- Top of page
Medical Education 2012: 46: 815–822
Context Schema-based instruction may alter knowledge organisation and diagnostic reasoning strategies through the provision of structured knowledge to novice trainees. The effects of schema-based instruction on diagnostic accuracy and knowledge organisation have not been rigorously tested.
Methods Year 2 medical students were randomised to learn four cardiac diagnoses using schema-based instruction (n = 26) or traditional instruction (n = 27) on a high-fidelity cardiopulmonary simulator (CPS). Students completed case-based learning in groups of two to five and underwent individual written and practical tests. The written test consisted of questions testing features that linked or distinguished diagnoses (structured knowledge) and questions testing features of individual diagnoses (factual knowledge). A practical test of diagnostic accuracy on the CPS was performed for two diagnoses present in the learning phase (taught lesions) and two untaught lesions. A majority of students (n = 37, 70%) voluntarily returned for follow-up written testing 2–4 weeks later.
Results Learning time and accuracy did not differ between students on schema-based and those on traditional instruction. Students receiving schema-based instruction performed better on structured knowledge questions (p < 0.001) and no differently on factual knowledge questions (p = 0.7). Relative differences between groups remained unchanged on follow-up testing. Diagnostic success was higher in the schema-based instruction group for taught lesions (mean difference = 38%, 95% confidence interval [CI] 20–56; p < 0.001) and untaught lesions (mean difference = 31%, 95% CI 15–48; p < 0.001).
Conclusions Schema-based instruction was associated with improved retention of structured knowledge and diagnostic performance among novices. This study provides important proof-of-concept for a schema-based approach and suggests there is substantial benefit to using this approach with novice trainees.
- Top of page
To our knowledge, this study is the first prospective, randomised study to evaluate schema-based instruction among pre-clerkship learners. Our results provide strong support for the use of schema-based instruction in cardiac auscultation among novice students. We found that students instructed with a schema achieved higher diagnostic success than students instructed with the traditional framework. Of note, the 26% diagnostic accuracy achieved by the control group was comparable with rates achieved in previous evaluations of diagnostic accuracy among Year 2 medical students (19–24%),12,19 which suggests that this group was not unfairly disadvantaged by our design.
How do schemas benefit novices? The hypothesis that schemas benefit novices by modifying knowledge organisation and diagnostic reasoning strategies is supported by our data. The results on the written test provide evidence that schemas modified knowledge organisation by increasing the amount of structured knowledge without changing factual knowledge of diagnoses. The magnitude of the effect (38–40%) is similar to that in a previous study in which schemas were found to improve scores on nephrology questions by 30%.8 Although reasoning strategy was not directly measured (for fear of overwhelming novices with a verbal protocol), schema-instructed students were more likely to report features from the schema in arriving at their diagnosis. This suggests a higher use of schema-based reasoning by schema-instructed students.
Is there a downside to schema-based instruction? We anticipated lower diagnostic accuracy on untaught lesions among students instructed with schemas because they may be less likely to generate non-schema-based correct features. Although the traditional instruction group did generate a greater number of correct features for untaught lesions, this did not translate into higher diagnostic success. In fact, the schema-based instruction group had 30% higher diagnostic success. Two explanations are possible. Firstly, the traditional instruction group identified more incorrect features. Incorrect features may have a derailing effect on diagnostic reasoning. Secondly, the schema-based instruction group identified more schema-based features. Schema-based features may be more helpful in the diagnostic process than non-schema-based features as they help distinguish diagnoses. Importantly, we cannot generalise these findings to all diagnoses as we did not assess diagnoses or feature combinations not provided in the instructional material.
Our study has several limitations. Firstly, it was conducted at a single centre. Our conclusions would be strengthened by replication elsewhere. Secondly, practical testing was immediate. With further time to study and digest material, the benefit of a schema might be reduced. Thirdly, only novice learners participated. It is unclear how our findings might apply to experienced learners, who may already possess compiled knowledge. Fourthly, Harvey® has prominent and standardised physical examination findings that are not always present in real patients. This may have allowed students to make the correct diagnosis with fewer variables, thereby favouring the schema group. Finally, not all complex clinical problems are easily translated into schemas. Schema development requires the identification of a small number of useful clinical variables, which may not always be apparent.
In summary, this study provides important proof-of-concept for a schema-based approach to clinical instruction. Like the traditional framework used in this study, current teaching methods often emphasise comprehensiveness and diagnosis-specific knowledge. Our randomised study suggests a substantial benefit in organising this knowledge using schemas. Future work should explore whether this benefit is found across other domains, in more advanced learners, and continues to persist over time.
Acknowledgements: the authors would like to acknowledge Tobi Lam, Institute of Medical Science, Faculty of Medicine, University of Toronto, Toronto, Canada, for her assistance in designing the instructional frameworks, and the University of Toronto Medical School and the Herbert Ho Ping Kong Centre for Excellence in Education and Practice for their support of this study.