Factors predictive of 90-day mortality after surgical resection for oral cavity cancer: Development of a recursive partitioning analysis for risk stratification

Background: Factors that influence postoperative mortality (POM) have been identified, but a predictive model to guide clinicians treating oral cavity cancer (OCC) has not been well established. Methods: Patients with OCC undergoing upfront surgical resection were included. Primary outcome was 90-day POM (90dPOM). Results: 33 845 were identified using the National Cancer Database. Rate of 90dPOM was 3.2%. Predictors of higher 90dPOM include older age, higher comorbidity scores, nonprivate insurance, lower income, treatment in an academic facility, higher T- and N-classification, radical excision, and presence of positive margins. On RPA, two high-risk (90dPOM > 10%) patient subsets were identified: patients ≥80 years of age with T3-4 disease and patients <80 years, with any comorbidity and T3-4, N2-3 disease. Conclusions: We identified a subset of patients in this cohort who are at high risk for 90dPOM. These patients may warrant additional perioperative and postoperative monitoring in addition to better preoperative assessment and screening.