Predicting Complications in Head and Neck Surgery: Comparing Calculators to Surgeons.

Objectives: Surgical outcomes determine national ranking, reputation, and funding, and are often assessed with objective surgical risk calculators (SRCs). Surgeons' assessments are not considered. This study aims to determine if surgeons or SRCs are more accurate in predicting outcomes. Methods: This prospective cohort study identified a surgeon's assessment on a patient's risk preoperatively. The patient's risk was also calculated using the SRC. Predictions were compared to patient outcomes and to each other to assess whether surgeons or the SRC were more accurate. Results: Of the 101 patients included, 37 (36.6%) experienced a complication of any kind and 18 (17.8%) experienced a serious complication. Smoking resulted in a 2.49 times higher overall complication rate (P = .04). Laryngectomy patients experienced the highest rate of complications (P = .02) compared to those undergoing free flap reconstruction [odds ratio (OR) 0.9] or any other surgery (OR 0.26). Both surgeons and the American College of Surgeons (ACS) tool performed poorly on the prediction of the outcome of any complication, with a receiver operating characteristic (ROC) area under the curve (AUC) of 0.51 [95% confidence interval (CI): 0.39-0.62] and 0.58 (95% CI: 0.47-0.70), respectively, which was not statistically significant (P = .34). For the prediction of the outcome of serious complication, the AUC for surgeons and the ACS tool were 0.55 (95% CI: 0.41-0.69) and 0.60 (95% CI: 0.46-0.74), respectively, which was not statistically significant (P = .58). Conclusions: Neither validated risk calculators nor surgeons are accurate in predicting perioperative risk. The only risk factor that contributes to improving predictions for complications is preoperative smoking, although age and type of surgery are also significant predictors. Risk calculators may therefore not be appropriate metrics for assessing hospital performance. These findings can help guide preoperative counseling and may help in the development of more accurate predictive tools as the healthcare field continues to incorporate artificial intelligence into surgical planning.

Duke Scholars

Author Christopher Lindsell Biostatistics & Bioinformatics, Division of Biostatistics