Abstract
This survey study examines the patterns of work-related burnout in physician-scientists.
Physician burnout, a pathological syndrome characterized by emotional exhaustion, depersonalization, and a reduced sense of personal accomplishment in response to prolonged occupational stress, is a national priority.1,2 Burnout appears to be more common in certain physician subgroups, including women.1 We extend prior research by reporting on the patterns of work-related burnout in physician-scientists.
Methods
As detailed elsewhere,3 the University of Michigan institutional review board provided approval and a waiver of documentation of consent. Between August 2010 and February 2011, we surveyed 1708 individuals who received new K08 and K23 awards from the National Institutes of Health during 2006 to 2009. In 2014, we conducted a follow-up survey that was administered exclusively to respondents to the initial questionnaire; 1066 (62.0% of the originally targeted population) responded to both surveys.3
We report analyses from the 816 physician-respondents who remained in academic positions at the time of follow-up. Multivariable logistic regression modeling evaluated individual, job, and environmental independent variables measured at baseline that were associated with the dependent variable of work-related burnout measured at follow-up using the validated Copenhagen Burnout Inventory subscale,4 with scores of 50 or greater taken to signify burnout, as in prior work.4,5 Analyses were conducted using SAS, version 9.4 (SAS Institute) and statistical significance was set at P < .05.
Results
The mean (SD) age of the 816 participants at baseline was 40.2 (3.7) years; other sample characteristics are detailed by sex in Table 1. Men were more likely to provide a positive appraisal of their work climate (scores of >4 among 35.8% of women vs 49.4% of men; P < .001). Women reported higher weekly hours spent on parenting and domestic tasks than men (≥44 hours by 41.7% of women vs 17.1% of men; P < .001) and lower weekly hours of patient care (≥15 hours by 14.2% of women vs 24.1% of men; P = .007) and work overall (≥60 hours by 34.6% of women vs 59.2% of men; P < .001).
Table 1. Distribution of Select Sample Characteristics at Baseline by Sexa.
| Characteristic | No. (%) | P Value | ||
|---|---|---|---|---|
| Total (N = 816) | Women (n = 340) | Men (n = 476) | ||
| Age, mean (SD), y | 40.2 (3.7) | 39.9 (3.7) | 40.5 (3.6) | .05 |
| No. | 811 | 338 | 473 | |
| Race/ethnicityb | ||||
| White | 559 (68.5) | 222 (65.3) | 337 (70.8) | .21 |
| Asian | 202 (24.8) | 91 (26.8) | 111 (23.3) | |
| Other | 55 (6.7) | 27 (7.9) | 28 (5.9) | |
| Degree | ||||
| MD | 583 (71.4) | 273 (80.3) | 310 (65.1) | <.001 |
| MD/PhD | 233 (28.6) | 67 (19.7) | 166 (34.9) | |
| Specialty | ||||
| Clinical specialties for women, children, and families | 215 (26.4) | 113 (33.4) | 102 (21.4) | <.001 |
| Hospital-based | 124 (15.2) | 45 (13.3) | 79 (16.6) | |
| Surgical | 59 (7.3) | 9 (2.7) | 50 (10.5) | |
| Medical | 416 (51.1) | 171 (50.6) | 245 (51.5) | |
| Grant type | ||||
| K08 | 440 (53.9) | 133 (39.1) | 307 (64.5) | <.001 |
| K23 | 376 (46.1) | 207 (60.9) | 169 (35.5) | |
| Work Climate Scale scorec | ||||
| ≤3 | 83 (10.2) | 51 (15.1) | 32 (6.8) | <.001 |
| >3-≤4 | 374 (46.1) | 166 (49.1) | 208 (43.9) | |
| >4-≤5 | 355 (43.7) | 121 (35.8) | 234 (49.4) | |
| Marital status | ||||
| Married/domestic partner | 745 (91.5) | 304 (89.7) | 441 (92.8) | .02 |
| Single | 49 (6.0) | 29 (8.6) | 20 (4.2) | |
| Divorced/widowed | 20 (2.5) | 6 (1.8) | 14 (3.0) | |
| Parenting and domestic task weekly, h | ||||
| <22 | 196 (24.4) | 65 (19.4) | 131 (27.9) | <.001 |
| 22-<34 | 197 (24.5) | 49 (14.6) | 148 (31.6) | |
| 34-<44 | 192 (23.9) | 82 (24.4) | 110 (23.5) | |
| ≥44 | 220 (27.3) | 140 (41.7) | 80 (17.1) | |
| Nightly hours of sleep | ||||
| <6 | 75 (9.3) | 26 (7.7) | 49 (10.4) | .01 |
| 6-<7 | 281 (34.7) | 99 (29.3) | 182 (38.5) | |
| 7-<8 | 338 (41.7) | 154 (45.6) | 184 (38.9) | |
| ≥8 | 117 (14.4) | 59 (17.5) | 58 (12.3) | |
| No. of children | ||||
| None | 141 (17.3) | 68 (20.0) | 73 (15.4) | <.001 |
| 1 | 106 (13.0) | 50 (14.7) | 56 (11.8) | |
| 2 | 379 (46.5) | 170 (50.0) | 209 (44.0) | |
| 3 | 155 (19.0) | 48 (14.1) | 107 (22.5) | |
| ≥4 | 34 (4.2) | 4 (1.2) | 30 (6.3) | |
| Vacation, wk | ||||
| <2 | 112 (13.8) | 45 (13.3) | 67 (14.2) | .75 |
| 2-<3 | 261 (32.2) | 107 (31.7) | 154 (32.6) | |
| 3-<4 | 212 (26.1) | 85 (25.2) | 127 (26.9) | |
| 4 or more | 226 (27.9) | 101 (29.9) | 125 (26.4) | |
| Typical weekly work hours | ||||
| <50 | 103 (12.7) | 70 (20.7) | 33 (7.0) | <.001 |
| 50-<55 | 225 (27.6) | 120 (35.4) | 105 (22.1) | |
| 55-<60 | 98 (12.0) | 42 (12.4) | 56 (11.8) | |
| 60 or more | 388 (47.7) | 107 (34.6) | 281 (59.2) | |
| Typical weekly hours spent on patient care | ||||
| <5 | 164 (20.2) | 74 (21.8) | 90 (19.0) | .01 |
| 5-<10 | 270 (33.2) | 122 (36.0) | 148 (31.2) | |
| 10-<15 | 217 (26.7) | 95 (28.0) | 122 (25.7) | |
| ≥15 | 162 (19.9) | 48 (14.2) | 114 (24.1) | |
Abbreviations: MD, Doctor of Medicine degree; PhD, Doctor of Philosophy.
Distributions exclude missing or unknown responses when present. The amount of missing data because of item nonresponse was low (<2% for all questions).
Race/ethnicity was defined in this sample by self-report from participants using options defined by the investigator (white, Asian, black or African American, Hispanic/Latino, American Indian or Alaska Native, Native Hawaiian or Pacific Islander, or Other with a write-in option for specific information). These were then categorized into 3 groups: any individual who indicated any category other than white or Asian was categorized as Other, then any individual who indicated Asian was categorized as Asian, and the remaining individuals who indicated only white were categorized as white.
Perceived work climate was assessed using a scale averaging responses to 7 questions that evaluate different aspects of the work climate, as developed by the University of Michigan ADVANCE program (https://advance.umich.edu/wp-content/uploads/2018/09/ADVANCE-2006-R1-FullReport.pdf): (1) “Racist” to ‘“Non-racist,” (2) “Homogenous” to “Diverse,” (3) “Non-sexist” to “Sexist,” (4) “Collaborative” to “Individualistic,” (5) “Cooperative” to “Competitive,” (6) “Homophobic” to “Non-homophobic,” and (7) “Not supportive” to “Supportive.” The 5-point response scale ranged from 1 to 5 and questions 3, 4, and 5 were inverse coded so that the higher value represented the more desirable climate (eg, “Collaborative” rather than “Individualistic”). The Cronbach α for the total scale in this sample was 0.75.
Continuous scores for work-related burnout on the Copenhagen Burnout Inventory were a mean (SD) of 45.3 (14.2) among women and 42.2 (14.4) among men. Work-related burnout (using the prespecified cutoff of 504,5) was reported by 139 of 336 women (41.4%) and 149 of 473 men (31.5%; P = .004).
On multivariable analysis (Table 2), factors associated with burnout were greater time spent on parenting and domestic tasks (odds ratio [OR], 1.98; 95% CI, 1.30-3.01 for ≥44 vs <22 weekly hours; P = .01), less vacation time (OR, 1.43; 95% CI, 1.06-1.93 for <3 vs ≥3 weeks; P = .02), more time spent on patient care (OR, 1.83; 95% CI, 1.14-2.92 for ≥15 vs 1-5 weekly hours; P = .04), and a negatively perceived work climate (OR, 2.00; 95% CI, 1.21-3.31 for Work Climate Scale scores of ≤3 vs >4-≤5).
Table 2. Bivariable and Multivariable Associations of Baseline Characteristics With Burnout.
| Characteristic | Bivariable Models | Multivariable Modela | ||
|---|---|---|---|---|
| Estimate (95% CI) | P Value | Estimate (95% CI) | P Value | |
| Sex | ||||
| Men | 1 [Reference] | .004 | NA | NA |
| Women | 1.53 (1.15-2.05) | NA | NA | |
| Parenting and domestic tasks weekly, h | ||||
| <22 | 1 [Reference] | .02 | 1 [Reference] | .01 |
| 22-<34 | 1.16 (0.75-1.78) | 1.29 (0.80-1.94) | ||
| 34-<44 | 1.42 (0.93-2.18) | 1.46 (0.94-2.26) | ||
| ≥44 | 1.85 (1.23-2.79) | 1.98 (1.30-3.01) | ||
| Vacation, wk | ||||
| <3 | 1.36 (1.02-1.82) | .04 | 1.43 (1.06-1.93) | .02 |
| ≥3 | 1[Reference] | 1 [Reference] | ||
| Weekly hours spent on patient care | ||||
| <5 | 1 [Reference] | .09 | 1 [Reference] | .04 |
| 5-<10 | 1.27 (0.74-1.72) | 1.13 (0.73-1.74) | ||
| 10-<15 | 1.38 (0.90-2.14) | 1.46 (0.93-2.27) | ||
| ≥15 | 1.71 (1.08-2.70) | 1.83 (1.14-2.92) | ||
| Work Climate Scale score | ||||
| ≤3 | 2.09 (1.29-3.40) | .01 | 2.00 (1.21-3.31) | .02 |
| >3-≤4 | 1.10 (0.81-1.50) | 1.07 (0.78-1.47) | ||
| >4-≤5 | 1 [Reference] | 1 [Reference] | ||
Abbreviations: NA, not applicable; NIH, National Institutes of Health.
Logistic regression model of burnout (defined as a Copenhagen Burnout Inventory work-related burnout scale score of ≥50) as the dependent variable. Independent variables measured at baseline by self-report and considered for inclusion were: sex, age, race/ethnicity, marital status, professional degree, specialty, academic rank, weekly hours spent on parenting and domestic tasks, nightly hours of sleep, presence and number of children, time spent on vacation, typical weekly work hours, typical weekly hours in patient care activities, frequency of K award mentor communications, satisfaction with primary K award mentor, and perceived work climate. Independent variables measured at baseline using public information were grant type, year of grant award, tier of institution at time of K award (based on total extramural NIH funding as previously described), and K-awarding NIH institute tier (based on the total amount of R01 awards granted by the institute). The final model was created after iterative removal of variables with model recalculation until only important covariates remained, as measured by comparing the Akaike information criterion between nested models. The model listed in the table is the final model; sex was not selected for inclusion in the final multivariable model but is included in the table to demonstrate its bivariable association because it was the primary independent variable of interest in the analysis.
Discussion
We observed work-related burnout in many men and women physician-scientists. The finding that sex was significantly associated with burnout on bivariable analysis but not on multivariable analysis suggests that the greater prevalence of burnout among women is driven by other differences in the experiences of men and women early in their careers. Specifically, we observed burnout to be associated with early-career reports of substantial competing demands (at home and work) and early-career perceptions of work climates. These findings suggest that mitigating competing demands and improving work climates might alleviate burnout and that such interventions may be particularly important for women, who are more likely to experience burnout than their male peers.1
As in any observational study, associations may not necessarily imply causation; it is possible that a common underlying confounding factor led to a negative appraisal of work climates and to an endorsement of burnout, although the longitudinal design mitigates this concern. Our findings may not be generalizable to physician-scientists who do not hold career development awards. However, the fact that many of those with considerable support for career development experience burnout suggests that those without such support may experience even higher levels of burnout than the sample described in this article.2
Conclusions
Our finding that perceptions of work climate are associated with burnout adds to the motivations to improve civility and respect in the academic medical workplace.6 Future research should focus on tailored, multifaceted systems-level interventions to decrease work-related burnout, which has substantial consequences not only for individual physician-scientists themselves but also the broader society they serve.
References
- 1.Templeton K, Bernstein CA, Sukhera J, et al. Gender-based Differences in Burnout: Issues Faced by Women Physicians. Washington, DC: National Academy of Medicine; 2019:1-16, https://nam.edu/gender-based-differences-in-burnout-issues-faced-by-women-physicians/. Accessed June 21, 2019. [Google Scholar]
- 2.Shanafelt TD, Hasan O, Dyrbye LN, et al. Changes in burnout and satisfaction with work-life balance in physicians and the general US working population between 2011 and 2014. Mayo Clin Proc. 2015;90(12):1600-1613. doi: 10.1016/j.mayocp.2015.08.023 [DOI] [PubMed] [Google Scholar]
- 3.Jagsi R, Griffith KA, Jones RD, Stewart A, Ubel PA. Factors associated with success of clinician-researchers receiving career development awards from the National Institutes of Health: a longitudinal cohort study. Acad Med. 2017;92(10):1429-1439. doi: 10.1097/ACM.0000000000001728 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kristensen TS, Borritz M, Villadsen E, Christensen KB. The Copenhagen Burnout Inventory: a new tool for the assessment of burnout. Work Stress. 2005;19(3):192-207. doi: 10.1080/02678370500297720 [DOI] [Google Scholar]
- 5.Klein J, Grosse Frie K, Blum K, von dem Knesebeck O. Burnout and perceived quality of care among German clinicians in surgery. Int J Qual Health Care. 2010;22(6):525-530. doi: 10.1093/intqhc/mzq056 [DOI] [PubMed] [Google Scholar]
- 6.National Academies of Sciences, Engineering, and Medicine . 2018. Sexual Harassment of Women: Climate, Culture, and Consequences in Academic Sciences, Engineering, and Medicine. Washington, DC: The National Academies Press; 2018. [PubMed] [Google Scholar]