A Network Analysis of Clinical Reasoning after Femoral Transverse Fracture Surgery: A Comparison by Years of Experience

Clinical reasoning requires organizing extensive patient information. However, Japanese physical therapist schools provide an insufficient curriculum for clinical reasoning, leading to less-experienced physical therapists struggling with information structuring. We examined the differences in clinical reasoning between less-experienced and more-experienced physical therapists to clarify how years of experience influence their reasoning capabilities. 22 physical therapists with 1-13 years of experience were presented with simulated patient data after femoral transverse fracture surgery and asked to select the necessary tests, possible problems, and body parts. Two groups were compared: a less-experienced group with 1-2 years of experience and a more-experienced group with 6-13 years of experience. Their reasoning structure was visualized using NetworkX, a Python library. A directed graph of "patient->problem->body part->tests" paths was created, and the number of nodes, edges, and average degree were calculated. Nodes indicate the participant's problems, body parts, and tests. Edges indicate connections among them. We investigated the lack-free paths from the problem nodes to the test nodes. A less-experienced group had 92 nodes, 599 edges, and an average degree of 12.7. A more-experienced group had 90 nodes, 801 edges, and an average degree of 17.8. In the less-experienced group, the combination of problem, body part, and test concentrated on a "muscle function -> hip muscle -> MMT." In contrast, in the more- experienced group, not only the "muscle function" path but also pain-related paths, such as "pain sensation -> hip muscle -> NRS and tenderness," were examined. In the network analysis, the number of edges and the average degree were higher in the more-experienced group, suggesting that many tests were associated with a single problem or body part. A less-experienced group was biased toward "muscle function" problems and tests, while a more-experienced group mentioned not only "muscle function" but also "pain" problems and tests. These results showed that with years of experience, the patient's problems and necessary tests were considered more widely. Visualizing the structure of clinical reasoning as a network is effective for supporting clinical reasoning education because it highlights reasoning biases and the recognized relevance of problem-to-test combinations.