Cadaver-based and high-fidelity model-based anatomy education: effects on practical performance and transfer of learning in medical students

Cadaveric dissection remains a cornerstone of anatomy education, whereas high-fidelity anatomical models are increasingly used as alternative or complementary teaching tools. However, evidence comparing their effects on practical performance and transfer of learning is limited. This study compared cadaver-based and model-based anatomy teaching in terms of theoretical knowledge, practical skills, and transfer of learning among first-year medical students. This prospective, randomized, assessor-blinded study included 71 first-year medical students with no prior formal anatomy education. Participants were randomly allocated to a Cadaver Group ( n = 36), trained using formalin-fixed cadavers and prosections, or a Model Group ( n = 35), trained using high-fidelity anatomical models. After a standardized two-week extremity anatomy course, students completed a multi-station Objective Structured Practical Examination (OSPE) with a crossover design incorporating both training-congruent and training-incongruent assessment conditions to evaluate direct learning outcomes and transfer of learning across different material contexts. Data were analyzed using a two-way mixed-design ANOVA to evaluate the effects of training modality and assessment condition on practical performance. Within-group comparisons were performed using paired-samples t-tests, and transfer scores were calculated to quantify changes between training-congruent and training-incongruent conditions. Independent-samples t-tests were used for between-group comparisons, with p < 0.05 considered statistically significant. The Cadaver Group achieved significantly higher practical examination scores than the Model Group (7.4 ± 2.3 vs. 5.3 ± 3.1; p = 0.002), and total examination scores were also significantly higher (12.8 ± 3.9 vs. 10.3 ± 5.0; p = 0.022). No significant difference was observed between the groups in theoretical examination scores (5.7 ± 2.5 vs. 5.0 ± 2.3; p = 0.210). A significant interaction between training modality and assessment condition was identified by mixed-design ANOVA ( p < 0.001). The interaction effect size was large (η² = 0.433). Within-group comparisons showed that the Cadaver Group performed better under training-incongruent conditions, whereas the Model Group performed better under training-congruent conditions (both p ≤ 0.001). Transfer scores also differed significantly between groups ( p < 0.001), indicating different patterns of performance change across assessment contexts according to training modality. In conclusion, this study suggests that although high-fidelity anatomical models are effective for the acquisition of theoretical anatomy knowledge, cadaver-based education provides superior outcomes in practical performance. Explicit transfer analyses demonstrated that performance changes across assessment contexts differed according to training modality, suggesting different patterns of performance adaptation across assessment contexts. Nevertheless, these findings should be interpreted cautiously and confirmed in larger studies involving different anatomical regions and longer-term outcomes. Rather than functioning as competing approaches, cadaveric and model-based methods appear to provide complementary educational benefits. Integrating both approaches according to targeted learning outcomes and stages of learner development may provide the most effective educational strategy.