Predictive performance of anthropometric measurements and bedside airway tests for difficult intubation and difficult mask ventilation: a prospective observational study

Difficult tracheal intubation and difficult mask ventilation remain leading contributors to anaesthesia-related morbidity, yet reliable preoperative prediction continues to be challenging. This study evaluated the predictive value of multiple preoperative anthropometric measurements and bedside clinical airway tests for difficult intubation and difficult mask ventilation. In this prospective observational study, 1,044 adult patients undergoing elective surgery under general anaesthesia with endotracheal intubation were enrolled. Fourteen preoperative anthropometric and bedside airway parameters were recorded. Difficult intubation was defined as two or more intubation attempts and/or clinically indicated use of a non-stylet airway-facilitating or alternative device, including use during the first attempt. Because device selection could be influenced by the preoperative assessment, Cormack-Lehane Grade ≥ 3 was evaluated as a complementary outcome less directly dependent on this management decision. Discrimination was assessed by receiver operating characteristic analysis, independent predictors by multivariable logistic regression, and both multivariable models underwent internal validation with 2,000 bootstrap resamples. Difficult intubation occurred in 194 patients (18.6%). STOP-BANG (AUC 0.611), Mallampati classification (AUC 0.608), and neck circumference (AUC 0.597) were the strongest single predictors. Mallampati > 2, STOP-BANG > 2, and bimandibular distance were independently associated with the composite outcome; their combination yielded an apparent AUC of 0.645 and an optimism-corrected AUC of 0.639. The complementary Cormack-Lehane Grade ≥ 3 model yielded an apparent AUC of 0.660 and an optimism-corrected AUC of 0.655. Neck circumference (AUC 0.741) and STOP-BANG (AUC 0.732) were the strongest predictors of difficult mask ventilation. Ratio-based thyromental indices showed limited discrimination, and agreement between Mallampati and Cormack-Lehane grading was weak (κ = 0.182). Neither individual bedside tests nor the combined model provided sufficient discrimination for standalone prediction of difficult intubation. Bootstrap validation showed limited optimism but confirmed modest model performance. The complementary Cormack-Lehane analysis, which was less directly influenced by the outcome-defining decision to use an additional airway device, led to a similar overall interpretation. Difficult intubation and difficult mask ventilation showed distinct predictor profiles and should therefore be considered separately within structured preoperative airway assessment. ClinicalTrials.gov, NCT07077317. Registered 11 July 2025 (prospectively registered).