The low data fidelity of human-based informatics in surgical instrument waste reduction: is it time for objective computer vision integration?

Unused but sterilized surgical instruments generate substantial financial and environmental waste. Tray optimization commonly relies on human-based education, feedback, and manual documentation, yet data fidelity remains uncertain. We conducted a retrospective pre–post observational analysis of a human-based tray optimization initiative involving six surgeons across multiple procedures. Utilization was defined as the ratio of instruments used to those opened per case. Adherence, data capture, and variability across surgeons and procedures were assessed. Mean utilization increased from 0.335 to 0.548 (+ 63.6%). However, changes varied widely across surgeons (+ 0.103 to + 0.418), and one procedure showed decreased utilization. Critically, data capture was low and declined from 24.7% to 15.3%, indicating inconsistent participation and limited data fidelity. Despite improved average utilization, poor data fidelity, participation imbalance, and outcome variability constrain interpretation of human-based optimization efforts. Objective, automated approaches such as computer vision may support more reliable and scalable waste reduction.