Discrepancy Between Gaze Fixation and Risk Perception in Fall Risk Scenarios Among Healthcare Professionals and Students

Background Medical errors and near misses remain critical patient safety concerns. While eye-gaze analysis has been used to study risk perception among healthcare professionals, the relationship between visual fixation time and actual risk judgment under time constraints remains unclear, particularly regarding experience-based differences in hazard recognition. Understanding healthcare workers' risk perception strategies is essential for effective training in preventing medical near-misses and adverse events in healthcare settings. This study addresses two crucial research questions that delve into the characteristics of risk perception when observing hazardous situations. It explores whether gaze fixation time on risks differs with experience and whether the tendency to overlook fall risks (hazards) despite observing them decreases with expertise. Methods The study involved 23 healthcare workers from three hospitals with convalescent rehabilitation units and 25 fourth-year students from rehabilitation training schools in Japan. The participants wore eye trackers and completed the Time Pressure-Kiken Yochi Training Effectiveness Measurement System (TP-KYT), which assesses hazard prediction skills. Healthcare professionals were required to have: (i) continuous clinical practice, (ii) TP-KYT scores ≥ 138 points (competency threshold), and (iii) analyzable eye-tracking data. Students were included if they were: (i) fourth-year rehabilitation students, (ii) able to continue academic activities, and (iii) had analyzable eye-tracking data. No TP-KYT score criterion was applied to students. We compared TP-KYT scores with gaze fixation duration, total amplitude of saccades, and fixation and non-cognition. We tested TP-KYT scores, gaze fixation duration, and total saccade amplitude using the Mann-Whitney U test. We examined fixation and non-cognition using risk ratios. Results Healthcare workers had higher TP-KYT scores than students (p < 0.01). Students had a shorter total fixation duration on hazards in one item of Scene 4 (p = 0.02), but there was no difference in total amplitude of saccades between groups in any item. Students demonstrated a significantly higher likelihood of not perceiving a risk despite fixating on it (Fixation and non-Cognition, FnC), particularly for environmental and contextual hazards (e.g., Scene 3 risk ratio for FnC: 2.68 (95%CI 1.46-4.91) for students vs. healthcare workers). Conclusion A key characteristic of student risk perception is an inability to contextualize relationships between objects and environmental factors. Risk education should therefore focus on teaching recognition methods that incorporate contextual understanding. These findings can inform the development of training strategies aimed at improving healthcare safety by preventing adverse events.