Discrepancies Between Self-Reported and High-Precision UWB-Measured Close Contacts: Evidence from a Japanese Junior High School

Background Accurate identification of close contacts is essential for effective infection control. However, school-based contact tracing typically relies on retrospective self-reports, which are prone to recall bias. Ultra-wideband (UWB) indoor positioning offers centimeter-level spatial resolution but has rarely been applied in educational settings. This study is the first to benchmark self-reported 'close contacts' against centimeter-level ultra-wideband (UWB) positioning in a real classroom over multiple consecutive days, extending beyond prior RFID- or Bluetooth-based school studies. Methods We conducted a five-day observational study in a junior high school classroom in Japan (36 participants; 33 with complete questionnaire and UWB data). Close contacts were identified via self-reported questionnaires and UWB positioning under multiple distance-time thresholds, with 1.0 meter for 15 minutes as the official baseline. Discrepancies were quantified using participant-level consistency rates, and sensitivity analyses evaluated alternative thresholds. Findings Discrepancies between the two methods were consistently large under the 1.0 m/15 min definition. The median discrepancy was -0.16, indicating systematic overreporting in questionnaires. The smallest discrepancies occurred at 1.5 m for ≥10 min, where UWB and recall most closely aligned. Notably, UWB continued to identify new unique contacts throughout the week, whereas questionnaire-based reporting plateaued after three days. Higher confidence in self-reports was paradoxically associated with greater overreporting. Teacher-student reciprocity was absent, despite the teacher reporting contact with all students. Interpretation In a fixed-seating classroom, self-reported and UWB-measured close contacts diverged substantially. By leveraging centimeter-precision UWB tracking and threshold calibration, our study provides the first empirical evidence that current operational definitions reflect memory alignment more than actual spatial interactions. Incorporating UWB positioning and empirically optimized criteria may improve the accuracy and feasibility of contact tracing in schools, while minimizing unnecessary quarantines and educational disruption.