Prevalence and bedside predictors of difficult direct laryngoscopy among 3,080 adult elective surgical patients at a Cambodian tertiary center: a retrospective cohort study.

backgroundFailure to anticipate difficult direct laryngoscopy (DDL) leads to catastrophic airway events. While prediction tools have been developed primarily in European and North American cohorts, evidence from Southeast Asia—particularly in settings with high volumes of head-and-neck pathology—is limited. We investigated DDL prevalence, identified independent bedside predictors, and evaluated a simplified cumulative risk score in a Cambodian tertiary center.

methodsWe conducted a retrospective cohort study of 3,080 consecutive adults undergoing elective surgery with planned Macintosh direct laryngoscopy (January–June 2023) at Preah Ang Duong Hospital, Phnom Penh. DDL was defined as Cormack–Lehane grade III/IV or three or more laryngoscopic attempts. Seven bedside predictors were analyzed using multivariable logistic regression. A simplified composite risk score (range 0–3) was derived from the strongest independent predictors and evaluated using receiver operating characteristic (ROC) analysis. To assess confounding by surgical case-mix, an additional model adjusted for head-and-neck surgical category. Model calibration was assessed with the Hosmer–Lemeshow goodness-of-fit test. Data were analyzed using SPSS V30.

resultsDDL prevalence was 9.03% (278/3,080; 95% CI: 8.01–10.04%), rising to 13.53% in maxillofacial and 11.34% in ENT procedures. The mean BMI was 26.5 ± 6.4 kg/m², with 35.0% classified as obese using the WHO Asian-specific threshold (≥ 27.5 kg/m²). In multivariable analysis, six independent predictors emerged: Mallampati III/IV (adjusted odds ratio [AOR] 4.15; 95% CI: 3.02–5.70), BMI ≥ 27.5 kg/m² (AOR 2.92; 95% CI: 2.25–3.79), limited neck mobility (AOR 2.13; 95% CI: 1.64–2.75), thyromental distance ≤ 6.5 cm (AOR 1.95; 95% CI: 1.46–2.59), neck circumference ≥ 40 cm (AOR 1.41; 95% CI: 1.09–1.84), and inter-incisor gap ≤ 3 cm (AOR 1.41; 95% CI: 1.06–1.88); all p < 0.02. The upper lip bite test was not independently predictive (p = 0.91). All six predictors retained significance after adjustment for head-and-neck surgical category (H&N AOR 2.42; 95% CI: 1.83–3.21). A simplified composite score (Mallampati + BMI + TMD) achieved an AUC of 0.72 (95% CI: 0.69–0.75; bootstrap-corrected 0.72), with 82.7% sensitivity, 54.0% specificity, and 96.9% negative predictive value at a cutoff of 2 or greater. The score demonstrated a clear dose–response relationship (DDL prevalence: 2.2% at score 0, 3.3% at score 1, 11.9% at score 2, 24.7% at score 3). The model showed acceptable calibration (Hosmer–Lemeshow p = 0.42). Pre-specified sensitivity analyses confirmed score stability in non-head-and-neck cases (AUC 0.75; n = 1,488) and under a Cormack–Lehane III/IV–only definition (AUC 0.72). Specialty-specific AUCs ranged from 0.68 to 0.82, supporting generalizability across surgical subgroups.

conclusionsDDL affects nearly 1 in 11 elective surgical patients in this high-acuity cohort, driven largely by the case-mix of head-and-neck procedures. A simple, three-component bedside score offers a zero-cost tool to enhance preoperative risk stratification in resource-limited settings, with BMI emerging as a particularly strong predictor when assessed using population-specific thresholds. Prospective, multicenter validation is required before routine clinical implementation.