240 Background: With increasing health care costs, risk-adjusted quality outcomes are essential. The American College of Surgeons – National Surgical Quality Improvement Program (ACS-NSQIP) is a robust perioperative risk-adjustment platform, but lacks capture of oncologic- and specialty-specific variables and has limited utility for risk adjustment in head and neck oncologic surgery. This study uses the specialty-specific Head and Neck-Reconstructive Surgery (HNSR) NSQIP to develop risk-adjusted models for patients undergoing head and neck oncologic surgery with reconstruction. Methods: Multiple logistic regression modelling using data from patients in the HNSR NSQIP between 8/2012-10/2016 identified predictors of postoperative morbidity. Final models were validated using a cohort of patients treated between 10/2016-12/2017. The concordance index (c-index) was used to evaluate the model performance. Results: The modelling cohort included 1095 patients and the validation cohort included 407. Models were created to predict probability of postoperative complications (presence of fistula, ventilator dependence > 48 hours, pneumonia, deep/superficial surgical site infection); presence of gastrostomy-jejunostomy (GJ), nasoenteric (NE), or tracheostomy tube at 30 days postoperatively; conversion from NE to GJ tube; unplanned return to the operating room; and length of stay > 7 days. Most discriminant models were those predicting presence of GJ tube (model c-index [MCI] = 0.91; validation c-index[VCI] = 0.93), NE tube (MCI= 0.83; VCI= 0.84), and conversion from NE to GJ tube (MCI= 0.86; VCI= 0.80). Prediction models were least discriminant for ventilator dependence (MCI= 0.63; VCI= 0.45), fistula (MCI= 0.58; VCI= 0.54), and unplanned return to the operating room (MCI= 0.59; VCI= 0.51). Conclusions: Reliable and discriminant risk prediction models were able to be created for a variety of perioperative complications incorporating oncologic- and specialty-specific variables in the HNSR NSQIP. These models help inform risk stratification strategies for patients undergoing head and neck reconstructive surgery and the development of a specialty-specific preoperative risk calculators.
Surgical technique refinements in head and neck oncologic surgery