Predictive Performance of Three Multivariate Difficult Tracheal Intubation Models: A Double-Blind, Case-Controlled Study

Abstract

We performed a case-controlled, double-blind study to examine the performance of three multivariate clinical models (Wilson, Arné, and Naguib models) in the prediction of unanticipated difficult intubation. The study group consisted of 97 patients in whom an unanticipated difficult intubation had occurred. For each difficult intubation patient, a matched control patient was selected in whom tracheal intubation had been easily accomplished. Postoperatively, a blinded investigator evaluated both patients. The clinical assessment included the patient's weight, height, age, Mallampati score, interincisor gap, thyromental distance, thyrosternal distance, neck circumference, Wilson risk sum score, history of previous difficult intubation, and diseases associated with difficult laryngoscopy or intubation. The Naguib model was significantly more sensitive (81.4%; P < 0.0001) than the Arné (54.6%) or Wilson (40.2%) models. Both the Naguib (76.8%) and Arné (74.7%) model classified more intubations correctly (P = 0.01) than the Wilson model (66.5%). The specificity of Arné, Wilson, and Naguib model was 94.9%, 92.8%, and 72.2%, respectively (P < 0.0001). The corresponding area under the receiver operating characteristic curve was 0.87, 0.79, and 0.82, respectively. Our new model for prediction of difficult intubation was developed using logistic regression and includes thyromental distance, Mallampati score, interincisor gap, and height. This model is 82.5% sensitive and 85.6% specific with an area under the receiver operating characteristic curve of 0.90.