests that individuals with HIV infection have multiple immunological defects that not only lead to increased susceptibility to bacterial infection but also to an unregulated inflammatory response, even in patients who are on ART and virologically suppressed.[28, 29] This highlights the need for hospital physicians to evaluate an HIV‐infected patient's risk for more routine infections that are not commonly considered AIDS related in addition to traditional opportunistic infections. It also implies that inpatient providers should carefully monitor HIV‐positive patients admitted for bacterial infections, as they remain at higher risk for the development of septic shock.

Cardiovascular and liver disease represented the next most common causes of death, which is similar to the New York City study and is consistent with other studies from the ART era.[15, 16, 17, 18] Although deaths due directly to cardiovascular and liver disease did not significantly change over time, these represented the major comorbidities associated with non‐AIDS mortality and, along with renal disease, increased significantly over the study period. There are accumulating studies indicating that HIV infection is associated with accelerated coronary artery disease due to the immune and inflammatory response to the viral replication.[30] Additionally, ART side effects such as hyperlipidemia, metabolic syndrome, and insulin resistance contribute to an increased cardiovascular risk profile.[31] Our findings emphasize the importance of assessing comorbidities not classically considered HIV related. For example, acute coronary syndrome should be in the differential diagnosis for HIV‐infected patients admitted with chest pain regardless of age. Furthermore, HIV‐infected patients are at increased risk for hepatitis B and C coinfection due to related behavioral risk, and coinfection is associated with rapid progression to liver cirrhosis[32, 33, 34] and increased risk for oncogenesis over time rapidly expanding therapeutic options will benefit patients with chronic liver disease.[35, 36, 37]

Although the numbers are relatively small, non‐AIDS malignancy deaths more than quadrupled from the early to the late ART eras. This finding likely underestimates the proportion of overall hospital deaths due to non‐AIDS malignancies given the increased use of hospice facilities and community‐based care,[38] though it is consistent with increasing trends noted in other studies.[39] Doubling of malignancy as a cause of death among AIDS patients from 2000 to 2010 was reported in a French study, as well as in a large multicohort study from 1999 to 2011, consistent with our findings.[16, 40] Developing and implementing screening guidelines for non‐AIDS malignancy among those with HIV at the primary care level may potentially reduce this upward trend.[41] Inpatient providers need to be aware of this trend and consider undiagnosed non‐AIDS malignancy as part of their differential diagnosis when evaluating HIV‐positive patients.

Although emphasis has been placed on non‐AIDS causes, nearly one‐half of all deaths for the entire period, and almost one‐third of deaths in the late ART era were still due to AIDS‐related causes. This is similar to a study of 40,000 patients in Europe and North America from 1996 to 2006, where AIDS deaths comprised almost half of all deaths,[7] as well as a French national study,[16] and remains characteristic of resource‐limited settings.[42] This indicates the need for continued vigilance toward earlier HIV case detection and retention in care to prevent disease progression and AIDS‐related mortality. Primary care and hospital physicians should assess risk for HIV infection in all patients and institute universal HIV testing in both the inpatient and outpatient settings.

Although the majority of our sample was nonwhite and male, there was sufficient demographic diversity to determine that race and gender differences were not statistically significant contributors to mortality. In contrast, hospital‐based and population‐based studies reporting racial and gender disparities in HIV‐associated mortality have attributed this to poor access to health care.[9, 17, 43, 44, 45, 46] Compared to the New York City study, patients in our study had comparable median age and CD4 cell count, but also had greater ART use and better virologic control.[17] We speculate that in our smaller urban area, characterized by strong community and clinical HIV programs, patients may have had improved access to care without regard to race and gender.

Our study strengths include a large sample size, a diverse population with a relatively high proportion of women, and varied age and race, as well as data acquired in a standardized fashion over a prolonged period of ART availability. Further, 2 clinicians classified causes of death independently, utilizing validated definitions to minimize bias. Our late ART era evaluation is consistent with other HIV cohort studies,[25] though we utilized multivariate analysis to uncover independent correlates of mortality, a feature not employed in other studies.[16, 17]

We also recognize several limitations in our study. Our study design was associated with the recognized limitations of retrospective research, including missing data. We examined in‐hospital deaths at a single urban hospital in the Northeastern United States only, affecting the generalizability of our findings. The study did not include a control group of hospitalized HIV‐infected patients who survived or hospitalized HIV‐negative patients who died, which might have further strengthened our findings. Despite these limitations, this study provides important observations that can inform strategies to impact HIV‐associated mortality in the inpatient setting.

In conclusion, the mortality profile of hospitalized HIV‐infected patients has evolved with the epidemic. Caring for the hospitalized HIV‐infected patient has become increasingly complex because patients are more likely to suffer from multiple comorbidities, especially cardiovascular and liver diseases, and to die from non‐AIDS causes. Inpatient providers need to understand the changing trends in chronic HIV disease management as patients are living longer with antiretroviral therapy and are increasingly likely to succumb to nonAIDS‐related causes of death. Clinicians can no longer remain focused on AIDS‐defining opportunistic infections and need to recognize the emerging importance of chronic comorbidities when developing a differential diagnosis, and the higher risk of death due to non‐AIDS infectious causes. Physicians caring for hospitalized patients should appreciate the current trends in the HIV epidemic to provide comprehensive and appropriate interventions that can reduce mortality for HIV‐infected inpatients.

Disclosures: This research was supported by the National Institute of Allergy and Infectious Diseases (S.S.; 1K23AI089260). The authors report no conflicts of interest.

Hospitalists are a growing workforce numbering over 40,000 physicians, one‐third of whom are women.[1] Flexibility of work schedules and control over personal time have been the traditional selling points of the specialty.[2] Multiple studies of physician work life reveal growing physician dissatisfaction and a high prevalence of burnout.[3] To mitigate burnout risk, leaders in hospital medicine recognize the importance of creating a sustainable profession that offers both job and career satisfaction as well as work‐life balance and, importantly, fairness within the work environment.[4] Although success in some of these endeavors has been realized sporadically, sustaining work‐life balance and fairness in the specialty remains a work in progress, whereas evidence of high job attrition and pay inequities remain.[5, 6]

Pay inequity for women relative to men continues to be pervasive in medicine, including among early‐career physicians, researchers, and various specialists.[6, 7, 8, 9, 10, 11, 12, 13] The earnings gap seems to persist for physicians, even as federal efforts such as the Fair Pay Act of 2013 and the Paycheck Fairness Act of 2014 aim to end wage discrimination.[11, 14] Differences in specialty, part‐time status, and practice type do not mitigate the disparity.[8, 10, 15] Additional explanations have been proposed to explain the variability, including gender differences in negotiating skills, lack of opportunities to join networks of influence within organizations, and implicit or explicit bias and discrimination.[12, 16, 17, 18, 19, 20]

The earnings gap is also a consequence of what is commonly called the glass ceiling.[18, 19] Most agree that obstacles to fair advancement of women include absence of collaborative environments and role models who have successfully achieved work‐life balance.[17, 20, 21, 22] Somewhat surprisingly, women leaders in medicine seem to suffer greater income disparity than nonleaders; this income gap is prevalent among leaders in other elite professions as well.[7, 23] It is unknown whether women physicians' emphasis on work‐life balance, seen repeatedly in surveys, explains any of the pay disparity.[24] Little research to date has examined whether work‐life priorities of women in hospital medicine differ from men.

In this study, we sought to examine differences in job priorities between men and women hospitalists. In particular, we examine the relative prioritization of substantial pay to job satisfaction. We also examined gender differences in work patterns and earnings to explore potential sources of the persistent gender earnings gap.

METHODS

We analyzed data from the 20092010 Hospital Medicine Physician Worklife Survey, the design of which is detailed elsewhere.[4] Briefly, a 118‐item survey was administered by mail to a stratified sample of hospitalists from the Society of Hospital Medicine database and 3 large multisite hospitalist groups. A single survey item asked respondents to identify up to 4 out of 12 most important domains to their satisfaction with a hospitalist job. The domains were distilled from focus groups of nationally representative hospitalists as described previously,[4] and the survey item allowed up to one‐third of these domains to be identified as respondents' personal priorities. The list included: optimal variety of tasks, optimal workload, substantial pay, collegiality with other physicians, recognition by leaders, rewarding relationships with patients, satisfaction with nurses, optimal autonomy, control over personal time, fairness within organization, ample availability of resources to do job, and organizational climate of trust and belonging. We tabulated and ranked the frequency with which respondents selected each satisfaction domain by gender. Due to the nonstandard format of the survey item, we a priori decided to analyze only responses that were completed as instructed.

We also used demographic data including detailed work characteristics, clinical and nonclinical workload, total pretax earnings in 2009 as a hospitalist, and self‐identification as leader of their hospital medicine group. Respondent characteristics were tabulated and gender differences were tested using the t test, rank sum test, and the Fischer exact test as appropriate. We also listed the number of nonrespondents for each item. In estimating gender differences in earnings, we opted to use multiple‐imputation techniques to more conservatively account for greater variance inherent in the presence of missing data. Consistent with existing guidelines,[25] we demonstrated that item responses were not missing monotonically by visually inspecting patterns of nonresponse. We further demonstrated that data were missing at random by showing that response patterns of completed survey items did not predict whether or not a given variable response was missing using logistic regression models. We found no significant differences between respondents with complete and missing data. We verified that appropriate regression models for each variable on every other variable converged. We used Stata 13.1 (StataCorp, College Station, TX) to perform multiple imputations using chain equations (mi impute chain) to create 10 imputed tables for 7 normally distributed continuous variables using the ordinary least squares method, 3 non‐normally distributed variables using the predictive mean matching method, 2 nonordinal categorical variables using the multinomial logit method, and 1 binary variable using the logit method.[26, 27] Gender, pediatric specialty status, region of practice, and whether or not respondents prioritized substantial pay for job satisfaction were used as regular variables without missing data points.

Differences in earnings were assessed using a multivariate ordinary linear regression model applied to the imputed datasets fitted by forward selection of explanatory variables using P < 0.20 in bivariate analysis for inclusion and manual backward elimination of all statistically nonsignificant variables. We tested the significance of the women leader interaction term in the final parsimonious model. We used the usual significance threshold of P < 0.05 for inferences. Our analysis of publicly available anonymous data was exempt from IRB review.

RESULTS

Of the 816 survey respondents (response rate 25.6%), 40 either omitted the item soliciting work priorities or completed it incorrectly. Data from the remaining 776 respondents were used for the present analysis. Respondent characteristics are tabulated in Table 1. The characteristics of hospitalists by age, gender, specialty, practice model, and practice region were representative of US hospitalists from other surveys.[28]

Several gender differences were seen in the characteristics of hospitalists and their work (Table 1). Women compared to men hospitalists were less likely to be leaders, more likely to be pediatricians, work in university settings, and practice in Western states. Women compared to men, on average, were younger by 3 years, worked fewer full‐time equivalents (FTEs), worked a greater percentage of nights, and reported fewer billable encounters per shift. They were also more likely to be divorced or separated.

Job satisfaction priorities differed for women and men hospitalists. Table 2 lists job satisfaction domains in descending order of the frequency prioritized by men. The largest proportion of women and men prioritized optimal workload. However, although substantial pay was prioritized next most frequently by men, more women prioritized collegiality and control over personal time than substantial pay.

Key differences in individual characteristics, work patterns, and indicating substantial pay as a priority were associated with self‐reported total earnings in 2009 from respondents' work as a hospitalist. As shown in Table 3, the inclusion of detailed productivity measures such as FTE, days of monthly clinical work, and estimated number of daily billable encounters yielded a model that explained 33% of variance in earnings. After adjusting for significant covariates including pediatric specialty, practice model, geography, and amount and type of clinical work, the estimated underpayment of women compared to men was $14,581. Hospitalists who prioritized substantial pay earned $10,771 more than those who did not. The female x leader interaction term testing the hypothesis that gender disparity is greater among leaders than frontline hospitalists was not statistically significant ($16,720, P = 0.087) and excluded from the final model.

DISCUSSION

In a national stratified sample of US hospitalists, we found gender differences in job satisfaction priorities and hospitalist work characteristics. We also confirmed the persistence of a substantial gender earnings disparity. Lower earnings among women compared to men hospitalists were present in our data after controlling for age, pediatric specialty, practice model, geography, type of clinical work, and productivity measures. The gender earnings disparity noted in 1999[6] persists, although it appears to have decreased, possibly indicating progress toward equity. We showed that women hospitalists' relative tendency not to prioritize pay explains a significant portion of the residual income gap.

Hoff examined hospitalist earnings in a large national survey of hospitalists in 1999. Our cohorts differed in age and experience (both lower in the Hoff study). An estimated $24,000 ($124,266 vs. $148,132) earnings gap between women and men was greater than our estimate of $14,581 following an interval of 10 years. Although survey items differed, both studies found that women were less interested in pay than men when considering a hospitalist job. Hoff also found that work setting and attitudes about pay and lifestyle were significantly related to earnings. We extended the previous analyses of gender differences in job satisfaction priorities, work, and demographic characteristics to explain the earnings gap and understand how it may be remedied.

When considering job satisfaction, we found that more men than women prioritized substantial pay and that prioritization of substantial pay was directly related to higher earnings. Therefore, fewer women prioritizing pay partly explains women's lower earnings. Reasons for why fewer women prioritize pay were not assessed in this study but may include factors like being part of 2‐income households and competing commitments.[29] Priorities may even be influenced by empirically observed gender differences in discussions of financial matters, governed by cultural norms. Such norms may implicitly sanction employers to offer women less pay than men for the same or similar work. Women may disadvantage themselves by negotiating less or less well than men for higher starting and promotion salaries. They may be perceived more negatively than men when they do negotiate pay, leading to unintended negative consequences such as loss of social networks, decreased likability, and even loss of job offers.[29, 30, 31, 32]

More women prioritized optimal variety of tasks (6th most prevalent among women and 10th among men). Women who highly rate optimal variety of tasks as a job satisfier may choose positions in which they teach, perform research, and participate in hospital committees and quality‐improvement work, but offer lower pay. Yet hours per month doing nonclinical work was not significantly different between men and women, nor associated with earnings differences in our earnings models. Understanding whether women self‐select into hospitalist jobs with like‐minded colleagues to achieve complementary fit or end up supplementing their skills with hospitalists with different priorities may inform strategies to reduce the gender income disparity.[5] Unlike disparities between various hospital medicine groups, systematic disparities within practices risk generating low levels of organizational fairness and burnout among employees.

Not surprisingly, productivity was positively associated with earnings but did not fully account for the gender earnings gap. Our data demonstrated that women, on average, were associated with work characteristics that expectedly generate less compensation. For example, women were younger, more often part time, academic, pediatric, less often leaders, and reported fewer billable encounters compared to men. These differences account for some of the earnings gap between men and women, but these factors were controlled for in the earnings model. In addition, our analysis may have underestimated the gap by not incorporating loss of fringe benefits from part‐time status and not comprehensively counting incentive pay associated with high productivity. Other work patterns more commonly associated with women suggest an imbalance in reimbursement. More women than men work nights that are often compensated at higher rates than daytime work, yet their average pay was less, suggesting that compensation for night work may need to be adjusted to reflect its unique burdens and responsibilities.[33]

Although the gender pay gap was not more extreme among leaders compared to frontline hospitalists in our data, the trend, nonetheless, underscores an important consideration. Whereas clinical work is paid for in mostly measurable ways, pay for leadership may be influenced by intangible factors such as reputation, negotiation, and confidence that may disadvantage women relative to men.[7, 19, 23, 34, 35, 36] Efforts to overcome implicit gender bias should be most effective when we consciously couple fair promotion of women to leadership with fair compensation commensurate with their male peers.[21]

Our data are vulnerable to nonresponse bias.[1, 4, 5] Post hoc analyses demonstrated that distributions of age, gender, practice model and region of our respondents were similar to other nationally representative cohorts of hospitalists. Consequently, we believe our data can make valid estimates about a nationally representative sample of hospitalists. However, we acknowledge several additional weaknesses of self‐reported data, including recall bias and accuracy of productivity figures, which were rounded to variable significant digits by respondents. Earnings analysis using this data was intended to be exploratory, but the findings echoed analyses using more authoritative data sources.[11] Still, we made inferences conservatively by adopting multiple imputation techniques for dealing with nonresponse surveys in adherence to established reporting guidelines.[25] We also note several limitations relevant to multiple imputations. The greater prevalence of missing data for survey items soliciting earnings and the number of billable encounters suggest they were not truly missing at random as assumed. However, we showed that missingness is unrelated to the variables under study, justifying use of the technique. The wider measures of variance derived from multiple imputations make us vulnerable to not detecting associations that may exist.

The gender earnings gap found in hospital medicine echoes the gap found in multiple medical specialties, including but not limited to pediatrics, academic medicine, gastroenterology, and plastic surgery.[7, 8, 9, 11, 12, 13, 37] Hospital medicine employment models and practice patterns have important structural differences compared to previously studied populations that could mitigate factors contributing to women physicians' lower earnings. However, despite well‐defined working hours, lack of control over the number of patient encounters per day and high prevalence of hospital‐employed practice models, the gender earnings gap persists. We showed that lower prioritization for pay may reflect the self‐selection of women into lower paying jobs. Unmeasured factors, including implicit bias and differences in negotiations, social networks and mentoring opportunities[38, 39] may also contribute to pay differences between men and women hospitalists. As hospital medicine tackles gender inequities and other disparities, strategies to assess and address fair physician compensation must be on the table.

Hospitalists are a growing workforce numbering over 40,000 physicians, one‐third of whom are women.[1] Flexibility of work schedules and control over personal time have been the traditional selling points of the specialty.[2] Multiple studies of physician work life reveal growing physician dissatisfaction and a high prevalence of burnout.[3] To mitigate burnout risk, leaders in hospital medicine recognize the importance of creating a sustainable profession that offers both job and career satisfaction as well as work‐life balance and, importantly, fairness within the work environment.[4] Although success in some of these endeavors has been realized sporadically, sustaining work‐life balance and fairness in the specialty remains a work in progress, whereas evidence of high job attrition and pay inequities remain.[5, 6]

Pay inequity for women relative to men continues to be pervasive in medicine, including among early‐career physicians, researchers, and various specialists.[6, 7, 8, 9, 10, 11, 12, 13] The earnings gap seems to persist for physicians, even as federal efforts such as the Fair Pay Act of 2013 and the Paycheck Fairness Act of 2014 aim to end wage discrimination.[11, 14] Differences in specialty, part‐time status, and practice type do not mitigate the disparity.[8, 10, 15] Additional explanations have been proposed to explain the variability, including gender differences in negotiating skills, lack of opportunities to join networks of influence within organizations, and implicit or explicit bias and discrimination.[12, 16, 17, 18, 19, 20]

The earnings gap is also a consequence of what is commonly called the glass ceiling.[18, 19] Most agree that obstacles to fair advancement of women include absence of collaborative environments and role models who have successfully achieved work‐life balance.[17, 20, 21, 22] Somewhat surprisingly, women leaders in medicine seem to suffer greater income disparity than nonleaders; this income gap is prevalent among leaders in other elite professions as well.[7, 23] It is unknown whether women physicians' emphasis on work‐life balance, seen repeatedly in surveys, explains any of the pay disparity.[24] Little research to date has examined whether work‐life priorities of women in hospital medicine differ from men.

In this study, we sought to examine differences in job priorities between men and women hospitalists. In particular, we examine the relative prioritization of substantial pay to job satisfaction. We also examined gender differences in work patterns and earnings to explore potential sources of the persistent gender earnings gap.

METHODS

We analyzed data from the 20092010 Hospital Medicine Physician Worklife Survey, the design of which is detailed elsewhere.[4] Briefly, a 118‐item survey was administered by mail to a stratified sample of hospitalists from the Society of Hospital Medicine database and 3 large multisite hospitalist groups. A single survey item asked respondents to identify up to 4 out of 12 most important domains to their satisfaction with a hospitalist job. The domains were distilled from focus groups of nationally representative hospitalists as described previously,[4] and the survey item allowed up to one‐third of these domains to be identified as respondents' personal priorities. The list included: optimal variety of tasks, optimal workload, substantial pay, collegiality with other physicians, recognition by leaders, rewarding relationships with patients, satisfaction with nurses, optimal autonomy, control over personal time, fairness within organization, ample availability of resources to do job, and organizational climate of trust and belonging. We tabulated and ranked the frequency with which respondents selected each satisfaction domain by gender. Due to the nonstandard format of the survey item, we a priori decided to analyze only responses that were completed as instructed.

We also used demographic data including detailed work characteristics, clinical and nonclinical workload, total pretax earnings in 2009 as a hospitalist, and self‐identification as leader of their hospital medicine group. Respondent characteristics were tabulated and gender differences were tested using the t test, rank sum test, and the Fischer exact test as appropriate. We also listed the number of nonrespondents for each item. In estimating gender differences in earnings, we opted to use multiple‐imputation techniques to more conservatively account for greater variance inherent in the presence of missing data. Consistent with existing guidelines,[25] we demonstrated that item responses were not missing monotonically by visually inspecting patterns of nonresponse. We further demonstrated that data were missing at random by showing that response patterns of completed survey items did not predict whether or not a given variable response was missing using logistic regression models. We found no significant differences between respondents with complete and missing data. We verified that appropriate regression models for each variable on every other variable converged. We used Stata 13.1 (StataCorp, College Station, TX) to perform multiple imputations using chain equations (mi impute chain) to create 10 imputed tables for 7 normally distributed continuous variables using the ordinary least squares method, 3 non‐normally distributed variables using the predictive mean matching method, 2 nonordinal categorical variables using the multinomial logit method, and 1 binary variable using the logit method.[26, 27] Gender, pediatric specialty status, region of practice, and whether or not respondents prioritized substantial pay for job satisfaction were used as regular variables without missing data points.

Differences in earnings were assessed using a multivariate ordinary linear regression model applied to the imputed datasets fitted by forward selection of explanatory variables using P < 0.20 in bivariate analysis for inclusion and manual backward elimination of all statistically nonsignificant variables. We tested the significance of the women leader interaction term in the final parsimonious model. We used the usual significance threshold of P < 0.05 for inferences. Our analysis of publicly available anonymous data was exempt from IRB review.

RESULTS

Of the 816 survey respondents (response rate 25.6%), 40 either omitted the item soliciting work priorities or completed it incorrectly. Data from the remaining 776 respondents were used for the present analysis. Respondent characteristics are tabulated in Table 1. The characteristics of hospitalists by age, gender, specialty, practice model, and practice region were representative of US hospitalists from other surveys.[28]

Several gender differences were seen in the characteristics of hospitalists and their work (Table 1). Women compared to men hospitalists were less likely to be leaders, more likely to be pediatricians, work in university settings, and practice in Western states. Women compared to men, on average, were younger by 3 years, worked fewer full‐time equivalents (FTEs), worked a greater percentage of nights, and reported fewer billable encounters per shift. They were also more likely to be divorced or separated.

Job satisfaction priorities differed for women and men hospitalists. Table 2 lists job satisfaction domains in descending order of the frequency prioritized by men. The largest proportion of women and men prioritized optimal workload. However, although substantial pay was prioritized next most frequently by men, more women prioritized collegiality and control over personal time than substantial pay.

Key differences in individual characteristics, work patterns, and indicating substantial pay as a priority were associated with self‐reported total earnings in 2009 from respondents' work as a hospitalist. As shown in Table 3, the inclusion of detailed productivity measures such as FTE, days of monthly clinical work, and estimated number of daily billable encounters yielded a model that explained 33% of variance in earnings. After adjusting for significant covariates including pediatric specialty, practice model, geography, and amount and type of clinical work, the estimated underpayment of women compared to men was $14,581. Hospitalists who prioritized substantial pay earned $10,771 more than those who did not. The female x leader interaction term testing the hypothesis that gender disparity is greater among leaders than frontline hospitalists was not statistically significant ($16,720, P = 0.087) and excluded from the final model.

DISCUSSION

In a national stratified sample of US hospitalists, we found gender differences in job satisfaction priorities and hospitalist work characteristics. We also confirmed the persistence of a substantial gender earnings disparity. Lower earnings among women compared to men hospitalists were present in our data after controlling for age, pediatric specialty, practice model, geography, type of clinical work, and productivity measures. The gender earnings disparity noted in 1999[6] persists, although it appears to have decreased, possibly indicating progress toward equity. We showed that women hospitalists' relative tendency not to prioritize pay explains a significant portion of the residual income gap.

Hoff examined hospitalist earnings in a large national survey of hospitalists in 1999. Our cohorts differed in age and experience (both lower in the Hoff study). An estimated $24,000 ($124,266 vs. $148,132) earnings gap between women and men was greater than our estimate of $14,581 following an interval of 10 years. Although survey items differed, both studies found that women were less interested in pay than men when considering a hospitalist job. Hoff also found that work setting and attitudes about pay and lifestyle were significantly related to earnings. We extended the previous analyses of gender differences in job satisfaction priorities, work, and demographic characteristics to explain the earnings gap and understand how it may be remedied.

When considering job satisfaction, we found that more men than women prioritized substantial pay and that prioritization of substantial pay was directly related to higher earnings. Therefore, fewer women prioritizing pay partly explains women's lower earnings. Reasons for why fewer women prioritize pay were not assessed in this study but may include factors like being part of 2‐income households and competing commitments.[29] Priorities may even be influenced by empirically observed gender differences in discussions of financial matters, governed by cultural norms. Such norms may implicitly sanction employers to offer women less pay than men for the same or similar work. Women may disadvantage themselves by negotiating less or less well than men for higher starting and promotion salaries. They may be perceived more negatively than men when they do negotiate pay, leading to unintended negative consequences such as loss of social networks, decreased likability, and even loss of job offers.[29, 30, 31, 32]

More women prioritized optimal variety of tasks (6th most prevalent among women and 10th among men). Women who highly rate optimal variety of tasks as a job satisfier may choose positions in which they teach, perform research, and participate in hospital committees and quality‐improvement work, but offer lower pay. Yet hours per month doing nonclinical work was not significantly different between men and women, nor associated with earnings differences in our earnings models. Understanding whether women self‐select into hospitalist jobs with like‐minded colleagues to achieve complementary fit or end up supplementing their skills with hospitalists with different priorities may inform strategies to reduce the gender income disparity.[5] Unlike disparities between various hospital medicine groups, systematic disparities within practices risk generating low levels of organizational fairness and burnout among employees.

Not surprisingly, productivity was positively associated with earnings but did not fully account for the gender earnings gap. Our data demonstrated that women, on average, were associated with work characteristics that expectedly generate less compensation. For example, women were younger, more often part time, academic, pediatric, less often leaders, and reported fewer billable encounters compared to men. These differences account for some of the earnings gap between men and women, but these factors were controlled for in the earnings model. In addition, our analysis may have underestimated the gap by not incorporating loss of fringe benefits from part‐time status and not comprehensively counting incentive pay associated with high productivity. Other work patterns more commonly associated with women suggest an imbalance in reimbursement. More women than men work nights that are often compensated at higher rates than daytime work, yet their average pay was less, suggesting that compensation for night work may need to be adjusted to reflect its unique burdens and responsibilities.[33]

Although the gender pay gap was not more extreme among leaders compared to frontline hospitalists in our data, the trend, nonetheless, underscores an important consideration. Whereas clinical work is paid for in mostly measurable ways, pay for leadership may be influenced by intangible factors such as reputation, negotiation, and confidence that may disadvantage women relative to men.[7, 19, 23, 34, 35, 36] Efforts to overcome implicit gender bias should be most effective when we consciously couple fair promotion of women to leadership with fair compensation commensurate with their male peers.[21]

Our data are vulnerable to nonresponse bias.[1, 4, 5] Post hoc analyses demonstrated that distributions of age, gender, practice model and region of our respondents were similar to other nationally representative cohorts of hospitalists. Consequently, we believe our data can make valid estimates about a nationally representative sample of hospitalists. However, we acknowledge several additional weaknesses of self‐reported data, including recall bias and accuracy of productivity figures, which were rounded to variable significant digits by respondents. Earnings analysis using this data was intended to be exploratory, but the findings echoed analyses using more authoritative data sources.[11] Still, we made inferences conservatively by adopting multiple imputation techniques for dealing with nonresponse surveys in adherence to established reporting guidelines.[25] We also note several limitations relevant to multiple imputations. The greater prevalence of missing data for survey items soliciting earnings and the number of billable encounters suggest they were not truly missing at random as assumed. However, we showed that missingness is unrelated to the variables under study, justifying use of the technique. The wider measures of variance derived from multiple imputations make us vulnerable to not detecting associations that may exist.

The gender earnings gap found in hospital medicine echoes the gap found in multiple medical specialties, including but not limited to pediatrics, academic medicine, gastroenterology, and plastic surgery.[7, 8, 9, 11, 12, 13, 37] Hospital medicine employment models and practice patterns have important structural differences compared to previously studied populations that could mitigate factors contributing to women physicians' lower earnings. However, despite well‐defined working hours, lack of control over the number of patient encounters per day and high prevalence of hospital‐employed practice models, the gender earnings gap persists. We showed that lower prioritization for pay may reflect the self‐selection of women into lower paying jobs. Unmeasured factors, including implicit bias and differences in negotiations, social networks and mentoring opportunities[38, 39] may also contribute to pay differences between men and women hospitalists. As hospital medicine tackles gender inequities and other disparities, strategies to assess and address fair physician compensation must be on the table.

Hospitalists are a growing workforce numbering over 40,000 physicians, one‐third of whom are women.[1] Flexibility of work schedules and control over personal time have been the traditional selling points of the specialty.[2] Multiple studies of physician work life reveal growing physician dissatisfaction and a high prevalence of burnout.[3] To mitigate burnout risk, leaders in hospital medicine recognize the importance of creating a sustainable profession that offers both job and career satisfaction as well as work‐life balance and, importantly, fairness within the work environment.[4] Although success in some of these endeavors has been realized sporadically, sustaining work‐life balance and fairness in the specialty remains a work in progress, whereas evidence of high job attrition and pay inequities remain.[5, 6]

Pay inequity for women relative to men continues to be pervasive in medicine, including among early‐career physicians, researchers, and various specialists.[6, 7, 8, 9, 10, 11, 12, 13] The earnings gap seems to persist for physicians, even as federal efforts such as the Fair Pay Act of 2013 and the Paycheck Fairness Act of 2014 aim to end wage discrimination.[11, 14] Differences in specialty, part‐time status, and practice type do not mitigate the disparity.[8, 10, 15] Additional explanations have been proposed to explain the variability, including gender differences in negotiating skills, lack of opportunities to join networks of influence within organizations, and implicit or explicit bias and discrimination.[12, 16, 17, 18, 19, 20]

The earnings gap is also a consequence of what is commonly called the glass ceiling.[18, 19] Most agree that obstacles to fair advancement of women include absence of collaborative environments and role models who have successfully achieved work‐life balance.[17, 20, 21, 22] Somewhat surprisingly, women leaders in medicine seem to suffer greater income disparity than nonleaders; this income gap is prevalent among leaders in other elite professions as well.[7, 23] It is unknown whether women physicians' emphasis on work‐life balance, seen repeatedly in surveys, explains any of the pay disparity.[24] Little research to date has examined whether work‐life priorities of women in hospital medicine differ from men.

In this study, we sought to examine differences in job priorities between men and women hospitalists. In particular, we examine the relative prioritization of substantial pay to job satisfaction. We also examined gender differences in work patterns and earnings to explore potential sources of the persistent gender earnings gap.

METHODS

We analyzed data from the 20092010 Hospital Medicine Physician Worklife Survey, the design of which is detailed elsewhere.[4] Briefly, a 118‐item survey was administered by mail to a stratified sample of hospitalists from the Society of Hospital Medicine database and 3 large multisite hospitalist groups. A single survey item asked respondents to identify up to 4 out of 12 most important domains to their satisfaction with a hospitalist job. The domains were distilled from focus groups of nationally representative hospitalists as described previously,[4] and the survey item allowed up to one‐third of these domains to be identified as respondents' personal priorities. The list included: optimal variety of tasks, optimal workload, substantial pay, collegiality with other physicians, recognition by leaders, rewarding relationships with patients, satisfaction with nurses, optimal autonomy, control over personal time, fairness within organization, ample availability of resources to do job, and organizational climate of trust and belonging. We tabulated and ranked the frequency with which respondents selected each satisfaction domain by gender. Due to the nonstandard format of the survey item, we a priori decided to analyze only responses that were completed as instructed.

We also used demographic data including detailed work characteristics, clinical and nonclinical workload, total pretax earnings in 2009 as a hospitalist, and self‐identification as leader of their hospital medicine group. Respondent characteristics were tabulated and gender differences were tested using the t test, rank sum test, and the Fischer exact test as appropriate. We also listed the number of nonrespondents for each item. In estimating gender differences in earnings, we opted to use multiple‐imputation techniques to more conservatively account for greater variance inherent in the presence of missing data. Consistent with existing guidelines,[25] we demonstrated that item responses were not missing monotonically by visually inspecting patterns of nonresponse. We further demonstrated that data were missing at random by showing that response patterns of completed survey items did not predict whether or not a given variable response was missing using logistic regression models. We found no significant differences between respondents with complete and missing data. We verified that appropriate regression models for each variable on every other variable converged. We used Stata 13.1 (StataCorp, College Station, TX) to perform multiple imputations using chain equations (mi impute chain) to create 10 imputed tables for 7 normally distributed continuous variables using the ordinary least squares method, 3 non‐normally distributed variables using the predictive mean matching method, 2 nonordinal categorical variables using the multinomial logit method, and 1 binary variable using the logit method.[26, 27] Gender, pediatric specialty status, region of practice, and whether or not respondents prioritized substantial pay for job satisfaction were used as regular variables without missing data points.

Differences in earnings were assessed using a multivariate ordinary linear regression model applied to the imputed datasets fitted by forward selection of explanatory variables using P < 0.20 in bivariate analysis for inclusion and manual backward elimination of all statistically nonsignificant variables. We tested the significance of the women leader interaction term in the final parsimonious model. We used the usual significance threshold of P < 0.05 for inferences. Our analysis of publicly available anonymous data was exempt from IRB review.

RESULTS

Of the 816 survey respondents (response rate 25.6%), 40 either omitted the item soliciting work priorities or completed it incorrectly. Data from the remaining 776 respondents were used for the present analysis. Respondent characteristics are tabulated in Table 1. The characteristics of hospitalists by age, gender, specialty, practice model, and practice region were representative of US hospitalists from other surveys.[28]