How do junior doctors' work characteristics relate to burnout and work engagement? Results from a German longitudinal cohort study

Background: Hospital doctors are at increased risk for stress-related illnesses [1]. Previous research shows that work conditions in the hospital are associated with doctors' ill health in terms of burnout and depression [2,3,4]. The Job Demands-Resources model (JD-R) postulates different premises for the impact of work conditions on ill-health [5,6,7]. Drawing on the JD-R model, we hypothesized that job demands and job resources are longitudinally related to burnout and work engagement, respectively.

Methods: The sample is part of a prospective cohort study of German junior doctors undergoing specialty training. We used follow-up data with a two-year time range. Information on work conditions were assessed with the established self-report tool Tätigkeits- und Arbeitsanalyseverfahren für das Krankenhaus Selbstbeobachtungsverfahren (TAA-KH-S). Job demands were operationalized with time pressure, interruptions by persons, and additional effort due to adverse working conditions. Relevant job resources were represented by social climate, autonomy, and supervisor feedback. Burnout (emotional exhaustion and depersonalization) was measured through the German translation of the Maslach Burnout Inventory (MBI-D) scales. Work engagement (vigor and absorption) was assessed with the Utrecht Work Engagement Scale (UWES). Control variables were age, sex, part-time employment, job tenure, and job change. Cross-lagged panel analyses between work conditions, burnout and work engagement were conducted with structural equation modeling (AMOS Version 22.0).

Results: Data from 507 doctors were included. At T1, 51.3% were female, mean age was 30.48 years (SD: 2.64), and mean job tenure at T2 amounted to 5.23 years (SD: 1.5). Different structural models were compared: The model with autoregressive paths, M0: χ²(df)= 357.32(204), and the causal model (M1, consistent with the JD-R hypotheses) showed similar good fit: χ²(df)= 355.50(202). However, the model with reverse associations (M2: χ²(df)= 346.77(202) and the model with complete lagged associations (M4: χ²(df)= 328.76(192) showed significantly improved fit. The reciprocal model (M3: χ²(df)= 344.39(200) was not superior to M2. Goodness-of-fit indices for M2 and M4 were good, e.g. for M2, IFI= 0.97, TLI= 0.96, CFI= 0.97, RMSEA= 0,038 (CI: 0,031 – 0,044). Concerning the estimated associations in the reverse model (M2), work engagement at baseline predicted job resources at follow-up (β= 0,18, p = 0,002). Further analyses (for M4) revealed that burnout and engagement at baseline predicted job demands (β= 0,28, p = 0,03) as well as job resources at follow-up (β= 0,21, p = 0,04). Additionally, job demands at baseline predicted work engagement at follow up (β= 0,23, p = 0,02).

Discussion: Our prospective results demonstrate that work conditions in hospitals have a significant association with junior doctors' burnout and engagement. In contrast to the JD-R approach, we conclude that burned-out doctors' inwardly experience increased job demands over time, while engaged doctors are able to elevate their job resources. Thus, attempts to strengthen well-being are advised to foster doctors' capabilities to manage work load efficiently and to promote their abilities to gain resources for successful coping. A discussion of potential limitations of the study rounds up the contribution.