Development of an HFACS-based risk assessment indicator system for unplanned endotracheal extubation in adult ICU patients: a Delphi and AHP study

Objective To construct a risk-assessment indicator system for unplanned endotracheal extubation (UEE) among adult ICU patients based on the Human Factors Analysis and Classification System (HFACS), and provide a structured multi-level framework for unit-level safety auditing and root-cause analysis of UEE events. Methods Initial indicators were generated via systematic literature review and retrospective analysis of local UEE adverse-event cases. A two-round Delphi expert consultation was performed to screen and revise indicators. The Analytic Hierarchy Process (AHP) was adopted to calculate indicator weights at each hierarchical level. Content validity was evaluated using the content validity index (CVI). Results The valid response rates for the two Delphi rounds were 100.0% (20/20) and 95.0% (19/20), respectively. The expert authority coefficients (Cr) were 0.893 and 0.915, indicating high expert authority. Kendall's coefficients of concordance ( W ) were 0.242 and 0.254 ( P < 0.001), showing statistically significant yet low expert agreement. Among the four primary dimensions, the combined weights were 0.448 for unsafe behavior of clinicians, 0.257 for inadequate supervision, 0.151 for preconditions for unsafe acts, and 0.144 for organizational influence. The final risk-assessment indicator system comprised 4 primary, 13 secondary, and 63 tertiary indicators with acceptable overall content validity. Conclusion Using HFACS, two-round Delphi consultation, and AHP, this study developed a consensus-driven preliminary indicator system for UEE risk-related safety management in adult ICUs. This framework is intended for system-oriented safety audits, training priority setting, and root-cause analysis rather than bedside individual-patient risk prediction. Further prospective clinical validation is required to verify its predictive validity, reliability and practical feasibility before widespread clinical implementation.