Daniel L. Roberts, MD

The advent of work‐hour restrictions and admission limits for teaching services has led many academic hospitals to implement hospitalist‐run staff (ie, nonteaching) services.[1] Although this practice is not new,[2] it is growing in popularity[3] and has been endorsed as a way to protect resident teaching and prevent excessive workload.[4] One potential benefit is the assignment of more educational cases to teaching services, whereas the nonteaching services receive more patients whose care is presumably relatively mundane or routine.[5]

Despite the rapid growth of this system of educational triage,[6] little is known about the factors considered when teaching versus nonteaching decisions are made. Studies of clinical outcomes for patients assigned to teaching versus nonteaching services have understandably used random assignment,[7, 8] whereas a study finding that patients with unhealthy substance use were more likely to be on teaching services than nonteaching services relied on patient assignment based on the identity of the patient's primary care provider or insurer.[9] In 2009, O'Connor et al. reported that implementation of nonteaching services at 2 hospitals had led to unequal distribution of patients in terms of demographics, diagnosis, and illness severity.[10] Triage decisions were made by either a nurse coordinator or a medical chief resident, and sicker patients (and occasionally good teaching cases) were preferentially placed on the teaching services, reportedly out of respect for the comfort level of the midlevel providers who staffed the nonteaching services.

Our institution has used a system of hospitalist educational triage since 1998. Over that time, residents have often expressed concerns about the assignment of patients to the teaching services, reporting in particular that they receive a disproportionate number of complex cases and outside transfers. In 2006, the hospitalist group attempted to address these concerns by collecting real‐time admission data, but the application of the data was limited by suspicion on both sides of a Hawthorne effect (data not published).

If trainee and hospitalist expectations for what constitutes a great teaching case differ substantially, that difference can have significant implications for resident and medical student teaching, self‐perceived roles, and satisfaction. More significantly, an understanding of what faculty perceive as ideal teaching cases would provide valuable information about the strengths and weaknesses of the teachingnonteaching model, which may prove useful to other academic institutions considering such a system. In this study, we endeavored to understand what residents and hospitalists consider an educational admission and to compare these expectations to the actual triage decisions of hospitalists.

METHODS

Mayo Clinic Hospital (Phoenix, Arizona) has used separate teaching and nonteaching services since opening in 1998. At our institution, like many others,[11] a hospitalist is assigned to take all calls for emergency department (ED) admissions, admissions from outpatient clinics, and transfer requests; this physician directs patients to the teaching or nonteaching service. At the time of our study, the 2 teaching services alternated days in which they admitted up to 7 patients, and the 5 nonteaching services admitted all other patients and provided medicine consultative services for the hospital. Teaching services consisted of 1 hospitalist, 2 senior residents, 2 or 3 first‐year residents, and sometimes 1 third‐ or fourth‐year medical student. Nonteaching services consisted of a hospitalist with intermittent assistance from a physician assistant or nurse practitioner.

Although there are no formal guidelines for the hospitalist triage role, hospitalists are encouraged to assign more educational cases to the teaching services and to allow the residents enough time to address the acute needs of the prior admission before receiving the next admission. Residents are not assigned any patients between 4:00 _am and 7:00 _am. The goals and objectives for the resident rotation on the medicine teaching service include a list of diagnoses with which residents are expected to become familiar during their residency; triage hospitalists have on‐line access to these goals and objectives.

To assess resident and hospitalist opinions about what types of patients should or should not be admitted to teaching services and to compare those characteristics with those of the patients actually admitted to teaching services, we began by administering a simple, open‐ended survey and asked both groups: (1) In an ideal world, what kinds of patients should be admitted to the internal medicine teaching services at Mayo Clinic Hospital? (2) In the real world, what kinds of patients are admitted to the internal medicine teaching services at Mayo Clinic Hospital?

Ample space was provided for free‐text entries. Residents were additionally asked their postgraduate year level. The survey was administered in April 2011, at which time all residents would have rotated on the medicine teaching services several times. Survey responses were anonymous and were compiled and retyped by someone unfamiliar with the subjects' handwriting.

Two authors (D.L.R. and H.R.L.) reviewed the results of the first survey and used conventional content analysis to group responses into categories and tally them.[12] Responses from hospitalists and residents were used to determine the content for a second, quantitative survey that asked respondents to rate specific possible factors that affected triage decisions on a Likert scale from 1 (Argues against teaching admission) to 5 (Argues for teaching admission). The second survey, administered to the same residents and hospitalists in May 2011, asked: (1) In an ideal world, how do these factors contribute to the decision about which patients should be admitted to the internal medicine teaching services at Mayo Clinic Hospital? (2) In the real world, how do these factors contribute to the decision about which patients are admitted to the internal medicine teaching services at Mayo Clinic Hospital?

Assuming a 3:1 ratio of nonteaching to teaching admissions, we calculated that we would need to analyze 1028 admissions to detect a 10% difference in the proportion of a specific trait present in 50% of patients admitted to the nonteaching service, with the use of a 2‐sided test with 80% statistical power and a significance level of 0.05.

We collected data on patient assignment via retrospective chart review to avoid the possibility of a Hawthorne effect. We studied all admissions to the internal medicine services for a 3‐month period before the administration of the first survey (January 1, 2011 through March 31, 2011). The following patient data were collected: service assignment (teaching vs nonteaching), age, sex, source of admission (ED, direct from clinic, outside transfer, internal transfer from another hospital service), first visit to our institution, prior hematology or oncology visit at our institution (as a surrogate for cancer), prior psychiatry visit at our institution (as a surrogate for psychiatric disease), transplantation history, human immunodeficiency virus (HIV) or acquired immune deficiency syndrome (AIDS) history, chronic or functional pain mentioned in ED or admission note, need for translator, and benefactor status. Additionally, an online calculator was used to determine the Charlson Comorbidity Index score for each patient.[13] We collected actual patient data corresponding to factors reported by survey respondents whenever possible and practical, but not every factor reported by survey respondents was amenable to rigorous analysis; for example, no unbiased method could be devised to rigorously categorize patients whose admissions are likely to take more time or difficult patients and families.

Responses to the second (quantitative) survey and patient data were compared using the Pearson [2] and Fisher exact test for categorical variables and the Student t test or Wilcoxon rank sum test for continuous variables. Categorical variables that achieved statistical significance for overall difference were analyzed on a post hoc basis using the Bonferroni method to control for the overall type I error rate. We also examined the differences between actual and ideal triage decisions using the Wilcoxon signed rank test. Data were analyzed using SAS 9.3 (SAS Institute, Inc., Cary, NC). Statistical significance was defined as P<0.05.

The project was deemed exempt by the Mayo Clinic institutional review board.

RESULTS

We surveyed all categorical internal medicine residents (n=30, 10 each from postgraduate year [PGY]‐1, PGY‐2, and PGY‐3) and hospitalists except the authors (n=21; average years since completing training=13.3; range, 129 years). For both surveys, responses were collected from 29 (96.7%) residents. The nonresponding resident was a PGY‐2. The response rate for hospitalists was 20/21 (95.2%) for the first survey and 16/21 (76.2%) for the second survey.

DISCUSSION

The results of our qualitative and quantitative surveys showed significant differences between resident and staff perceptions of the faculty triage role. Although both groups similarly valued many features, residents expressed a clear preference for more bread‐and‐butter admissions, whereas the staff prioritized selecting the most complex, challenging, and rare cases from among the day's admissions to give to the residents. (Residents were also very interested in rare cases, suggesting that they saw benefit to admitting patients with a variety of degrees of rarity and complexity.) Residents and faculty seemed to agree that the number of social admissions and outside transfers admitted to teaching services was not ideal.

These perceptions have substantial implications. If the current triage process is to continue, there may be benefit to designing a faculty development project focused on the triage process, which previously has been largely unexamined. Efforts to remove or limit time barriers that prevent perceived educational cases from being admitted to teaching services is also a worthy endeavor (eg, structuring the 2 teams to admit simultaneously so that teaching teams can admit patients back to back without exceeding capacity). In addition, residents may benefit from teaching hospitalists who concentrate educational efforts on the learning that can be extracted from the care of any patient, including admissions that initially seem mundane or purely social.[14] A concerted effort to divert more traditional medicine admissions and fewer unusual cases to the teaching service might improve resident perceptions of the triage process. Further, although the care of any patient can have education benefit, the fact that both groups perceived excessive social admissions in the teaching service suggests that a potential benefit of a nonteaching service (ie, absorbing the most mundane admissions) may not yet be fully realized.

Despite the perceived differences noted on the surveys, we found remarkably few differences between patients admitted to the teaching and nonteaching services. Although both groups rated complexity; outside transfers; being seen at the institution for the first time; and histories of transplantation, cancer, chronic or functional pain, and psychiatric disease as increasing the likelihood of admission to a teaching service, no differences were observed for these factors or their quantifiable surrogates. (Although the overall test for admission type achieved marginal statistical significance, none of the individual admission types were significantly different in post hoc analysis.) Residents, but not faculty, thought that older patients were over‐represented on the teaching service, but their assigned patients were significantly younger than those on the nonteaching service.

These findings have several possible explanations. First, although most hospitalists spend time on teaching and nonteaching services (and therefore are familiar with the patient composition of each), residents get very little exposure to the nonteaching services (until they are senior residents with a rotation on a consulting service). Their impression of inequity may be due to misunderstanding the patient composition of the nonteaching services. Second, the mere existence of a triage role may create false expectations about patient composition; that is, simply by knowing that every admission was chosen for its educational merit, residents may have disproportionate perceptions about those cases judged to have less educational value, even ifas our data suggestassignments to teaching versus nonteaching services are occurring fairly equitably.

CONCLUSION

Triage hospitalists distributed patients fairly evenly between teaching and nonteaching services, but residents and faculty alike perceived that residents would benefit from more bread‐and‐butter cases. Hospitals considering the addition of a nonteaching service may want to incorporate a faculty development project focused on the triage process to ensure that these traditional medicine cases are assigned to resident services and to ensure that the great teaching case is not considered such because of complexity and acuity alone.

The advent of work‐hour restrictions and admission limits for teaching services has led many academic hospitals to implement hospitalist‐run staff (ie, nonteaching) services.[1] Although this practice is not new,[2] it is growing in popularity[3] and has been endorsed as a way to protect resident teaching and prevent excessive workload.[4] One potential benefit is the assignment of more educational cases to teaching services, whereas the nonteaching services receive more patients whose care is presumably relatively mundane or routine.[5]

Despite the rapid growth of this system of educational triage,[6] little is known about the factors considered when teaching versus nonteaching decisions are made. Studies of clinical outcomes for patients assigned to teaching versus nonteaching services have understandably used random assignment,[7, 8] whereas a study finding that patients with unhealthy substance use were more likely to be on teaching services than nonteaching services relied on patient assignment based on the identity of the patient's primary care provider or insurer.[9] In 2009, O'Connor et al. reported that implementation of nonteaching services at 2 hospitals had led to unequal distribution of patients in terms of demographics, diagnosis, and illness severity.[10] Triage decisions were made by either a nurse coordinator or a medical chief resident, and sicker patients (and occasionally good teaching cases) were preferentially placed on the teaching services, reportedly out of respect for the comfort level of the midlevel providers who staffed the nonteaching services.

Our institution has used a system of hospitalist educational triage since 1998. Over that time, residents have often expressed concerns about the assignment of patients to the teaching services, reporting in particular that they receive a disproportionate number of complex cases and outside transfers. In 2006, the hospitalist group attempted to address these concerns by collecting real‐time admission data, but the application of the data was limited by suspicion on both sides of a Hawthorne effect (data not published).

If trainee and hospitalist expectations for what constitutes a great teaching case differ substantially, that difference can have significant implications for resident and medical student teaching, self‐perceived roles, and satisfaction. More significantly, an understanding of what faculty perceive as ideal teaching cases would provide valuable information about the strengths and weaknesses of the teachingnonteaching model, which may prove useful to other academic institutions considering such a system. In this study, we endeavored to understand what residents and hospitalists consider an educational admission and to compare these expectations to the actual triage decisions of hospitalists.

METHODS

Mayo Clinic Hospital (Phoenix, Arizona) has used separate teaching and nonteaching services since opening in 1998. At our institution, like many others,[11] a hospitalist is assigned to take all calls for emergency department (ED) admissions, admissions from outpatient clinics, and transfer requests; this physician directs patients to the teaching or nonteaching service. At the time of our study, the 2 teaching services alternated days in which they admitted up to 7 patients, and the 5 nonteaching services admitted all other patients and provided medicine consultative services for the hospital. Teaching services consisted of 1 hospitalist, 2 senior residents, 2 or 3 first‐year residents, and sometimes 1 third‐ or fourth‐year medical student. Nonteaching services consisted of a hospitalist with intermittent assistance from a physician assistant or nurse practitioner.

Although there are no formal guidelines for the hospitalist triage role, hospitalists are encouraged to assign more educational cases to the teaching services and to allow the residents enough time to address the acute needs of the prior admission before receiving the next admission. Residents are not assigned any patients between 4:00 _am and 7:00 _am. The goals and objectives for the resident rotation on the medicine teaching service include a list of diagnoses with which residents are expected to become familiar during their residency; triage hospitalists have on‐line access to these goals and objectives.

To assess resident and hospitalist opinions about what types of patients should or should not be admitted to teaching services and to compare those characteristics with those of the patients actually admitted to teaching services, we began by administering a simple, open‐ended survey and asked both groups: (1) In an ideal world, what kinds of patients should be admitted to the internal medicine teaching services at Mayo Clinic Hospital? (2) In the real world, what kinds of patients are admitted to the internal medicine teaching services at Mayo Clinic Hospital?

Ample space was provided for free‐text entries. Residents were additionally asked their postgraduate year level. The survey was administered in April 2011, at which time all residents would have rotated on the medicine teaching services several times. Survey responses were anonymous and were compiled and retyped by someone unfamiliar with the subjects' handwriting.

Two authors (D.L.R. and H.R.L.) reviewed the results of the first survey and used conventional content analysis to group responses into categories and tally them.[12] Responses from hospitalists and residents were used to determine the content for a second, quantitative survey that asked respondents to rate specific possible factors that affected triage decisions on a Likert scale from 1 (Argues against teaching admission) to 5 (Argues for teaching admission). The second survey, administered to the same residents and hospitalists in May 2011, asked: (1) In an ideal world, how do these factors contribute to the decision about which patients should be admitted to the internal medicine teaching services at Mayo Clinic Hospital? (2) In the real world, how do these factors contribute to the decision about which patients are admitted to the internal medicine teaching services at Mayo Clinic Hospital?

Assuming a 3:1 ratio of nonteaching to teaching admissions, we calculated that we would need to analyze 1028 admissions to detect a 10% difference in the proportion of a specific trait present in 50% of patients admitted to the nonteaching service, with the use of a 2‐sided test with 80% statistical power and a significance level of 0.05.

We collected data on patient assignment via retrospective chart review to avoid the possibility of a Hawthorne effect. We studied all admissions to the internal medicine services for a 3‐month period before the administration of the first survey (January 1, 2011 through March 31, 2011). The following patient data were collected: service assignment (teaching vs nonteaching), age, sex, source of admission (ED, direct from clinic, outside transfer, internal transfer from another hospital service), first visit to our institution, prior hematology or oncology visit at our institution (as a surrogate for cancer), prior psychiatry visit at our institution (as a surrogate for psychiatric disease), transplantation history, human immunodeficiency virus (HIV) or acquired immune deficiency syndrome (AIDS) history, chronic or functional pain mentioned in ED or admission note, need for translator, and benefactor status. Additionally, an online calculator was used to determine the Charlson Comorbidity Index score for each patient.[13] We collected actual patient data corresponding to factors reported by survey respondents whenever possible and practical, but not every factor reported by survey respondents was amenable to rigorous analysis; for example, no unbiased method could be devised to rigorously categorize patients whose admissions are likely to take more time or difficult patients and families.

Responses to the second (quantitative) survey and patient data were compared using the Pearson [2] and Fisher exact test for categorical variables and the Student t test or Wilcoxon rank sum test for continuous variables. Categorical variables that achieved statistical significance for overall difference were analyzed on a post hoc basis using the Bonferroni method to control for the overall type I error rate. We also examined the differences between actual and ideal triage decisions using the Wilcoxon signed rank test. Data were analyzed using SAS 9.3 (SAS Institute, Inc., Cary, NC). Statistical significance was defined as P<0.05.

The project was deemed exempt by the Mayo Clinic institutional review board.

RESULTS

We surveyed all categorical internal medicine residents (n=30, 10 each from postgraduate year [PGY]‐1, PGY‐2, and PGY‐3) and hospitalists except the authors (n=21; average years since completing training=13.3; range, 129 years). For both surveys, responses were collected from 29 (96.7%) residents. The nonresponding resident was a PGY‐2. The response rate for hospitalists was 20/21 (95.2%) for the first survey and 16/21 (76.2%) for the second survey.

DISCUSSION

The results of our qualitative and quantitative surveys showed significant differences between resident and staff perceptions of the faculty triage role. Although both groups similarly valued many features, residents expressed a clear preference for more bread‐and‐butter admissions, whereas the staff prioritized selecting the most complex, challenging, and rare cases from among the day's admissions to give to the residents. (Residents were also very interested in rare cases, suggesting that they saw benefit to admitting patients with a variety of degrees of rarity and complexity.) Residents and faculty seemed to agree that the number of social admissions and outside transfers admitted to teaching services was not ideal.

These perceptions have substantial implications. If the current triage process is to continue, there may be benefit to designing a faculty development project focused on the triage process, which previously has been largely unexamined. Efforts to remove or limit time barriers that prevent perceived educational cases from being admitted to teaching services is also a worthy endeavor (eg, structuring the 2 teams to admit simultaneously so that teaching teams can admit patients back to back without exceeding capacity). In addition, residents may benefit from teaching hospitalists who concentrate educational efforts on the learning that can be extracted from the care of any patient, including admissions that initially seem mundane or purely social.[14] A concerted effort to divert more traditional medicine admissions and fewer unusual cases to the teaching service might improve resident perceptions of the triage process. Further, although the care of any patient can have education benefit, the fact that both groups perceived excessive social admissions in the teaching service suggests that a potential benefit of a nonteaching service (ie, absorbing the most mundane admissions) may not yet be fully realized.

Despite the perceived differences noted on the surveys, we found remarkably few differences between patients admitted to the teaching and nonteaching services. Although both groups rated complexity; outside transfers; being seen at the institution for the first time; and histories of transplantation, cancer, chronic or functional pain, and psychiatric disease as increasing the likelihood of admission to a teaching service, no differences were observed for these factors or their quantifiable surrogates. (Although the overall test for admission type achieved marginal statistical significance, none of the individual admission types were significantly different in post hoc analysis.) Residents, but not faculty, thought that older patients were over‐represented on the teaching service, but their assigned patients were significantly younger than those on the nonteaching service.

These findings have several possible explanations. First, although most hospitalists spend time on teaching and nonteaching services (and therefore are familiar with the patient composition of each), residents get very little exposure to the nonteaching services (until they are senior residents with a rotation on a consulting service). Their impression of inequity may be due to misunderstanding the patient composition of the nonteaching services. Second, the mere existence of a triage role may create false expectations about patient composition; that is, simply by knowing that every admission was chosen for its educational merit, residents may have disproportionate perceptions about those cases judged to have less educational value, even ifas our data suggestassignments to teaching versus nonteaching services are occurring fairly equitably.

CONCLUSION

Triage hospitalists distributed patients fairly evenly between teaching and nonteaching services, but residents and faculty alike perceived that residents would benefit from more bread‐and‐butter cases. Hospitals considering the addition of a nonteaching service may want to incorporate a faculty development project focused on the triage process to ensure that these traditional medicine cases are assigned to resident services and to ensure that the great teaching case is not considered such because of complexity and acuity alone.

The advent of work‐hour restrictions and admission limits for teaching services has led many academic hospitals to implement hospitalist‐run staff (ie, nonteaching) services.[1] Although this practice is not new,[2] it is growing in popularity[3] and has been endorsed as a way to protect resident teaching and prevent excessive workload.[4] One potential benefit is the assignment of more educational cases to teaching services, whereas the nonteaching services receive more patients whose care is presumably relatively mundane or routine.[5]

Despite the rapid growth of this system of educational triage,[6] little is known about the factors considered when teaching versus nonteaching decisions are made. Studies of clinical outcomes for patients assigned to teaching versus nonteaching services have understandably used random assignment,[7, 8] whereas a study finding that patients with unhealthy substance use were more likely to be on teaching services than nonteaching services relied on patient assignment based on the identity of the patient's primary care provider or insurer.[9] In 2009, O'Connor et al. reported that implementation of nonteaching services at 2 hospitals had led to unequal distribution of patients in terms of demographics, diagnosis, and illness severity.[10] Triage decisions were made by either a nurse coordinator or a medical chief resident, and sicker patients (and occasionally good teaching cases) were preferentially placed on the teaching services, reportedly out of respect for the comfort level of the midlevel providers who staffed the nonteaching services.

Our institution has used a system of hospitalist educational triage since 1998. Over that time, residents have often expressed concerns about the assignment of patients to the teaching services, reporting in particular that they receive a disproportionate number of complex cases and outside transfers. In 2006, the hospitalist group attempted to address these concerns by collecting real‐time admission data, but the application of the data was limited by suspicion on both sides of a Hawthorne effect (data not published).

If trainee and hospitalist expectations for what constitutes a great teaching case differ substantially, that difference can have significant implications for resident and medical student teaching, self‐perceived roles, and satisfaction. More significantly, an understanding of what faculty perceive as ideal teaching cases would provide valuable information about the strengths and weaknesses of the teachingnonteaching model, which may prove useful to other academic institutions considering such a system. In this study, we endeavored to understand what residents and hospitalists consider an educational admission and to compare these expectations to the actual triage decisions of hospitalists.

METHODS

Mayo Clinic Hospital (Phoenix, Arizona) has used separate teaching and nonteaching services since opening in 1998. At our institution, like many others,[11] a hospitalist is assigned to take all calls for emergency department (ED) admissions, admissions from outpatient clinics, and transfer requests; this physician directs patients to the teaching or nonteaching service. At the time of our study, the 2 teaching services alternated days in which they admitted up to 7 patients, and the 5 nonteaching services admitted all other patients and provided medicine consultative services for the hospital. Teaching services consisted of 1 hospitalist, 2 senior residents, 2 or 3 first‐year residents, and sometimes 1 third‐ or fourth‐year medical student. Nonteaching services consisted of a hospitalist with intermittent assistance from a physician assistant or nurse practitioner.

Although there are no formal guidelines for the hospitalist triage role, hospitalists are encouraged to assign more educational cases to the teaching services and to allow the residents enough time to address the acute needs of the prior admission before receiving the next admission. Residents are not assigned any patients between 4:00 _am and 7:00 _am. The goals and objectives for the resident rotation on the medicine teaching service include a list of diagnoses with which residents are expected to become familiar during their residency; triage hospitalists have on‐line access to these goals and objectives.

To assess resident and hospitalist opinions about what types of patients should or should not be admitted to teaching services and to compare those characteristics with those of the patients actually admitted to teaching services, we began by administering a simple, open‐ended survey and asked both groups: (1) In an ideal world, what kinds of patients should be admitted to the internal medicine teaching services at Mayo Clinic Hospital? (2) In the real world, what kinds of patients are admitted to the internal medicine teaching services at Mayo Clinic Hospital?

Ample space was provided for free‐text entries. Residents were additionally asked their postgraduate year level. The survey was administered in April 2011, at which time all residents would have rotated on the medicine teaching services several times. Survey responses were anonymous and were compiled and retyped by someone unfamiliar with the subjects' handwriting.

Two authors (D.L.R. and H.R.L.) reviewed the results of the first survey and used conventional content analysis to group responses into categories and tally them.[12] Responses from hospitalists and residents were used to determine the content for a second, quantitative survey that asked respondents to rate specific possible factors that affected triage decisions on a Likert scale from 1 (Argues against teaching admission) to 5 (Argues for teaching admission). The second survey, administered to the same residents and hospitalists in May 2011, asked: (1) In an ideal world, how do these factors contribute to the decision about which patients should be admitted to the internal medicine teaching services at Mayo Clinic Hospital? (2) In the real world, how do these factors contribute to the decision about which patients are admitted to the internal medicine teaching services at Mayo Clinic Hospital?

Assuming a 3:1 ratio of nonteaching to teaching admissions, we calculated that we would need to analyze 1028 admissions to detect a 10% difference in the proportion of a specific trait present in 50% of patients admitted to the nonteaching service, with the use of a 2‐sided test with 80% statistical power and a significance level of 0.05.

We collected data on patient assignment via retrospective chart review to avoid the possibility of a Hawthorne effect. We studied all admissions to the internal medicine services for a 3‐month period before the administration of the first survey (January 1, 2011 through March 31, 2011). The following patient data were collected: service assignment (teaching vs nonteaching), age, sex, source of admission (ED, direct from clinic, outside transfer, internal transfer from another hospital service), first visit to our institution, prior hematology or oncology visit at our institution (as a surrogate for cancer), prior psychiatry visit at our institution (as a surrogate for psychiatric disease), transplantation history, human immunodeficiency virus (HIV) or acquired immune deficiency syndrome (AIDS) history, chronic or functional pain mentioned in ED or admission note, need for translator, and benefactor status. Additionally, an online calculator was used to determine the Charlson Comorbidity Index score for each patient.[13] We collected actual patient data corresponding to factors reported by survey respondents whenever possible and practical, but not every factor reported by survey respondents was amenable to rigorous analysis; for example, no unbiased method could be devised to rigorously categorize patients whose admissions are likely to take more time or difficult patients and families.

Responses to the second (quantitative) survey and patient data were compared using the Pearson [2] and Fisher exact test for categorical variables and the Student t test or Wilcoxon rank sum test for continuous variables. Categorical variables that achieved statistical significance for overall difference were analyzed on a post hoc basis using the Bonferroni method to control for the overall type I error rate. We also examined the differences between actual and ideal triage decisions using the Wilcoxon signed rank test. Data were analyzed using SAS 9.3 (SAS Institute, Inc., Cary, NC). Statistical significance was defined as P<0.05.

The project was deemed exempt by the Mayo Clinic institutional review board.

RESULTS

We surveyed all categorical internal medicine residents (n=30, 10 each from postgraduate year [PGY]‐1, PGY‐2, and PGY‐3) and hospitalists except the authors (n=21; average years since completing training=13.3; range, 129 years). For both surveys, responses were collected from 29 (96.7%) residents. The nonresponding resident was a PGY‐2. The response rate for hospitalists was 20/21 (95.2%) for the first survey and 16/21 (76.2%) for the second survey.

DISCUSSION

The results of our qualitative and quantitative surveys showed significant differences between resident and staff perceptions of the faculty triage role. Although both groups similarly valued many features, residents expressed a clear preference for more bread‐and‐butter admissions, whereas the staff prioritized selecting the most complex, challenging, and rare cases from among the day's admissions to give to the residents. (Residents were also very interested in rare cases, suggesting that they saw benefit to admitting patients with a variety of degrees of rarity and complexity.) Residents and faculty seemed to agree that the number of social admissions and outside transfers admitted to teaching services was not ideal.

These perceptions have substantial implications. If the current triage process is to continue, there may be benefit to designing a faculty development project focused on the triage process, which previously has been largely unexamined. Efforts to remove or limit time barriers that prevent perceived educational cases from being admitted to teaching services is also a worthy endeavor (eg, structuring the 2 teams to admit simultaneously so that teaching teams can admit patients back to back without exceeding capacity). In addition, residents may benefit from teaching hospitalists who concentrate educational efforts on the learning that can be extracted from the care of any patient, including admissions that initially seem mundane or purely social.[14] A concerted effort to divert more traditional medicine admissions and fewer unusual cases to the teaching service might improve resident perceptions of the triage process. Further, although the care of any patient can have education benefit, the fact that both groups perceived excessive social admissions in the teaching service suggests that a potential benefit of a nonteaching service (ie, absorbing the most mundane admissions) may not yet be fully realized.

Despite the perceived differences noted on the surveys, we found remarkably few differences between patients admitted to the teaching and nonteaching services. Although both groups rated complexity; outside transfers; being seen at the institution for the first time; and histories of transplantation, cancer, chronic or functional pain, and psychiatric disease as increasing the likelihood of admission to a teaching service, no differences were observed for these factors or their quantifiable surrogates. (Although the overall test for admission type achieved marginal statistical significance, none of the individual admission types were significantly different in post hoc analysis.) Residents, but not faculty, thought that older patients were over‐represented on the teaching service, but their assigned patients were significantly younger than those on the nonteaching service.

These findings have several possible explanations. First, although most hospitalists spend time on teaching and nonteaching services (and therefore are familiar with the patient composition of each), residents get very little exposure to the nonteaching services (until they are senior residents with a rotation on a consulting service). Their impression of inequity may be due to misunderstanding the patient composition of the nonteaching services. Second, the mere existence of a triage role may create false expectations about patient composition; that is, simply by knowing that every admission was chosen for its educational merit, residents may have disproportionate perceptions about those cases judged to have less educational value, even ifas our data suggestassignments to teaching versus nonteaching services are occurring fairly equitably.

CONCLUSION

Triage hospitalists distributed patients fairly evenly between teaching and nonteaching services, but residents and faculty alike perceived that residents would benefit from more bread‐and‐butter cases. Hospitals considering the addition of a nonteaching service may want to incorporate a faculty development project focused on the triage process to ensure that these traditional medicine cases are assigned to resident services and to ensure that the great teaching case is not considered such because of complexity and acuity alone.

An increasingly robust body of literature has identified burnout as a substantial problem for physicians across specialties and practice settings.[1, 2, 3, 4] Burnout, a work‐related condition characterized by emotional exhaustion, depersonalization, and lack of a sense of personal accomplishment,[5] has been tied to negative consequences for patients, physicians, and the medical profession including medical errors,[6] poor physician health,[7, 8] and decreased professionalism.[9] Studies of burnout among general internists have pointed to time pressures, lack of work control, and difficult patient encounters as possible contributors.[10, 11]

Burnout has been demonstrated to affect a sizable proportion of hospitalists, with prevalence estimates from prior studies varying from 12.9% to 27.2%, although nearly all studies of US hospitalists have relied on single‐item instruments.[12, 13, 14, 15] Hospital‐based physicians have represented a rapidly expanding segment of the internist workforce for more than a decade,[14] but studies of the impact of inpatient vs outpatient practice location on burnout and career satisfaction are limited. A meta‐analysis of the impact of practice location on burnout relied almost exclusively on noncomparative studies from outside the United States.[15] A recent study of US physician burnout and satisfaction with work‐life balance showed that general internists expressed below average satisfaction with work‐life balance and had the second highest rate of burnout among 24 specialties.[4] However, this report did not differentiate between general internists working in inpatient vs outpatient settings.

We therefore examined burnout, satisfaction with work‐life balance, and other aspects of well‐being among internal medicine hospitalists relative to outpatient general internists, using a national sample developed in partnership with the American Medical Association.

METHODS

RESULTS

In the full survey across all specialties, 7288 physicians (26.7%) provided survey responses.[4] There were 448 outpatient internists and 130 internal medicine hospitalists who agreed to participate. Demographically, hospitalists were younger, worked longer hours, and were less likely to work in private practice than outpatient general internists (Table 1).

DISCUSSION

The primary result of this national study applying well‐validated metrics is that the overall rates of burnout among hospitalists and outpatient general internal medicine physicians were similar, as were rates of positive depression screening and QOL. Although these groups did not differ, the absolute rates of distress found in this study were high. Prior research has suggested that possible explanations for these high rates of distress include excessive workload, loss of work‐associated control and meaning, and difficulties with work‐home balance.[4] The present study, in the context of prior work showing that general internists have higher rates of burnout than almost any other specialty, suggests that the front‐line nature of the work of both hospitalists and outpatient general internists may exacerbate these previously cited factors. These results suggest that efforts to address physician well‐being are critically needed for both inpatient and outpatient physicians.

Despite the noted similarities, differences between hospitalists and outpatient general internists in certain aspects of well‐being merit further attention. For example, the lower rate of personal accomplishment among hospitalists relative to outpatient generalists is consistent with prior evidence.[15] The reasons for this difference are unknown, but the relative youth and inexperience of the hospitalists may be a factor. US hospitalists have been noted to feel like glorified residents in at least 1 report,[31] a factor that might also negatively impact personal accomplishment.

It is also worthwhile to place the burnout results for both groups in context with prior studies. Although we found high rates of burnout among outpatient physicians, our outpatient sample's mean MBI subset scores are not higher than previous samples of American[32] and Canadian[33] outpatient physicians, suggesting that this finding is neither new nor artifactual. Placing the hospitalist sample in perspective is more difficult, as very few studies have administered the MBI to US hospitalists, and those that have either administered 1 component only to an exclusive academic sample[34] or administered it to a small mixture of hospitalists and intensivists.[35] The prevalence of burnout we report for our hospitalist sample is higher than that reported by studies that utilized single‐item survey items¹²¹⁴; it is likely that the higher prevalence we report relates more to a more detailed assessment of the components of burnout than to a temporal trend, although this cannot be determined definitively from the data available.

The finding that 9.2% of hospitalists and 5.8% of outpatient general internists reported suicidal thoughts in the past 12 months is alarming, though consistent with prior data on US surgeons.[35] Although the higher rate of suicidal thoughts among hospitalists was not statistically significant, a better understanding of the factors associated with physician suicidality should be the focus of additional research.

Hospitalists were more likely than outpatient internists to report plans to leave their current practice in this study, although their plans after leaving differed. The fact that they were more likely to report plans to find a different role in medicine (rather than to leave medicine entirely or retire) is likely a function of age and career stage. The finding that hospitalists with an interest in changing jobs were more likely than their outpatient colleagues to consider administrative, leadership, education, and research roles may partially reflect the greater number of hospitalists at academic medical centers in this study, but suggests that hospitalists may indeed benefit from the availability of opportunities that have been touted as part of hospitalist diastole.[36]

Finally, rates of alcohol misuse and abuse found in this study were consistent with those reported in prior studies.[37, 38, 39] These rates support ongoing efforts to address alcohol‐related issues among physicians. In addition, the proportion of outpatient general internists and hospitalists reporting having a primary care provider was similar to that seen in prior research.[40] The fact that 1 in 3 physicians in this study did not have a primary care provider suggests there is great room for improvement in access to and prioritization of healthcare for physicians in general. However, it is noteworthy that hospitalists were less likely than outpatient general internists to have a primary care provider even after adjusting for their younger age as a group. The reasons behind this discrepancy are unclear but worthy of further investigation.

Several limitations of our study should be considered. The response rate for the entire study sample was 26.7%, which is similar to other US national physician surveys in this topic area.[41, 42, 43] Demographic comparisons with national data suggest the respondents were reasonably representative of physicians nationally,[4] and all analyses were adjusted for recognized demographic factors affecting our outcomes of interest. We found no statistically significant differences in demographics of early responders compared with late responders (a standard approach to evaluate for response bias),[14, 31] further supporting that responders were representative of US physicians. Despite this, response bias remains possible. For example, it is unclear if burned out physicians might be more likely to respond (eg, due to the personal relevance of the survey topic) or less likely to respond (eg, due to being too overwhelmed to open or complete the survey).

A related limitation is the relatively small number of hospitalists included in this sample, which limits the power of the study to detect differences between the study groups. The hospitalists in this study were also relatively experienced, with a median of 10 years in practice, although the overall demographics match closely to a recent national survey of hospitalists. Although age was considered in the analyses, this study may not fully characterize burnout patterns among very junior or very senior hospitalists. In addition, although analyses were adjusted for observed differences between the study groups for a number of covariates, there may be differences between the study groups in other, unmeasured factors that could act as confounders of the observed results. For example, the allocation of each individual's time to different activities (eg, clinical, research, education, administration), workplace flexibility and control, and meaning may all contribute to distress and well‐being, and could not be assessed in this study.

In conclusion, the degree of burnout, depression, and suicidal ideation in both hospitalists and outpatient general internists is similar and substantial. Urgent attention directed at better understanding the causes of distress and identifying solutions for all internists is needed.

An increasingly robust body of literature has identified burnout as a substantial problem for physicians across specialties and practice settings.[1, 2, 3, 4] Burnout, a work‐related condition characterized by emotional exhaustion, depersonalization, and lack of a sense of personal accomplishment,[5] has been tied to negative consequences for patients, physicians, and the medical profession including medical errors,[6] poor physician health,[7, 8] and decreased professionalism.[9] Studies of burnout among general internists have pointed to time pressures, lack of work control, and difficult patient encounters as possible contributors.[10, 11]

Burnout has been demonstrated to affect a sizable proportion of hospitalists, with prevalence estimates from prior studies varying from 12.9% to 27.2%, although nearly all studies of US hospitalists have relied on single‐item instruments.[12, 13, 14, 15] Hospital‐based physicians have represented a rapidly expanding segment of the internist workforce for more than a decade,[14] but studies of the impact of inpatient vs outpatient practice location on burnout and career satisfaction are limited. A meta‐analysis of the impact of practice location on burnout relied almost exclusively on noncomparative studies from outside the United States.[15] A recent study of US physician burnout and satisfaction with work‐life balance showed that general internists expressed below average satisfaction with work‐life balance and had the second highest rate of burnout among 24 specialties.[4] However, this report did not differentiate between general internists working in inpatient vs outpatient settings.

We therefore examined burnout, satisfaction with work‐life balance, and other aspects of well‐being among internal medicine hospitalists relative to outpatient general internists, using a national sample developed in partnership with the American Medical Association.

METHODS

RESULTS

In the full survey across all specialties, 7288 physicians (26.7%) provided survey responses.[4] There were 448 outpatient internists and 130 internal medicine hospitalists who agreed to participate. Demographically, hospitalists were younger, worked longer hours, and were less likely to work in private practice than outpatient general internists (Table 1).

DISCUSSION

The primary result of this national study applying well‐validated metrics is that the overall rates of burnout among hospitalists and outpatient general internal medicine physicians were similar, as were rates of positive depression screening and QOL. Although these groups did not differ, the absolute rates of distress found in this study were high. Prior research has suggested that possible explanations for these high rates of distress include excessive workload, loss of work‐associated control and meaning, and difficulties with work‐home balance.[4] The present study, in the context of prior work showing that general internists have higher rates of burnout than almost any other specialty, suggests that the front‐line nature of the work of both hospitalists and outpatient general internists may exacerbate these previously cited factors. These results suggest that efforts to address physician well‐being are critically needed for both inpatient and outpatient physicians.

Despite the noted similarities, differences between hospitalists and outpatient general internists in certain aspects of well‐being merit further attention. For example, the lower rate of personal accomplishment among hospitalists relative to outpatient generalists is consistent with prior evidence.[15] The reasons for this difference are unknown, but the relative youth and inexperience of the hospitalists may be a factor. US hospitalists have been noted to feel like glorified residents in at least 1 report,[31] a factor that might also negatively impact personal accomplishment.

It is also worthwhile to place the burnout results for both groups in context with prior studies. Although we found high rates of burnout among outpatient physicians, our outpatient sample's mean MBI subset scores are not higher than previous samples of American[32] and Canadian[33] outpatient physicians, suggesting that this finding is neither new nor artifactual. Placing the hospitalist sample in perspective is more difficult, as very few studies have administered the MBI to US hospitalists, and those that have either administered 1 component only to an exclusive academic sample[34] or administered it to a small mixture of hospitalists and intensivists.[35] The prevalence of burnout we report for our hospitalist sample is higher than that reported by studies that utilized single‐item survey items¹²¹⁴; it is likely that the higher prevalence we report relates more to a more detailed assessment of the components of burnout than to a temporal trend, although this cannot be determined definitively from the data available.

The finding that 9.2% of hospitalists and 5.8% of outpatient general internists reported suicidal thoughts in the past 12 months is alarming, though consistent with prior data on US surgeons.[35] Although the higher rate of suicidal thoughts among hospitalists was not statistically significant, a better understanding of the factors associated with physician suicidality should be the focus of additional research.

Hospitalists were more likely than outpatient internists to report plans to leave their current practice in this study, although their plans after leaving differed. The fact that they were more likely to report plans to find a different role in medicine (rather than to leave medicine entirely or retire) is likely a function of age and career stage. The finding that hospitalists with an interest in changing jobs were more likely than their outpatient colleagues to consider administrative, leadership, education, and research roles may partially reflect the greater number of hospitalists at academic medical centers in this study, but suggests that hospitalists may indeed benefit from the availability of opportunities that have been touted as part of hospitalist diastole.[36]

Finally, rates of alcohol misuse and abuse found in this study were consistent with those reported in prior studies.[37, 38, 39] These rates support ongoing efforts to address alcohol‐related issues among physicians. In addition, the proportion of outpatient general internists and hospitalists reporting having a primary care provider was similar to that seen in prior research.[40] The fact that 1 in 3 physicians in this study did not have a primary care provider suggests there is great room for improvement in access to and prioritization of healthcare for physicians in general. However, it is noteworthy that hospitalists were less likely than outpatient general internists to have a primary care provider even after adjusting for their younger age as a group. The reasons behind this discrepancy are unclear but worthy of further investigation.

Several limitations of our study should be considered. The response rate for the entire study sample was 26.7%, which is similar to other US national physician surveys in this topic area.[41, 42, 43] Demographic comparisons with national data suggest the respondents were reasonably representative of physicians nationally,[4] and all analyses were adjusted for recognized demographic factors affecting our outcomes of interest. We found no statistically significant differences in demographics of early responders compared with late responders (a standard approach to evaluate for response bias),[14, 31] further supporting that responders were representative of US physicians. Despite this, response bias remains possible. For example, it is unclear if burned out physicians might be more likely to respond (eg, due to the personal relevance of the survey topic) or less likely to respond (eg, due to being too overwhelmed to open or complete the survey).

A related limitation is the relatively small number of hospitalists included in this sample, which limits the power of the study to detect differences between the study groups. The hospitalists in this study were also relatively experienced, with a median of 10 years in practice, although the overall demographics match closely to a recent national survey of hospitalists. Although age was considered in the analyses, this study may not fully characterize burnout patterns among very junior or very senior hospitalists. In addition, although analyses were adjusted for observed differences between the study groups for a number of covariates, there may be differences between the study groups in other, unmeasured factors that could act as confounders of the observed results. For example, the allocation of each individual's time to different activities (eg, clinical, research, education, administration), workplace flexibility and control, and meaning may all contribute to distress and well‐being, and could not be assessed in this study.

In conclusion, the degree of burnout, depression, and suicidal ideation in both hospitalists and outpatient general internists is similar and substantial. Urgent attention directed at better understanding the causes of distress and identifying solutions for all internists is needed.

An increasingly robust body of literature has identified burnout as a substantial problem for physicians across specialties and practice settings.[1, 2, 3, 4] Burnout, a work‐related condition characterized by emotional exhaustion, depersonalization, and lack of a sense of personal accomplishment,[5] has been tied to negative consequences for patients, physicians, and the medical profession including medical errors,[6] poor physician health,[7, 8] and decreased professionalism.[9] Studies of burnout among general internists have pointed to time pressures, lack of work control, and difficult patient encounters as possible contributors.[10, 11]

Burnout has been demonstrated to affect a sizable proportion of hospitalists, with prevalence estimates from prior studies varying from 12.9% to 27.2%, although nearly all studies of US hospitalists have relied on single‐item instruments.[12, 13, 14, 15] Hospital‐based physicians have represented a rapidly expanding segment of the internist workforce for more than a decade,[14] but studies of the impact of inpatient vs outpatient practice location on burnout and career satisfaction are limited. A meta‐analysis of the impact of practice location on burnout relied almost exclusively on noncomparative studies from outside the United States.[15] A recent study of US physician burnout and satisfaction with work‐life balance showed that general internists expressed below average satisfaction with work‐life balance and had the second highest rate of burnout among 24 specialties.[4] However, this report did not differentiate between general internists working in inpatient vs outpatient settings.

We therefore examined burnout, satisfaction with work‐life balance, and other aspects of well‐being among internal medicine hospitalists relative to outpatient general internists, using a national sample developed in partnership with the American Medical Association.

METHODS

RESULTS

In the full survey across all specialties, 7288 physicians (26.7%) provided survey responses.[4] There were 448 outpatient internists and 130 internal medicine hospitalists who agreed to participate. Demographically, hospitalists were younger, worked longer hours, and were less likely to work in private practice than outpatient general internists (Table 1).

DISCUSSION

The primary result of this national study applying well‐validated metrics is that the overall rates of burnout among hospitalists and outpatient general internal medicine physicians were similar, as were rates of positive depression screening and QOL. Although these groups did not differ, the absolute rates of distress found in this study were high. Prior research has suggested that possible explanations for these high rates of distress include excessive workload, loss of work‐associated control and meaning, and difficulties with work‐home balance.[4] The present study, in the context of prior work showing that general internists have higher rates of burnout than almost any other specialty, suggests that the front‐line nature of the work of both hospitalists and outpatient general internists may exacerbate these previously cited factors. These results suggest that efforts to address physician well‐being are critically needed for both inpatient and outpatient physicians.

Despite the noted similarities, differences between hospitalists and outpatient general internists in certain aspects of well‐being merit further attention. For example, the lower rate of personal accomplishment among hospitalists relative to outpatient generalists is consistent with prior evidence.[15] The reasons for this difference are unknown, but the relative youth and inexperience of the hospitalists may be a factor. US hospitalists have been noted to feel like glorified residents in at least 1 report,[31] a factor that might also negatively impact personal accomplishment.

It is also worthwhile to place the burnout results for both groups in context with prior studies. Although we found high rates of burnout among outpatient physicians, our outpatient sample's mean MBI subset scores are not higher than previous samples of American[32] and Canadian[33] outpatient physicians, suggesting that this finding is neither new nor artifactual. Placing the hospitalist sample in perspective is more difficult, as very few studies have administered the MBI to US hospitalists, and those that have either administered 1 component only to an exclusive academic sample[34] or administered it to a small mixture of hospitalists and intensivists.[35] The prevalence of burnout we report for our hospitalist sample is higher than that reported by studies that utilized single‐item survey items¹²¹⁴; it is likely that the higher prevalence we report relates more to a more detailed assessment of the components of burnout than to a temporal trend, although this cannot be determined definitively from the data available.

The finding that 9.2% of hospitalists and 5.8% of outpatient general internists reported suicidal thoughts in the past 12 months is alarming, though consistent with prior data on US surgeons.[35] Although the higher rate of suicidal thoughts among hospitalists was not statistically significant, a better understanding of the factors associated with physician suicidality should be the focus of additional research.

Hospitalists were more likely than outpatient internists to report plans to leave their current practice in this study, although their plans after leaving differed. The fact that they were more likely to report plans to find a different role in medicine (rather than to leave medicine entirely or retire) is likely a function of age and career stage. The finding that hospitalists with an interest in changing jobs were more likely than their outpatient colleagues to consider administrative, leadership, education, and research roles may partially reflect the greater number of hospitalists at academic medical centers in this study, but suggests that hospitalists may indeed benefit from the availability of opportunities that have been touted as part of hospitalist diastole.[36]

Finally, rates of alcohol misuse and abuse found in this study were consistent with those reported in prior studies.[37, 38, 39] These rates support ongoing efforts to address alcohol‐related issues among physicians. In addition, the proportion of outpatient general internists and hospitalists reporting having a primary care provider was similar to that seen in prior research.[40] The fact that 1 in 3 physicians in this study did not have a primary care provider suggests there is great room for improvement in access to and prioritization of healthcare for physicians in general. However, it is noteworthy that hospitalists were less likely than outpatient general internists to have a primary care provider even after adjusting for their younger age as a group. The reasons behind this discrepancy are unclear but worthy of further investigation.

Several limitations of our study should be considered. The response rate for the entire study sample was 26.7%, which is similar to other US national physician surveys in this topic area.[41, 42, 43] Demographic comparisons with national data suggest the respondents were reasonably representative of physicians nationally,[4] and all analyses were adjusted for recognized demographic factors affecting our outcomes of interest. We found no statistically significant differences in demographics of early responders compared with late responders (a standard approach to evaluate for response bias),[14, 31] further supporting that responders were representative of US physicians. Despite this, response bias remains possible. For example, it is unclear if burned out physicians might be more likely to respond (eg, due to the personal relevance of the survey topic) or less likely to respond (eg, due to being too overwhelmed to open or complete the survey).

A related limitation is the relatively small number of hospitalists included in this sample, which limits the power of the study to detect differences between the study groups. The hospitalists in this study were also relatively experienced, with a median of 10 years in practice, although the overall demographics match closely to a recent national survey of hospitalists. Although age was considered in the analyses, this study may not fully characterize burnout patterns among very junior or very senior hospitalists. In addition, although analyses were adjusted for observed differences between the study groups for a number of covariates, there may be differences between the study groups in other, unmeasured factors that could act as confounders of the observed results. For example, the allocation of each individual's time to different activities (eg, clinical, research, education, administration), workplace flexibility and control, and meaning may all contribute to distress and well‐being, and could not be assessed in this study.

In conclusion, the degree of burnout, depression, and suicidal ideation in both hospitalists and outpatient general internists is similar and substantial. Urgent attention directed at better understanding the causes of distress and identifying solutions for all internists is needed.

Hospital medicine is a rapidly growing field of US clinical practice.[1] Almost since its advent, concerns have been expressed about the potential for hospitalists to burn out.[2] Hospitalists are not unique in this; similar concerns heralded the arrival of other location‐defined specialties, including emergency medicine[3] and the full‐time intensivist model,[4] a fact that has not gone unnoted in the literature about hospitalists.[5]

The existing international literature on physician burnout provides good reason for this concern. Inpatient‐based physicians tend to work unpredictable schedules, with substantial impact on home life.[6] They tend to be young, and much burnout literature suggests a higher risk among younger, less‐experienced physicians.[7] When surveyed, hospitalists have expressed more concerns about their potential for burnout than their outpatient‐based colleagues.[8]

In fact, data suggesting a correlation between inpatient practice and burnout predate the advent of the US hospitalist movement. Increased hospital time was reported to correlate with higher rates of burnout in internists,[9] family practitioners,[10] palliative physicians,[11] junior doctors,[12] radiologists,[13] and cystic fibrosis caregivers.[14] In 1987, Keinan and Melamed[15] noted, Hospital work by its very nature, as compared to the work of a general practitioner, deals with the more severe and complicated illnesses, coupled with continuous daily contacts with patients and their anxious families. In addition, these physicians may find themselves embroiled in the power struggles and competition so common in their work environment.

There are other features, however, that may protect inpatient physicians from burnout. Hospital practice can facilitate favorable social relations involving colleagues, co‐workers, and patients,[16] a factor that may be protective.[17] A hospitalist schedule also can allow more focused time for continuing medical education, research, and teaching,[18] which have all been associated with reduced risk of burnout in some studies.[17] Studies of psychiatrists[19] and pediatricians[20] have shown a lower rate of burnout among physicians with more inpatient duties. Finally, a practice model involving a seemingly stable cadre of inpatient physicians has existed in Europe for decades,[2] indicating at least a degree of sustainability.

Information suggesting a higher rate of burnout among inpatient physicians could be used to target therapeutic interventions and to adjust schedules, whereas the opposite outcome could refute a pervasive myth. We therefore endeavored to summarize the literature on burnout among inpatient versus outpatient physicians in a systematic fashion, and to include data not only from the US hospitalist experience but also from other countries that have used a similar model for decades. Our primary hypothesis was that inpatient physicians experience more burnout than outpatient physicians.

It is important to distinguish burnout from depression, job dissatisfaction, and occupational stress, all of which have been studied extensively in physicians. Burnout, as introduced by Freudenberger[21] and further characterized by Maslach,[22] is a condition in which emotional exhaustion, depersonalization, and a low sense of personal accomplishment combine to negatively affect work life (as opposed to clinical depression, which affects all aspects of life). Job satisfaction can correlate inversely with burnout, but it is a separate process[23] and the subject of a recent systematic review.[24] The importance of distinguishing burnout from job dissatisfaction is illustrated by a survey of head and neck surgeons, in which 97% of those surveyed indicated satisfaction with their jobs and 34% of the same group answered in the affirmative when asked if they felt burned out.[25]

One obstacle to the meaningful comparison of burnout prevalence across time, geography, and specialty is the myriad ways in which burnout is measured and reported. The oldest and most commonly used instrument to measure burnout is the Maslach Burnout Inventory (MBI), which contains 22 items assessing 3 components of burnout (emotional exhaustion, depersonalization, and low personal accomplishment).[26] Other measures include the Copenhagen Burnout Inventory[27] (19 items with the components personal burnout, work‐related burnout, and client‐related burnout), Utrecht Burnout Inventory[28] (20‐item modification of the MBI), Boudreau Burnout Questionnaire[29] (30 items), Arbeitsbezogenes Verhaltens und Erlebensmuster[30] (66‐item questionnaire assessing professional commitment, resistance to stress, and emotional well‐being), Shirom‐Melamed Burnout Measure[31] (22 items with subscales for physical fatigue, cognitive weariness, tension, and listlessness), and a validated single‐item questionnaire.[32]

METHODS

Electronic searches of MEDLINE, EMBASE, PsycINFO, SCOPUS, and PubMed were undertaken for articles published from January 1, 1974 (the year in which burnout was first described by Freudenberger[21]) to 2012 (last accessed, September 12, 2012) using the Medical Subject Headings (MeSH) terms stress, psychological; burnout, professional; adaptation, psychological; and the keyword burnout. The same sources were searched to create another set for the MeSH terms hospitalists, physician's practice patterns, physicians/px, professional practice location, and the keyword hospitalist#. Where exact subject headings did not exist in databases, comparable subject headings or keywords were used. The 2 sets were then combined using the operator and. Abstracts from the Society of Hospital Medicine annual conferences were hand‐searched, as were reference lists from identified articles. To ensure that pertinent international literature was captured, there was no language restriction in the search.

A 2‐stage screening process was used. The titles and abstracts of all articles identified in the search were independently reviewed by 2 investigators (D.L.R. and K.J.C.) who had no knowledge of each other's results. An article was obtained when either reviewer deemed it worthy of full‐text review.

All full‐text articles were independently reviewed by the same 2 investigators. The inclusion criterion was the measurement of burnout in physicians who are stated to or can be reasonably assumed to spend the substantial majority of their clinical practice exclusively in either the inpatient or the outpatient setting. Studies of emergency department physicians or specialists who invariably spend substantial amounts of time in both settings (eg, surgeons, anesthesiologists) were excluded. Studies limited to trainees or nonphysicians were also excluded. For both stages of review, agreement between the 2 investigators was assessed by calculating the statistic. Disagreements about inclusion were adjudicated by a third investigator (A.I.B.).

Because our goal was to establish and compare the rate of burnout among US hospitalists and other inpatient physicians around the world, we included studies of hospitalists according to the definition in use at the time of the individual study, noting that the formal definition of a hospitalist has changed over the years.[33] Because practice patterns for physicians described as primary care physicians, family doctors, hospital doctors, and others differ substantially from country to country, we otherwise included only the studies where the practice location was stated explicitly or where the authors confirmed that their study participants either are known or can be reasonably assumed to spend more than 75% of their time caring for hospital inpatients, or are known or can be reasonably assumed to spend the vast majority of their time caring for outpatients.

Data were abstracted using a standardized form and included the measure of burnout used in the study, results, practice location of study subjects, and total number of study subjects. When data were not clear (eg, burnout measured but not reported by the authors, practice location of study subjects not clear), authors were contacted by email, or when no current email address could be located or no response was received, by telephone or letter. In instances where burnout was measured repeatedly over time or before and after a specific intervention, only the baseline measurement was recorded. Because all studies were expected to be nonrandomized, methodological quality was assessed using a version of the tool of Downs and Black,[34] adapted where necessary by omitting questions not applicable to the specific study type (eg randomization for survey studies)[35] and giving a maximum of 1 point for the inclusion of a power calculation.

Two a priori analyses were planned: (1) a statistical comparison of articles directly comparing burnout among inpatient and outpatient physicians, and (2) a statistical comparison of articles measuring burnout among inpatient physicians with articles measuring burnout among outpatient physicians by the most frequently reported measuremean subset scores for emotional exhaustion, depersonalization, and personal accomplishment on the MBI.

The primary outcome measures were the differences between mean subset scores for emotional exhaustion, depersonalization, and personal accomplishment on the MBI. All differences are expressed as (outpatient meaninpatient mean). The variance of each outcome was calculated with standard formulas.[36] To calculate the overall estimate, each study was weighted by the reciprocal of its variance. Studies with fewer than 10 subjects were excluded from statistical analysis but retained in the systematic review.

For studies that reported data for both inpatient and outpatient physicians (double‐armed studies), Cochran Q test and the I² value were used to assess heterogeneity.[37, 38] Substantial heterogeneity was expected because these individual studies were conducted for different populations in different settings with different study designs, and this expectation was confirmed statistically. Therefore, we used a random effects model to estimate the overall effect, providing a conservative approach that accounted for heterogeneity among studies.[39]

To assess the durability of our findings, we performed separate multivariate meta‐regression analyses by including single‐armed studies only and including both single‐armed and double‐armed studies. For these meta‐regressions, means were again weighted by the reciprocal of their variances, and the arms of 2‐armed studies were considered separately. This approach allowed us to generate an estimate of the differences between MBI subset scores from studies that did not include such an estimate when analyzed separately.[40]

We examined the potential for publication bias in double‐armed studies by constructing a funnel plot, in which mean scores were plotted against their standard errors.[41] The trim‐and‐fill method was used to determine whether adjustment for publication bias was necessary. In addition, Begg's rank correlation test[42] was completed to test for statistically significant publication bias.

Stata 10.0 statistical software (StataCorp, College Station, TX) was used for data analyses. A P value of 0.05 or less was deemed statistically significant. The Preferred Reporting Items for Systematic Reviews and Meta‐analysis checklist was used for the design and execution of the systematic review and meta‐analysis.[43]

Subgroup analyses based on location were undertaken a posteriori. All data (double‐armed meta‐analysis, meta‐regression of single‐armed studies, and meta‐regression of single‐ and double‐armed studies) were analyzed by location (United States vs other; United States vs Europe vs other).

RESULTS

The search results are outlined in Figure 1. In total, 1704 articles met the criteria for full‐text review. A review of pertinent reference lists and author contacts led to the addition of 149 articles. Twenty‐nine references could not be located by any means, despite repeated attempts. Therefore, 1824 articles were subjected to full‐text review by the 2 investigators.

Figure 1

Initially, 57 articles were found that met criteria for inclusion. Of these, 2 articles reported data in formats that could not be interpreted.[44, 45] When efforts to clarify the data with the authors were unsuccessful, these studies were excluded. A study specifically designed to assess the response of physicians to a recent series of terrorist attacks[46] was excluded a posteriori because of lack of generalizability. Of the other 54 studies, 15 reported burnout data on both outpatient physicians and inpatient physicians, 22 reported data on outpatient physicians only, and 17 reported data on inpatient physicians only. Table 1 summarizes the results of the 37 studies involving outpatient physicians; Table 2 summarizes the 32 studies involving inpatient physicians.