Optimizing classical risk scores to predict complications in head and neck surgery: a new approach

Purpose To validate tools to identify patients at risk for perioperative complications to implement prehabilitation programmes in head and neck surgery (H&N). Methods Retrospective cohort including 128 patients submitted to H&N, with postoperative Intermediate Care Unit admittance. The accuracy of the risk calculators ASA, P-POSSUM, ACS-NSQIP and ARISCAT to predict postoperative complications and mortality was assessed. A multivariable analysis was subsequently performed to create a new risk prediction model for serious postoperative complications in our institution. Results Our 30-day morbidity and mortality were 45.3% and 0.8%, respectively. The ACS-NSQIP failed to predict complications and had an acceptable discrimination ability for predicting death. The discrimination ability of ARISCAT for predicting respiratory complications was acceptable. ASA and P-POSSUM were poor predictors for mortality and morbidity. Our new prediction model included ACS-NSQIP and ARISCAT (area under the curve 0.750, 95% confidence intervals: 0.63–0.87). Conclusion Despite the insufficient value of these risk calculators when analysed individually, we designed a risk tool combining them which better predicts the risk of serious complications.