Integration of a focused podcast curriculum (PC) to improve hematology oncology fellow (HOF) knowledge: A multi-center cluster randomized controlled trial.

Abstract

9041 Background: Several studies demonstrate widespread use of medical podcasts as learning resources. Utility of integration of educational podcasts in fellowship curricula remains unknown. Here we describe a multicenter cluster randomized trial comparing a supplemental PC with standard curriculum (SC) to SC alone for HOF. Methods: HO fellowship programs in the United States were eligible. One HOF from each program was invited via email to participate as a local PI. The trial protocol was reviewed by program directors for approval. Programs were cluster randomized to novel PC with SC (intervention) or SC alone (control). Permuted block randomization stratified by program size was performed using STATA. The PC arm was given access to a website developed by the study team with links to podcasts episodes (PE) covering 4 topics: breast oncology, myeloma, bleeding disorders, and HIT. Written show notes (SN) were provided as separate links. PE and SNs were previously developed employing an inductive approach and released by The Fellow on Call and Two Onc Docs podcasts and are freely available. Pre and post qualitative surveys (QS) and knowledge assessments (KA) were administered via REDCAP. QS utilized a 7-point Likert score. KAs were peer reviewed by expert disease-specific faculty. QS and KA links were included on the PC website. To maintain blinding, the SC arm was given a separate website with these links. HOF participation was voluntary. To improve enrollment, at minimum, QS completion was required; KA was optional. Baseline data was obtained from October to November 2023. Post data cut off will be June 2024. Power calculation was based on mean difference in comfort level and knowledge. Assuming a mean improvement by 0.5 points for QS, a total sample size of 210 HOF, 11 clusters in each arm would provide at least 80% power assuming a standard deviation (SD) of 1, intracluster correlation coefficient (ICC) of 0.05, and coefficient of variation of cluster sizes of 0.7. For improvement in KA by 10%, a total sample size of 60 HOF would provide the same power assuming SD of 15%. We therefore planned to recruit 30 HO programs to account for non-response rates. IRB review deemed the study exempt. Results: Baseline characteristics from 28 randomized programs are presented below. Conclusions: This is the first randomized education trial in graduate medical education of its kind with a priori power calculations. Our innovative use of HOF as local PIs may have led to increased participation in KAs. Recruited sites will serve as a consortium for future multicenter education research. Interim analysis will be presented at ASCO 2024. [Table: see text]