Introduction

Burnout is a work-related syndrome characterized by emotional exhaustion, depersonalization, and a sense of low personal accomplishment.¹ Physician burnout is associated with increased medical errors,^2,3 job attrition,^4,5 depression,⁶ and suicide,⁷ as well as reduced quality outcomes,^2,3 patient satisfaction,^2,8 and productivity.^4,9 Family physicians have higher rates of burnout than other specialties,¹⁰ and studies show burnout can begin in training^11,12 and may persist into practice.^13,14

To address trainee and physician burnout, a substantial investment in residency wellness initiatives for burnout mitigation have been implemented,^4,15,16 and, in addition, the Accreditation Council for Graduate Medical Education (ACGME) has mandated that residency programs address physician well-being and burnout.¹⁷ Various interventions have been introduced, such as individual-level interventions addressing mindfulness, resilience, and meaning in work, along with organizational interventions targeting duty hours, workload, efficiency, and social supports.^15,16 Yet, there is limited data about the long-term effects of these wellness interventions, especially after graduation.¹⁸

Our objective was to examine whether specific residency workload or wellness curricula were associated with self-reported burnout measures among US family medicine residency graduates 3 years into practice.

KEY POINTS

Methods

Eligible participants included all 2018 US family medicine residency graduates certified by the American Board of Family Medicine (ABFM) who were 3 years in practice after graduation. We used data from 2 national surveys. First, residency-level data were collected from a 2018 Council of Academic Family Medicine Education Research Alliance (CERA) Program Director (PD) survey, the methodology of which can be found elsewhere.¹⁹ The 2018 CERA PD survey included a module assessing workload and wellness curricula during residency (online supplementary data).

These residency data were combined with graduate data from the 2021 National Graduate Survey (NGS).^20,21 The NGS is sent to ABFM diplomates 3 years after residency graduation and completed voluntarily. Since the NGS distribution coincided with the COVID-19 pandemic in 2021, the survey remained open longer than previous years, 18 months instead of 12. The NGS assessed burnout through single-item proxy measures to evaluate respondents’ self-assessment of emotional exhaustion (“feeling burned out”) and depersonalization (“callousness”). These single-item measures have been predictive of full Maslach Burnout Inventory scores.²²

The overall response rate for the 2018 CERA PD survey was 43.3% (254 of 587). The overall response rate to the 2021 NGS was 45.1% (1623 of 3596). The NGS and CERA data were then merged, linked by residency program, and the final analytic sample included only graduates with NGS data who also had residency data from the CERA survey. Of the 254 CERA responses, 211 were successfully matched to graduates on NGS, though 9 of these surveys were missing all program-related variables. After excluding these 9 programs and the 41 NGS graduate responses linked to them, the final CERA ABFM National Family Medicine Residency Outcomes Project (FM-ROP) analytic sample included 738 graduates from 202 (of 587) residency programs.

We characterized the sample with descriptive statistics for residency programs and graduates. Residency program characteristics included program type, number of total residents, proportion of non-US medical school graduates, whether the program was unopposed, the geographic region of the program, and the type of residency clinic. For residency workload characteristics, we collected the number of inpatient months throughout residency, the average number of hours worked per week during each residency year, the average number of half-day continuity clinics per week during postgraduate year (PGY) 3, and the average number of patients scheduled for continuity clinics. For individual-level wellness curricula, we collected the number of hours of formal didactic/instructional time about wellness and burnout, PD assumptions about the prioritization of wellness in their residency program compared to other programs, presence of specific programming (facilitated small groups; non-facilitated small groups; structured wellness curricula, such as Rachel Remen’s Meaning in Medicine; stress management curricula; mindfulness approaches, such as meditation or yoga; communication skills trainings; presence of protected work hour time for nonclinical work, such as advocacy or research; presence of protected time for personal tasks) and frequency of team-building activities with individuals outside of the immediate residency program. Graduate demographics included age, gender, being underrepresented in medicine, type of medical degree (MD/DO), international medical graduate status, and current primary practice site.

After exploring the spread of these characteristics, we collapsed categories of some independent categorical variables. For example, months of inpatient across PGY-1 to 3 were collapsed into tertiles, average hours per week residents work were collapsed into categories of 60 hours or less or 61 hours or more, and an average of more than 7 half-days of continuity clinic per week in the final year of residency were grouped. We also collapsed the dependent variables of graduate-reported “feeling burned out” and “callousness” into a bivariate measure, consistent with previous work.²³ Graduates were defined as having burnout if they reported feeling burned out once a week or more, callous once a week or more, or both. Those who reported these feelings less frequently were grouped separately. This self-reported burnout is referred to as burnout hereafter.

Since many programs had multiple elements of wellness curricula, we created a summative score of wellness curricular offerings, as well as separate summative scores for the number of individual- and organizational-level curricula. We created a final scaled variable to combine the reported program amount of time allotted for wellness curricula and the number of types of curricula.

We conducted bivariate analyses addressing workload and wellness curricular offerings against our main dependent variable, burnout. We used Kendall-tau tests to compare ordinal variables (to account for smaller datasets and the presence of ties among the data)²⁴ and the chi-square test for other nominal variables. We then conducted logistic regression on all variables that were significant in the bivariate models, while controlling for current graduate practice factors and demographics. The data were analyzed using IBM SPSS Statistics for Windows, Version 28 (IBM Corp).

This study was approved by the American Academy of Family Physicians Institutional Review Board.

Results

In our national Family Medicine Residency Outcomes Project analytic sample of 738 graduates, 57.3% (n=423) of respondents were female, the average age was 35 years (range 29-61), and 31.7% (n=234) were currently practicing in a hospital or health system–owned practice. Forty percent of graduates (n=290) met our definition of burnout. Most residency programs were community-based, university-affiliated programs (133 of 202, 65.8%) and included a wide range of wellness curricula options and workload hours. See additional characteristics in Table 1 and the online supplementary data.

Table 1 Characteristics of Residency Programs From Special CERA Program Director Survey of 2018 With at Least One Graduate Respondent on the 2021 National Graduate Survey (N=202)

View Fullscreen

In exploring the association of program workload on graduates’ burnout, only having less than 60 hours of work per week during PGY-1 was associated with decreased burnout in practice (65.7% (115 of 175) not burned out vs 57.4% (298 of 519) not burned out if working 61 or more hours per week, P=.05; Table 2). There was no association seen for additional workload variables, including hours worked per week during PGY-2 or PGY-3. Most individual-level residency wellness offerings were also not associated with graduate burnout, with 2 exceptions: stress management curricula was associated with fewer graduates reporting burnout (63.8% [203 of 318] not burned out vs 55.9% [218 of 390] of those without this curricula not burned out, P=.03), while having non-facilitated small groups (ie, resident-only groups for sharing experiences) was associated with more graduates reporting burnout (43.7% [185 of 423] of those with these groups report burnout vs 35.8% [102 of 285] of those without, P=.04; Table 2). The remaining wellness curricula were not associated with graduates’ reports of burnout.

Table 2 Residency Workload and Wellness Curricula Correlated With Burnout 3 Years After Residency Graduation, N=738

View Fullscreen

In multivariate analyses, wellness curricula that showed significance in bivariate analyses (non-facilitated small groups and stress management curricula) were no longer associated with burnout (Table 3). Working more than 60 hours per week during PGY-1 remained associated with more burnout (OR=1.5; 95% CI, 1.04-2.31; P=.03). Female residents also had higher rates of burnout (OR=1.48; 95% CI, 1.04-2.10; P=.03), and a primary practice site at the time of response showed those at community health centers and hospitals/HMO sites each reported higher burnout (OR=2.20; 95% CI, 1.13-4.30; P=.021 and OR=2.24; 95% CI, 1.24-4.00; P=.007), respectively.

Table 3 Adjusted Associations Between Personal and Practice Characteristics With Burnout Among 2018 Family Medicine Residency Graduates Who Practice Continuity of Care, N=738

View Fullscreen

Discussion

In this large, national, prospective study of family medicine graduates, we found no associations between specific wellness curricula or total time of wellness programming and reduced burnout in family medicine physicians 3 years into practice, after controlling for practice and demographic characteristics. Of residency workload factors, only working an average of less than 60 hours of work per week during PGY-1 was associated with decreased burnout in graduates. No residency program characteristics were associated with burnout in graduates.

In bivariate analysis, the presence of stress management curricula was associated with lower burnout among graduates, but was not associated in multivariate analysis. In our survey stress management curricula were not defined and may have been interpreted differentially by PDs. A previous CERA study surveyed PDs on stress management programs and reported that these sessions included self-care, residency support groups, Balint groups, and residency retreats.²⁵ Many of these elements were called out separately in our survey options and none were associated with burnout among graduates. Related studies relevant to stress management have supported the role of personal behaviors limiting burnout, such as restful sleep, regular exercise, and self-care.¹⁵ Others have focused on increasing physician resilience through the promotion of psychological flexibility, self-reflection, and gratitude.²⁶ Given that it is unknown how PDs classified the components of stress management curricula, it is unclear whether, and which, combinations of initiatives may have been helpful.

Graduates who experienced non-facilitated resident forums (eg, resident-only groups for sharing experiences) were more likely to report being burned out in bivariate comparison, but this was no longer associated in multivariate analysis. From the survey data, it was unclear how often these non-facilitated resident forums occurred or whether they included all available residents or were split out by individual PGY class. Additionally, the structure, whether there was an open forum or structured components, was not specified. Team-building activities, which are likely to happen less frequently and with a larger mix of residents, faculty, and staff, were not associated with burnout in practice.

Since our 2018 PD survey there has been an expansion of wellness interventions, such as group coaching or interventions focused on increasing resilience, which were not included in this study. We also did not measure the hours spent on each type of intervention, how interventions were distributed throughout residency, or the resident uptake of the offered interventions, which all may affect how skills are learned, incorporated, and used by residents. For example, a higher number of interventions during PGY-3, when residents are anticipating practice, may have higher retention given approaching applicability. However, simple quantity of wellness curricula seems unlikely to reduce later burnout: none of our 4 summative variables looking at quantity of wellness curricula—total offerings overall, offerings targeting individual factors, offerings targeting organizational factors, or any of these related to amount of time spent—were associated with lower burnout after graduation.

Only work hours during PGY-1 was associated with burnout, in both bivariate and multivariate analyses. Historic decreases in work hours are partly aimed at stress management: increased time for self-care activities, such as sleep and exercise. The significance of work hours during PGY-1, compared to PGY-2 and PGY-3, is important to assess. It may be that the stark transition from being a medical student to a resident requires adjustment, and thus, subsequent longer hours are less influential. Alternatively, those with longer working hours during PGY-1 may have become burned out, which continued longitudinally. Outside of the United States, where work hours are even more restricted, resident fatigue and work-life balance also remains an issue.²⁷ Work hours, however, must be balanced with the need for adequate clinical exposure and training. Importantly, in contrast to residency, clinical practice settings have no clear work hour restrictions.

Although we assessed wellness interventions and workload, we were unable to assess the resident perspective. A recent study in family medicine residency programs showed that resident satisfaction with wellness programming was significantly lower than PD satisfaction,²⁸ which is consistent with similar studies in other specialties.²⁹ Thus, the lack of association between wellness curricula employed in 2018 and later burnout may be a true finding. It also may be that interventions introduced during residency may be effective within that context, but have little effect after graduation. Our study found increased burnout was highly associated with practice site; those at community health centers and hospital or HMO practices reported more than twice as much burnout. Thus, current practice characteristics may dominate other factors.

This study is limited by response bias, as the responses do not include most family medicine residency programs. The use of 2 single-item measures for burnout may over-simplify classification of physicians compared to multi-item surveys. Additionally, the study coincided with the COVID-19 pandemic, which may have affected graduate reactions to wellness curricula in unknown ways. As noted, CERA survey items were limited to the perceptions of PDs who may have interpreted our curricular descriptive terms differently, and the survey did not measure how engaged residents were in the curricular offerings.

This study demonstrates the critical need for longitudinal studies when curricula are introduced with a goal of affecting future physicians’ wellness, resilience, or work-life balance. Next steps could include examining whether targeted teaching of the application of wellness strategies after graduation is related to physician burnout in practice or whether any of the factors examined in this study are associated with resident burnout while in residency. Alternatively, a focus on altering structural aspects of clinical practice, particularly for certain practice sites, may prove more fruitful.

Conclusions

This national, longitudinal study of associations between wellness curricula during family medicine residency and burnout in practicing physicians 3 years after graduation showed no association with wellness curricula, whether specific types or total quantity, and later burnout, after controlling for practice and demographic factors. Only the program factor of less than 60 hours per week work in PGY-1 was associated with fewer graduates reporting burnout, in multivariable analysis.