Burnout Among Mid-Career Academic Medical Faculty
The authors surveyed 841 mid-career faculty who remained in academia after receiving NIH K08 and K23 grants between 2006 and 2009. Work-related burnout was 40.8% for women and 23.0% for men, a gender difference that persisted after adjusting for covariates. A favorable workplace climate and longer sleep duration were associated with lower odds of burnout, whereas clinical hours, experiences of sexual harassment, and increased work hours since the pandemic were associated with work-related burnout. The authors proposed evidence-based interventions to support the careers of female faculty and improve workplace climate.
Introduction
Physician burnout, career dissatisfaction, and attrition have been studied for the past two decades. Previous studies reported that over half of physicians experienced burnout symptoms. The National Academies of Sciences, Engineering, and Medicine defined burnout as a workplace syndrome characterized by emotional exhaustion, depersonalization such as cynicism, and low personal accomplishment. The National Plan for Health Workforce Well-Being, published in 2022, prioritized measuring burnout with validated tools and understanding the effects of discrimination, sexism, and the pandemic.
Mid-career clinical researchers are more than 10 years beyond the completion of their training. Caregiving responsibilities, reduced mentoring support, and increased expectations for institutional service and research productivity can place burdens on these researchers. Women may experience greater caregiving burdens, marginalization, and reduced recognition. Promotion, academic reputation, and expanded mentoring networks provide stability.
The authors examined whether burnout among mid-career clinical researchers differed by gender and whether these differences persisted after accounting for work hours, domestic labor, and workplace climate. The study population consisted of recipients of the National Institutes of Health (NIH) K08 and K23 career development grants.
Methods
The authors sent surveys by mail or email between August 2021 and August 2022 to 1,430 individuals who received new K08 and K23 grants from 2006 to 2009 and whose contact information had been verified. A $50 incentive was provided regardless of participation. The analysis was limited to respondents who remained in academic medicine. The University of Michigan Institutional Review Board approved the study and waived the requirement for written informed consent. Data were collected and managed using REDCap.
The survey covered careers, time allocation, mentoring, family, workplace environment, and well-being. The study outcomes were personal burnout and work-related burnout, measured using the Copenhagen Burnout Inventory (CBI). Personal burnout measures fatigue and exhaustion regardless of cause, including work-related fatigue.
The CBI assesses personal burnout with 6 items and work-related burnout with 7 items on a 5-point scale; respondents were classified as having burnout if their mean score on a scale from 0 to 100 was 50 or higher. Responses were converted to scores at 25-point intervals, and one reverse-worded item was reverse-scored.
Covariates included gender, race and ethnicity, sexual orientation and gender identity, spouse’s employment status, caregiving responsibilities, and academic rank. They also included time spent on work, clinical practice, and domestic labor, sleep duration, satisfaction with mentoring, and experiences of workplace sexual harassment in the past 2 years. The survey also asked whether weekly time spent on work or domestic labor had increased by 8 hours or more compared with before the pandemic. The general workplace climate scale assessed kindness, respect, and collegiality, while the DEI scale evaluated homogeneity, sexism, and homophobia. Higher scores on both scales indicated more favorable perceptions.
The authors generated descriptive statistics by gender and conducted bivariate analyses before constructing multivariable models for the two burnout outcomes. Holm corrections were applied for multiple testing, and the significance level was set at P<.05. The authors supplemented the complete-case analysis with multiple imputation for missing covariates and analyses of continuous burnout scores. Analyses were conducted from June to October 2023 using Stata 16.1 and SAS 9.4.
Results
Of the 1,430 individuals, 926 responded, yielding a response rate of 65%. The analysis included 841 respondents (90.8%) who remained in academia. These included 426 men (50.7%), 392 women (46.6%), 2 nonbinary individuals (0.2%), and 21 individuals who did not report their gender (2.5%). There were 172 Asian individuals (20.5%), 66 individuals from racial and ethnic groups underrepresented in medicine (URIM) (7.8%), 580 White individuals (69.0%), and 23 individuals who did not report their race and ethnicity (2.7%). There were 783 cisgender and heterosexual respondents (93.1%), 34 LGBTQ+ respondents (4.0%), and 24 individuals who did not report their sexual orientation (2.9%).
The table presents respondent characteristics by gender. Values are presented as number (%) unless otherwise indicated.
| Characteristic | Women (n=392) | Men (n=426) | Nonbinary or Unreported Gender (n=23) | P Value, Women vs Men | Adjusted P Value, Women vs Men |
|---|---|---|---|---|---|
| Gender | |||||
| Men | 0 | 426 (100) | 0 | NA | NA |
| Women | 392 (100) | 0 | 0 | NA | NA |
| Nonbinary | 0 | 0 | 2 (8.70) | NA | NA |
| Missing or unreported | 0 | 0 | 21 (91.30) | NA | NA |
| Race and ethnicity | |||||
| Asian | 83 (21.17) | 88 (20.66) | 1 (4.35) | .77 | .77 |
| Underrepresented in medicine group | 29 (7.39) | 37 (8.69) | 0 | ||
| White | 280 (71.43) | 296 (69.48) | 4 (17.39) | ||
| Missing or unreported | 0 | 5 (1.17) | 18 (78.26) | ||
| Sexual orientation | |||||
| Heterosexual | 374 (95.41) | 409 (96.01) | 2 (8.70) | .28 | >.99 |
| Gay or lesbian | 8 (2.04) | 13 (3.05) | 0 | ||
| Bisexual | 6 (1.53) | 1 (0.23) | 0 | ||
| Not applicable | 1 (0.26) | 1 (0.23) | 2 (8.70) | ||
| Missing or unreported | 3 (0.77) | 2 (0.47) | 19 (82.61) | ||
| Marital status and spouse employment | |||||
| Single, divorced, or widowed | 48 (12.24) | 22 (5.16) | 4 (17.39) | <.001 | <.001 |
| Married with unemployed or part-time employed spouse/partner | 74 (18.88) | 186 (43.66) | 7 (30.43) | ||
| Married with full-time employed spouse/partner | 266 (67.86) | 214 (50.23) | 10 (43.48) | ||
| Missing or unreported | 4 (1.02) | 4 (0.94) | 2 (8.70) | ||
| Children requiring adult supervision | |||||
| None | 175 (44.64) | 225 (52.82) | 15 (65.22) | .02 | .14 |
| Yes | 213 (54.34) | 196 (46.01) | 6 (26.09) | ||
| Missing or unreported | 4 (1.02) | 5 (1.17) | 2 (8.70) | ||
| Other dependents | |||||
| None | 299 (76.28) | 352 (82.63) | 18 (78.26) | .006 | .06 |
| Yes | 84 (21.43) | 59 (13.85) | 2 (8.70) | ||
| Missing or unreported | 9 (2.30) | 15 (3.52) | 3 (13.04) | ||
| Academic rank | |||||
| Professor | 202 (51.53) | 254 (59.62) | 10 (43.48) | .04 | .30 |
| Associate professor | 175 (44.64) | 153 (35.92) | 0 | ||
| Assistant professor or other | 15 (3.83) | 19 (4.46) | 1 (4.35) | ||
| Specialty | |||||
| Basic science or nonclinical | 114 (29.08) | 55 (12.91) | 5 (21.74) | <.001 | <.001 |
| Women’s, children’s, and family-related clinical specialties | 101 (25.77) | 88 (20.66) | 4 (17.39) | ||
| Hospital-based specialties | 34 (8.67) | 60 (14.08) | 5 (21.74) | ||
| Medical specialties | 137 (34.95) | 184 (43.19) | 9 (39.13) | ||
| Surgical specialties | 6 (1.53) | 39 (9.15) | 0 | ||
| Region | |||||
| West | 102 (26.02) | 102 (23.94) | 9 (39.13) | .58 | >.99 |
| Midwest | 77 (19.64) | 99 (23.24) | 3 (13.04) | ||
| Northeast | 128 (32.65) | 124 (29.11) | 7 (30.43) | ||
| South | 83 (21.17) | 98 (23.00) | 4 (17.39) | ||
| International | 2 (0.51) | 3 (0.70) | 0 | ||
| Substantial change in work hours since the pandemic (>8 h) | |||||
| None | 315 (80.36) | 356 (83.57) | 18 (78.26) | .30 | .89 |
| Yes | 74 (18.88) | 69 (16.20) | 4 (17.39) | ||
| Missing or unreported | 3 (0.77) | 1 (0.23) | 1 (4.35) | ||
| Substantial change in caregiving/domestic labor hours since the pandemic (>8 h) | |||||
| None | 303 (77.30) | 378 (88.73) | 16 (69.57) | <.001 | <.001 |
| Yes | 83 (21.17) | 40 (9.39) | 5 (21.74) | ||
| Missing or unreported | 6 (1.53) | 8 (1.88) | 2 (8.7) | ||
| Weekly work hours | |||||
| Number | 389 | 425 | 22 | .006 | .06 |
| Mean (SD) [95% CI] | 55.52 (13.36) [54.19-56.85] | 58.04 (12.68) [56.84-59.25] | 62.27 (15.10) [55.96-68.58] | ||
| Weekly caregiving/domestic labor hours | |||||
| Number | 387 | 418 | 21 | <.001 | <.001 |
| Mean (SD) [95% CI] | 30.31 (21.05) [28.21-32.41] | 22.99 (16.88) [21.37-24.61] | 27.81 (18.83) [19.76-35.86] | ||
| Weekly clinical hours | |||||
| Number | 382 | 421 | 22 | <.001 | <.001 |
| Mean (SD) [95% CI] | 13.1 (12.96) [11.80-14.40] | 17.67 (14.51) [16.28-19.05] | 16.6 (14.37) [10.59-22.60] | ||
| Daily sleep duration | |||||
| Number | 388 | 422 | 21 | .07 | .40 |
| Mean (SD) [95% CI] | 6.85 (1.11) [6.74-6.97] | 6.72 (0.93) [6.63-6.81] | 6.36 (1.34) [5.78-6.93] | ||
| Satisfied with mentoring received | |||||
| No | 111 (28.32) | 100 (23.47) | 11 (47.83) | .22 | >.99 |
| Yes | 268 (68.37) | 294 (69.01) | 9 (39.13) | ||
| Missing or unreported | 13 (3.32) | 32 (7.51) | 3 (13.04) | ||
| General workplace climate score | |||||
| Number | 379 | 423 | 19 | <.001 | <.001 |
| Mean (SD) [95% CI] | 3.68 (0.92) | 3.97 (0.85) [3.89-4.05] | 3.20 (0.97) [2.77-3.64] | ||
| DEI workplace climate score | |||||
| Number | 377 | 423 | 19 | <.001 | <.001 |
| Mean (SD) [95% CI] | 3.72 (0.82) [3.63-3.80] | 4.17 (0.67) [4.10-4.23] | 3.81 (0.90) [3.41-4.21] | ||
| Experience of sexual harassment | |||||
| None | 107 (27.30) | 231 (54.23) | 5 (21.74) | <.001 | <.001 |
| Yes | 281 (71.68) | 194 (45.54) | 13 (56.52) | ||
| Missing or unreported | 4 (1.02) | 1 (0.23) | 5 (21.74) |
Table notes: DEI indicates diversity, equity, and inclusion; NA, not applicable. P values were adjusted using the Holm method. Free-text responses for “not applicable” in sexual orientation included queer (n=2), man (n=1), and asexual (n=1). The “other” category for academic rank includes department chairs. Higher scores on both workplace climate scales indicate more favorable perceptions. The table uses “>8 h” as the criterion for a change in time allocation. The analysis in the main text described this as “an increase of 8 hours or more per week.”
The mean (SD) personal burnout scores were 46.6 (19.4) for women and 37.5 (17.2) for men. The corresponding work-related burnout scores were 43.7 (20.4) and 34.6 (19.7) (P<.001 for all). At the 50-point threshold, personal burnout prevalence was 46.7% (183/392) among women and 25.6% (109/426) among men, while work-related burnout prevalence was 40.8% (160/392) among women and 23.0% (98/426) among men (P<.001 for all).
In the bivariate analysis, increases in time spent on work and domestic labor, long work hours, short sleep duration, dissatisfaction with mentoring, a negative workplace climate, and experiences of sexual harassment were associated with burnout. The odds ratio for personal burnout among those who experienced sexual harassment was 2.20 (95% CI, 1.59–3.04). The odds ratio for work-related burnout was 2.84 (95% CI, 1.99–4.05).
In the multivariable analysis, the adjusted odds ratio (AOR) for personal burnout among women compared with men was 2.29 (95% CI, 1.54–3.41; P<.001). The AOR was 1.07 (1.00–1.15; P=.04) for a 5-hour increase in weekly clinical hours. It was 0.68 (0.56–0.81; P<.001) for a 1-hour increase in daily sleep and 0.64 (0.48–0.85; P=.002) for a 1-point increase in the general workplace climate score.
For work-related burnout, the AORs were 1.77 (95% CI, 1.17–2.69; P=.007) for women and 1.87 (1.13–3.08; P=.01) for an increase of 8 or more hours in weekly work since the pandemic. The AOR was 1.11 (1.03–1.19; P=.007) for a 5-hour increase in weekly clinical hours and 1.71 (1.11–2.62; P=.01) for experiences of sexual harassment in the past 2 years. It was 0.80 (0.66–0.96; P=.02) for a 1-hour increase in daily sleep and 0.49 (0.36–0.65; P<.001) for a 1-point increase in the general workplace climate score.
Although the proportion of missing data for each covariate was below 6%, the complete-case model included 690 individuals (82%). The magnitude and statistical significance of the associations remained similar after multiple imputation. The analyses of continuous scores showed the same patterns.
Discussion
The association between increased work since the pandemic and work-related burnout was consistent with previous research surveying over 40,000 healthcare workers. A study of K grant recipients from 2013 to 2019 also found that women had higher rates of burnout and stronger intentions to leave academia.
The general workplace climate encompassed kindness, respect, collegiality, collaboration, support, and a welcoming atmosphere. The estimated association for this scale was larger than those for 5-hour increases in weekly work, caregiving, or clinical hours. In a prior analysis of the same cohort, women rated the workplace climate more negatively and experienced more gender harassment.
The authors reviewed organizational responses including regular data collection and sharing, equity-centered policies, faculty support systems, and faculty development. A 10-year longitudinal study at a single institution by Wingard et al. from 2004 to 2015 found that although the proportion of female and URIM faculty increased, gender pay equity was not achieved. A 2019 study by Burns et al. reported an association between burnout and low self-efficacy in responding to unprofessional behavior. A 2012 survey of 13 medical schools by Carapinha et al. found that women’s support networks, trust in leadership, and reduced discrimination and work-family conflict were associated with a favorable climate.
Limitations
The analysis included those who remained in academia, and its generalizability is limited by the cohort’s composition: researchers who secured research funding and mentoring resources early in their careers. Despite the 65% response rate, selection bias remains possible. Recall bias could affect responses to questions about past time allocation. A single cutoff cannot capture the complexity of burnout. Different definitions make comparisons between studies difficult.
Conclusions
High levels of burnout among mid-career female faculty and gender disparities prompted suggestions for ongoing monitoring of post-pandemic experiences and evidence-based interventions to support female faculty throughout their careers.
Reading notes
General workplace climate and clinical hours were both associated with work-related burnout, with adjusted odds ratios of 0.49 for a 1-point increase in the climate score and 1.11 for a 5-hour increase in weekly clinical hours. Since a 1-point change in the climate scale and a 5-hour change in clinical hours represent different units, the two odds ratios are best interpreted as estimates of the association between each condition and burnout. Perceived workplace climate and burnout were self-reported at the same time. Burnout may influence assessments of the environment.
The adjusted odds ratio of 1.77 for work-related burnout among women in this study of 841 K08 and K23 career development grant recipients who remained in academia indicates a mid-career gender gap even among those who secured early funding and mentoring resources. The finding is based on CBI burnout scores from respondents currently working in academia. It describes their experiences during active employment, distinct from the effects of grant support or the process of leaving academia.
References
- Paradis, K. C., Kerr, E. A., Griffith, K. A., Cutter, C. M., Feldman, E. L., Singer, K., Spector, N. D., Ubel, P. A., & Jagsi, R. (2024). Burnout among mid-career academic medical faculty. JAMA Network Open, 7(6), e2415593. https://doi.org/10.1001/jamanetworkopen.2024.15593
Cite the paper under review
Paradis, Kelly C. 외 (2024). Burnout Among Mid-Career Academic Medical Faculty. JAMA Network Open, 7(6), e2415593.
@article{paradis2024,
author = {Paradis, Kelly C. and Kerr, Eve A. and Griffith, Kent A. and Cutter, Christina M. and Feldman, Eva L. and Singer, Kanakadurga and Spector, Nancy D. and Ubel, Peter A. and Jagsi, Reshma},
title = {Burnout Among Mid-Career Academic Medical Faculty},
journal = {JAMA Network Open},
year = {2024},
volume = {7},
number = {6},
pages = {e2415593},
doi = {10.1001/jamanetworkopen.2024.15593}
}TY - JOUR AU - Paradis, Kelly C. AU - Kerr, Eve A. AU - Griffith, Kent A. AU - Cutter, Christina M. AU - Feldman, Eva L. AU - Singer, Kanakadurga AU - Spector, Nancy D. AU - Ubel, Peter A. AU - Jagsi, Reshma TI - Burnout Among Mid-Career Academic Medical Faculty JO - JAMA Network Open VL - 7 IS - 6 SP - e2415593 PY - 2024 DO - 10.1001/jamanetworkopen.2024.15593 ER -