What physicians know about quality: Transmission of specialist preferences from preceptors to residents
Date:
Pany, M., McWilliams, J. M., Weinreb, G. G.
Note: this research project evolved between the abstract submission and final poster presentation. The latter supersedes.
Research Objective: Although evidence suggests physicians learn from the practice patterns of peers around them, physician practices vary widely within the same practice or institution and have not converged to evidence-based standards to the extent desired. This suggests the potential for structuring intra-professional learning to elevate clinical decision-making in aggregate by directing clinicians acting under uncertainty to coworkers who exemplify desired practices and can effectively transmit them, or to coaches who are tasked with knowing and teaching them. Outpatient medical training pairs residents with a longitudinal preceptor, presenting an opportunity to explore the potential gains from structuring intra-professional information flow, but little is known about the transmission of practice patterns from preceptor to resident. In this study, we examine transmission of specialist preferences from preceptors to residents in the setting of a large primary care teaching practice.
Study Design: Using data from physician shift schedules and EHRs of a large academic primary care practice in 2016– 2019, we leveraged quasi-random assignment of residents to preceptors to examine similarity between residents and preceptors in the specialists they selected in referral orders across different levels of longitudinal resident-preceptor interactions. Specifically, we compared similarity in specialist preferences across residents who (1) shared patients with their primary preceptors, (2) shared patients with other preceptors, (3) shared shifts but not patients with preceptors, and (4) did not share patients or shifts with preceptors (reference group). Because residents are assigned to preceptors without regard to patient populations, similarity in referral preferences across resident-preceptor dyad groups should not vary with strength of resident-preceptor interaction unless preceptors transmit their preferences for specialists to residents. We assessed differences in patient characteristics across dyad groups, and used generalized linear models with a logit link function and robust standard errors for our main analysis.
Population Studied: A total of 102,518 clinic encounters for 18,586 patients occurred over 9,522 shifts with 2,344 unique resident-preceptor dyads consisting of 127 residents and 49 preceptors over 2016–2019.
Principle Findings: Physician and patient characteristics were similar across dyad groups. Among resident-preceptor dyads who did not share shifts, the mean preference dissimilarity score was 0.21. Strength of residentpreceptor interaction was associated with similarity of preferences for specialists. Compared to residentpreceptor dyads without shared shifts, residents who shared patients with their primary preceptors had, on average, 72.7% (95%CI,9.4%-135.9%;p=0.025) more similar preferences. Residents who shared patients with other preceptors and residents who shared shifts but not patients with preceptors had, respectively, a 42.2% (95%CI,-5.5-89.8%; p=0.08) and a 31.2% (95%CI,-16.4-78.9%;p=0.2) higher preference similarity score compared to resident-preceptor dyads without shared shifts.
Conclusions: Specialist preferences were more similar among residents and preceptors who interacted more. Given that physicians were quasi-randomly assigned to each other and given that patient characteristics were balanced across resident-preceptor dyads, these findings likely reflect transmission of specialist preferences from preceptors to residents as opposed to confounding by patient or disease characteristics.
Implications for Policy or Practice: That residents’ specialist preferences converge to those of the preceptors they interact with most exemplifies an important mechanism of practice pattern formation that could be leveraged to facilitate more effective peer learning among physicians beyond residency.