Mao Xu1, Xiaoxi Li2, Jun Wang1, Xiangyang Guo3. 1. Department of Anesthesiology, Peking University Third Hospital, Beijing 100191, China. 2. Department of Anesthesiology, Beijing Cancer Hospital & Institute, Beijing 100142, China. 3. Department of Anesthesiology, Peking University Third Hospital, Beijing 100191, China. Email: puthmzk@163.com.
Abstract
BACKGROUND: Airway management is crucial in clinical anesthesia. Many complications associated with airway management result from unexpected difficult airway, but predicting a difficult airway is a major challenge. We investigated the efficacy of a new combined model including radiological indicators to predict difficult airway in patients undergoing surgery for cervical spondylosis, a population with a high incidence of difficult airway. METHODS: We randomly enrolled 303 patients scheduled for elective surgery for cervical spondylosis at Peking University Third Hospital between August 2012 and March 2013. Preoperatively, patients were evaluated for difficult airway according to a clinical index and parameters on lateral cervical radiographs and magnetic resonance images. Difficult airway was defined as Cormack-Lehane grades III-IV. Logistic regression was used to identify a combined (clinical and radiological) model for difficult airway. A receiver operating characteristic (ROC) curve was used to describe the effectiveness of prediction. RESULTS: We identified three clinical predictive factors using the ROC curve: mouth opening, sternomental distance, and neck mobility. We created a clinical model using three factors: gender, age, and mouth opening, with odds ratios (OR) of 0.370, 1.034, and 0.358, respectively. Using the clinical and radiological parameters, we formulated a combined model with five risk factors: gender, mouth opening, atlanto-occipital gap, the angle from the second to sixth cervical vertebraes in the neutral position, and the angle difference of d (the angle between the laryngeal axis and the epiglottic axis) from the neutral position to extension (OR: 0.107, 0.355, 0.846, 1.057, and 0.952, respectively). The sensitivity and specificity of the combined model were 80.0% and 65.7%, respectively, and the ROC curve confirmed that the combined model was better than any single clinical predictor and the clinical model. CONCLUSION: The efficacy of the combined model including both clinical and radiological indicators was better than any single clinical predictor or the clinical model in patients undergoing elective surgery for cervical spondylosis.
RCT Entities:
BACKGROUND: Airway management is crucial in clinical anesthesia. Many complications associated with airway management result from unexpected difficult airway, but predicting a difficult airway is a major challenge. We investigated the efficacy of a new combined model including radiological indicators to predict difficult airway in patients undergoing surgery for cervical spondylosis, a population with a high incidence of difficult airway. METHODS: We randomly enrolled 303 patients scheduled for elective surgery for cervical spondylosis at Peking University Third Hospital between August 2012 and March 2013. Preoperatively, patients were evaluated for difficult airway according to a clinical index and parameters on lateral cervical radiographs and magnetic resonance images. Difficult airway was defined as Cormack-Lehane grades III-IV. Logistic regression was used to identify a combined (clinical and radiological) model for difficult airway. A receiver operating characteristic (ROC) curve was used to describe the effectiveness of prediction. RESULTS: We identified three clinical predictive factors using the ROC curve: mouth opening, sternomental distance, and neck mobility. We created a clinical model using three factors: gender, age, and mouth opening, with odds ratios (OR) of 0.370, 1.034, and 0.358, respectively. Using the clinical and radiological parameters, we formulated a combined model with five risk factors: gender, mouth opening, atlanto-occipital gap, the angle from the second to sixth cervical vertebraes in the neutral position, and the angle difference of d (the angle between the laryngeal axis and the epiglottic axis) from the neutral position to extension (OR: 0.107, 0.355, 0.846, 1.057, and 0.952, respectively). The sensitivity and specificity of the combined model were 80.0% and 65.7%, respectively, and the ROC curve confirmed that the combined model was better than any single clinical predictor and the clinical model. CONCLUSION: The efficacy of the combined model including both clinical and radiological indicators was better than any single clinical predictor or the clinical model in patients undergoing elective surgery for cervical spondylosis.