[Evaluation of a protocol for predicting difficult airway in routine practice: interobserver agreement]

To evaluate a protocol for routine assessment of potential risk factors for difficult airway intubation in an anesthesia department, by measuring interobserver agreement and the behavior of the factors in a predictive model.

A cross-sectional study group of 320 consecutive patients undergoing major surgery requiring orotracheal intubation was assessed for possible difficult airway. We calculated interrater agreement for recording of the Mallampati score, thyromental distance less than 6 cm, thick neck, kyphosis, small mouth, macroglossia, and dental prosthesis during the preanesthesia examination (by an anesthetist) and on the day of the operation (by an anesthetist and a resident). We constructed a model to predict difficult intubation (requiring 3 or more attempts).

The kappa indices of agreement between the anesthetists at the preoperative examination and in the operating room or the resident were all less than 0.6. Factors like thyromental distance, small mouth, and kyphosis had kappa indices less than 0.21. The kappa index between the resident and the anesthetist in the operating room was over 0.55. The only factor that had a different level of agreement was the presence or not of a dental prosthesis. None of the studied individual factors, nor these factors in association with the Mallampati score, achieved significance in a bivariate regression model to predict difficult intubation.

There is poor interobserver agreement on factors for predicting difficult airway in comparisons between preoperative and operating room assessment by an anesthetist or a resident. The individual predictive factors and their association with the Mallampati score did not prove useful for predicting difficult intubation.