Purpose: To identify preferred burnout interventions within a resident physician population, utilizing the Nominal Group Technique. The results will be used to design a discrete choice experiment study to inform the development of resident burnout prevention programs.

Methods: Three resident focus groups met (10-14 participants/group) to prioritize a list of 23 factors for burnout prevention programs. The Nominal Group Technique consisted of three steps: an individual, confidential ranking of the 23 factors by importance from 1 to 23, a group discussion of each attribute, including a group review of the rankings, and an opportunity to alter the original ranking across participants.

Results: The total number of residents (36) were a representative sample of specialty, year of residency, and sex. There was strong agreement about the most highly rated attributes which grouped naturally into themes of autonomy, meaning, competency and relatedness. There was also disagreement on several of the attributes that is likely due to the differences in residency specialty and subsequently rotation requirements.

Conclusion: This study identified the need to address multiple organizational factors that may lead to physician burnout. There is a clear need for complex interventions that target systemic and program level factors rather than focus on individual interventions. These results may help residency program directors understand the specific attributes of a burnout prevention program valued by residents. Aligning burnout interventions with resident preferences could improve the efficacy of burnout prevention programs by improving adoption of, and satisfaction with, these programs.

Focus group discussions were audio-recorded and uploaded to a secure server for storage. Audio files were transcribed and deidentified. Deidentification involved removing participant names. Research team reviewed transcripts for quality (i.e. accuracy) and confidentiality (i.e., deidentification). We used a 6-step thematic analysis approach to capture the diversity and depth of the data relating to the reasons underpinning the ranking of outcomes. For step 1, we familiarized ourselves with the transcripts. For step 2, we confirmed the selection of codes and themes and made necessary amendments to reach a consensus. Initial coding was performed in Atlas.ti based on the deductive coding framework, with varied responses interpreted inductively into new codes as needed. Codes were matched between authors to ensure consistency and confirm the definition of the full set of themes. In step 3, we clustered nodes into a common theme based on coherent patterns. We used some of the quotes in the Results section to demonstrate the legitimacy of the defined themes. For step 4, we reduced the themes into the most prevalent implicit and explicit ideas while deleting redundant themes. In step 5, we described the names and parameters of the identified themes. For step 6, we reported the analysis performed.

See attached Figures 1 & 2, and ReadMe file for additional information.