Qualitative differences in knowledge structure are associated with diagnostic performance in medical students

Diagnosing is a knowledge-based skill: to diagnose one must retrieve knowledge from long-term memory and then apply this to a new clinical problem. Prior research on expertise found differences in knowledge structure between experts and novices, and it is assumed that the superior diagnostic performance of experts is somehow related to their superior knowledge structure. Here our objective was to study knowledge structure in final year medical students and to examine the association between knowledge structure and diagnostic performance. Ninety-one students participated. We used concept sorting to assess knowledge structures for four clinical problems. We performed qualitative analysis of knowledge structures, categorizing these as either problem-specific, where knowledge was predominantly structured around concepts specific to that clinical problem, or generic, where knowledge was structured around general concepts that could apply to all clinical problems. We evaluated diagnostic performance using problem-solving questions. Knowledge structure varied between different problems, but for each problem most students had problem-specific knowledge structure. These students had better diagnostic performance than those with generic structure (68.5 vs. 55.3%, d = 0.45, P = 0.004). This difference persisted after adjusting for overall medical knowledge (performance on the Medical Council of Canada Part 1 examination) and clinical problem. We found that most students organize their knowledge around problem-specific concepts, and that this type of knowledge was associated with better diagnostic performance. This may be due to easier knowledge retrieval if there is congruence between how knowledge is stored and how it is applied when diagnosing.