A Systems-Based Mixed-Methods Approach to Identifying Work-System Factors Contributing to Burnout Among Internal Medicine and Internal Medicine Pediatrics Residents

Background: Burnout among internal medicine (IM) and internal medicine pediatrics (Med-Peds) residents has reached epidemic levels. While burnout prevalence is well documented, limited research has employed systems-based approaches to identify the specific, contextually grounded work-system factors that drive burnout in these residents. This study aimed to identify and prioritize modifiable work-system factors contributing to burnout among IM and Med-Peds residents using a theory-based, participatory, and data-driven mixed-methods approach. Methods: A sequential mixed-methods design incorporating the National Academy of Medicine's (NAM) systems model of clinician burnout was employed at a single academic medical center across five phases: (1) survey, (2) focus groups, (3) contextual inquiry, (4) modeling and validation, and (5) prioritization and recommendations. IM and Med-Peds residents (N=119) were administered a 25-item survey including demographics, a 2-item Maslach Burnout Inventory (MBI) measuring emotional exhaustion (EE) and deperson-alization (DP), a 2-item Connor-Davidson Resilience Scale, and 21 work-system factor items rated for severity and improvement priority. Five focus groups gathered contextual information. Contextual inquiries (CIs) involving 4–6 hours of in-situ shadowing were conducted with 14 residents. Qualitative data were synthesized into an affinity model, which residents validated and used to rank their five priorities for improvement. Par-ticipants then rated each priority by level of impact and effort to generate prioritized improvement recommendations. Results: The survey response rate was 27% (n=32/119). Of respondents, 56% met criteria for burnout (combined EE and DP score >3), with mean EE of 3.84 (SD 1.19) and mean DP of 3.22 (SD 1.47). Mean resilience score was 8.09 (SD 0.96). The four workplace factors contributing most to burnout by severity were interruptions and distractions (mean 3.75, SD 0.95), excessive workload (3.59, SD 0.84), time pressure (3.38, SD 1.10), and poor work-life integration (3.28, SD 1.37). Focus groups (25%, n=30/119) and CIs (11.7%, n=14/119) provided rich contextual data organized into an affinity model with 167 discrete breakdowns across four major work-system categories. Validation sessions (6.7%, n=8/119) yielded 11 distinct improvement priorities. Impact-effort analysis identified high-impact, low-effort priorities (improve clinic workflows; reduce in-basket workload) and high-impact, high-effort priorities (improve work conditions; improve staffing; reduce patient load; improve Epic Chat and paging). Conclusions: This study demonstrates the feasibility and value of a contextual design approach for identifying actionable work-system factors contributing to resident burnout. By integrating quantitative data with rich qualitative observations, the methodology produces prioritized, context-specific recommendations for targeted improvement. Healthcare systems may utilize this participatory approach to efficiently identify and address the most impactful sources of resident burnout.