Cognitive biases and clinical reasoning errors among medical students in a simulated leprosy case: a cross-sectional decision-tree study

Abstract Background Neglected tropical diseases carry a substantial global burden yet remain underrepresented in medical curricula. Leprosy poses a particular stress test for clinical reasoning: diagnosis depends on careful neurological assessment, yet training increasingly privileges imaging and laboratory investigations over bedside evaluation. We examined how medical students at different training stages reason through an atypical presentation of leprosy. Methods We mapped the clinical decision-making of 405 medical students from 14 Brazilian medical programmes using an interactive decision tree with 22 diverging pathways, simulating an atypical neurological presentation of leprosy. Participants were sampled across three training stages: basic sciences (n = 104), clinical clerkship (n = 101), and internship (n = 200). Trajectories were classified into seven distinct decision-making profiles using deductive framework analysis. Associations with training stage and institutional leprosy exposure were tested using Pearson’s chi-square test, the Cochran–Armitage test for trend, and Monte Carlo simulation. Results Only 12.8% (52/405) of participants navigated the complete target management pathway (pathway 9), with selection rates remaining low across all training stages ( X ²= 0.65, p = 0.72; linear trend p = 0.51). The distribution of clinical reasoning profiles varied across the curriculum, although this overall association was exploratory ( X ² = 20.43, p = 0.059). Pre-clinical students showed an excess of cognitive-bias errors (standardised residual + 2.48), whereas interns showed a deficit in these novice errors (-3.03) alongside an excess of early diagnoses followed by inappropriate management (+ 3.07). Most students (71.4%, 289/405) completed the case without performing the physical examination, having requested high-complexity imaging first. Conclusions Linear curricular progression did not inherently foster adaptive clinical problem-solving on this simulated case. Advanced training produced a pattern of confident incompetence: senior students avoided novice heuristic biases yet frequently diverged into non-guideline management regimens, a pattern consistent with technological over-reliance and automated decision-making. Developing diagnostic autonomy requires shifting from passive content accumulation toward cultivating cognitive flexibility and metacognitive awareness.