Neck circumference for prediction of difficult laryngoscopies: a prospective cohort

Abstract

Background: Airway prediction still remains challenging and further approaches are worth investigating. We evaluated the predictive values of the neck circumference (NC) for prediction of difficult laryngoscopies (DL) based on diverse statistic models. Methods: A prospective cohort was conducted with patients undergoing general anesthesia for surgical procedures. We preoperatively collected data on sex, age, weight, height, ASA physical status, and NC. The main outcome was DL defined as Cormack and Lehane's classes 3 or 4. Uni and multivariable analyses were performed to evaluate association between variables and build predictive models. Results: From a total of 130 patients, 9 (6.9%) presented DL. Sex (p=0.008) and NC (p=0.008) were associated with DL. Age (p=0.306), weight (p=0.090), height (p=0.156), and ASA (p=0.624) did not present significant difference between the 2 groups. NC was weakly correlated with Cormack and Lehane's classification modified by Cook (r=0.271; p=0.001). Six predictive models were built including only NC. The ROC curves for the predictive models are presented in Figure 1 and their AUC were as follows: Logistic Regression model - LR (81.86%; 95% CI: 70.97-92.76%); Linear Discriminant Analysis - LDA (81.86%; 95% CI: 70.97-92.76%); Classification and Regression Trees - CART (50%; 95% CI: 50-50%); k-Nearest Neighbors - kNN (88.93%; 95% CI: 81.71-96.16%); Random Forest - RF (93.2%; 95% CI: 87.11-99.3%); Generalized Boosted Model - GBM (87.14%; 95% CI: 77.66-96.63%). The optimal cut-off point for NC according to Youden index was 39.1 cm. For such threshold, the predictive values were sensitivity of 88.88%, specificity of 74.38%, PPV of 20.51, NPV of 98.90%, overall accuracy of 75.38%, and balanced accuracy of 81.63%. Positive likelihood ratio and negative likelihood ratio were 3.46 and 0.14 respectively. Conclusions: NC is associated with DL and demonstrated helpful predictive performance. Therefore, it might be useful into multivariable models. Notwithstanding, the ability to segregate easy and difficult laryngoscopies remains cripple.