Scholars@Duke publication: Validity Evidence for Using Virtual, Interactive Patient Encounters to Teach and Assess Clinical Reasoning for First-Year Medical Students.

Validity Evidence for Using Virtual, Interactive Patient Encounters to Teach and Assess Clinical Reasoning for First-Year Medical Students.

Publication , Journal Article

Roberts, JK; Weigle, N; Fox, JW; Natesan, S; Gordon, D; Chudgar, SM

Published in: Acad Med

April 1, 2025

PURPOSE: Despite universal agreement on the importance of clinical reasoning skills, inadequate curricular attention to these skills remains a problem. To facilitate integration of clinical reasoning instruction and assessment into the preclerkship phase, the authors created a clinical reasoning curriculum using technology-enhanced patient simulations. METHOD: In 2023, first-year medical students at Duke University School of Medicine were enrolled in a biomedical science course using diagnostic reasoning sessions and 16 virtual, interactive patient (VIP) encounters to teach and assess clinical reasoning. The encounters were enhanced with interactive pop-in windows that assessed multiple clinical reasoning domains. Student responses were independently evaluated by faculty. Cumulative VIP clinical reasoning composite (CRC) scores were calculated, and growth mixture modeling was used to define students by growth trajectory. Clinical reasoning was assessed in a summative objective structured clinical examination (OSCE). RESULTS: Of the 118 students who participated in the curriculum, 1 was excluded from analysis for inadequate participation in the VIP encounters, leaving 117 students. The aggregate VIP encounter response rate was 95% (1,783 of 1,872 assessments completed). Clinical reasoning was assessed through cumulative performance across multiple domains. The mean (SD) scores were 58 (13) for information gathering, 46 (13) for illness script identification, 64 (14) for hypothesis generation, 59 (12) for differential diagnosis, and 77 (21) for management and plan. To identify latent classes of growth in cumulative VIP-CRC scores, growth mixture modeling was performed for 1-, 2-, and 3-class models. The 2-class model showed the best fit due to having the lowest bayesian information criterion (11,765.17) and Akaike information criteria (11,737.55). CONCLUSIONS: Integrated clinical reasoning instruction and deliberate, formative practice through authentic simulations were effective at teaching and assessing clinical reasoning in the preclerkship phase. VIP and OSCE can be used to identify students at risk of low performance in the clerkship year.