Bias Experiences and Burnout: A Multispecialty, Cross-Sectional Survey of Residents and Fellows at a Multisite US Academic Medical Center.

OBJECTIVE

To explore associations of bias frequency, sources, and types with burnout in a large, multispecialty sample of residents and fellows and to determine whether and how odds of burnout change after adjustment for bias experiences in multiple demographic subgroups.

METHODS

Trainees in graduate medical education programs at Mayo Clinic sites were surveyed between October 12, 2020, and November 22, 2020. Survey items measured personal experiences with bias (frequency, sources, types), burnout (2 Maslach Burnout Inventory items), and demographic characteristics (age, gender, race/ethnicity, lesbian-gay-bisexual-transgender-queer-nonbinary-other identification, disability, socioeconomic background, year in school, specialty). The χ2 test and logistic regression analyses examined relationships between variables.

RESULTS

Of 1825 trainees surveyed, 942 (52%) from 77 programs responded. Overall, 16% (137/881) of respondents reported 1 or more personal bias experiences. Trainees reporting bias experiences weekly or more often had markedly higher odds of burnout in adjusted analyses (odds ratio [OR], 8.00; 95% CI, 2.68 to 23.89; P<.001). Bias from education leaders/staff, fellow learners, and faculty was independently associated with burnout, whereas bias from patients/companions and other staff/employees was not. Trainees with a disability (OR, 3.16; 95% CI, 1.05 to 9.53; P=.04) and trainees from a low-income background (OR, 1.53; 95% CI, 1.03 to 2.26; P=.03) had higher odds of burnout in unadjusted analyses, but these associations were no longer statistically significant after adjustment for bias frequency.

CONCLUSION

Self-reported bias experiences relate strongly to trainee burnout. Some bias sources may be more strongly associated with burnout than others. More frequent bias experiences could account, at least in part, for higher odds of burnout in some demographic subgroups.