Are Baseline Learning Approaches Associated with Cognitive Load, and Do They Moderate the Cognitive Effects of Contextual Vertical Integration? A Prespecified Secondary Analysis of the CONVINCED-OG Trial.

Introduction: Effective clinical teaching depends on managing cognitive load. Contextual vertical integration (CVI), which interweaves basic-science mechanisms with clinical content within a teaching session, was associated with lower self-reported Intrinsic Cognitive Load (ICL) and Extraneous Cognitive Load (ECL) scores and higher self-reported Germane Cognitive Load (GCL) scores in the CONVINCED-OG cluster-randomized crossover trial. Whether baseline deep and surface learning approaches are associated with cognitive load, or moderate CVI's benefits, is unknown. This prespecified secondary analysis addressed both questions. Materials & Methods: The dataset comprised 103 third-year MBBS students (824 student-topic observations) taught eight core obstetrics and gynecology topics across two teaching hospitals, each topic delivered under CVI-enabled and non-integrated conditions in a topic-level crossover. Baseline deep and surface learning approaches were measured once using the Revised Two-Factor Study Process Questionnaire; ICL, ECL and GCL were self-reported after each session using the Cognitive Load Questionnaire. For each outcome, two linear mixed-effects models were fitted with random intercepts for participant and cluster-period: a main-effects model, and an interaction model adding condition-by-approach interaction terms. Results: Averaged across instructional conditions, a deeper learning approach was associated with lower ICL (b = -0.040; 95% CI, -0.071 to -0.009; p = 0.011), lower ECL (b = -0.060; 95% CI, -0.101 to -0.019; p = 0.004) and higher GCL (b = 0.037; 95% CI, 0.003 to 0.072; p = 0.033). There was no clear evidence that surface approach was associated with any outcome. No interaction between instructional condition and either learning approach was significant for any outcome, giving no clear evidence that CVI's cognitive-load benefit varied with learning approach. Sensitivity analyses were consistent. With multiplicity adjustments, ECL survived all adjustments, ICL only Benjamini-Hochberg, and GCL none. Conclusion: There was no clear evidence of moderation of CVI-related differences by learning approaches. Deep-approach scores were associated with lower ICL and ECL and higher GCL; surface approach showed no clear association.