Burnout pathway differences across dual-track residency training: a network analysis.

Abstract

Background: China's standardized residency training operates under a dual-track system: one track integrates clinical training with concurrent degree pursuit (student group), while the other is purely vocational without degree conferral (non-student group). Prior burnout research has largely relied on aggregate total or subscale scores, which cannot detect qualitative differences in symptom-to-symptom connectivity even when mean burnout severity is similar. Despite identical clinical rotations and competency requirements, it remains unclear whether these structurally distinct training pathways are associated with different burnout symptom-connectivity patterns, with direct implications for track-specific educational support.

Methods: We conducted a secondary analysis of data from Fan et al.'s previously published cross-sectional survey of 636 residents from a Chinese tertiary teaching hospital. Burnout was measured using the 22-item Maslach Burnout Inventory-Human Services Survey (MBI-HSS). Networks were estimated using Graphical Lasso, and bridge centrality was computed to identify symptoms that most strongly connect burnout dimensions. Analyses were stratified by gender and specialty, and supplemented by logistic regression and bootstrap resampling.

Results: Non-students showed higher emotional exhaustion and lower personal accomplishment scores than students, whereas depersonalization scores did not differ significantly between groups. Burnout prevalence was higher among non-students (50.2% vs. 42.4%), but this difference was not statistically significant. In both groups, DP2 ("I don't really care what happens to patients") was the dominant bridge node. However, PA4 ("I don't easily understand how my patients feel") entered the top five bridge items only in non-students. Three of four subgroups followed the DP2 pathway, whereas non-student males showed a distinct pattern with PA4 as the most prominent feature (BS = 1.445). Logistic regression found no significant association between training identity and burnout status after adjustment (OR = 1.07; ΔAUC = +0.002, p = 0.68), whereas network analyses revealed distinct symptom-connectivity patterns.

Conclusion: These findings suggest that burnout prevention should account for training structure and institutional context. Rather than uniform wellness programs, medical education should offer track-specific support, particularly for residents whose burnout is characterized by empathy-related frustration rather than depersonalization.