Burnout in Graduate Medical Education: Uncovering Resident Burnout Profiles Using Cluster Analysis

Burnout is common among residents and negatively impacts patient care and professional development. Residents vary in terms of their experience of burnout. Our objective was to employ cluster analysis, a statistical method of separating participants into discrete groups based on response patterns, to uncover resident burnout profiles using the exhaustion and engagement sub-scales of the Oldenburg Burnout Inventory (OLBI) in a cross-sectional, multispecialty survey of United States medical residents.

The 2017 ACGME resident survey provided residents with an optional, anonymous addendum containing 3 engagement and 3 exhaustion items from the OBLI, a 2-item depression screen (PHQ-2), general queries about health and satisfaction, and whether respondents would still choose medicine as a career. Gaussian finite mixture models were fit to exhaustion and disengagement scores, with the resultant clusters compared across PHQ-2 depression screen results. Other variables were used to demonstrate evidence for the validity and utility of this approach.

From 14 088 responses, 4 clusters were identified as statistically and theoretically distinct: Highly Engaged (25.8% of respondents), Engaged (55.2%), Disengaged (9.4%), and Highly Exhausted (9.5%). Only 2% of Highly Engaged respondents screened positive for depression, compared with 8% of Engaged respondents, 29% of Disengaged respondents, and 53% of Highly Exhausted respondents. Similar patterns emerged for the general query about health, satisfaction, and whether respondents would choose medicine as a career again.

Clustering based on exhaustion and disengagement scores differentiated residents into 4 meaningful groups. Interventions that mitigate resident burnout should account for differences among clusters.