Algorithm to predict postoperative complications in oropharyngeal and oral cavity carcinoma.

BACKGROUND Preoperative data in patients with oral cavity/oropharyngeal cancer may predict postoperative complications that may modify therapeutic choices and improve patient care. METHOD We reviewed 320 consecutive patients with oral cavity/oropharyngeal cancer, operated on 2003 through 2006 at the European Institute of Oncology. By multivariate analysis of preoperative patient and tumor characteristics, we developed an algorithm to predict postoperative complications. We tested the algorithm on a new series of 307 patients operated on 2007 through 2010. RESULTS The final algorithm used to produce a nomogram was comprised of: alcohol consumption (p = .01), site of primary (p = .03), interaction of clinical T classification to sex (p = .007), and type of neck dissection (p < .0001). The algorithm had good ability to predict complications (concordance index [c-index] 0.74) in the new series. CONCLUSION The nomogram accurately predicts presurgical risk of postoperative local/systemic complications in patients with oral cavity/oropharyngeal cancer and can be used to adapt therapy to patient characteristics, optimize ward admissions, and improve care.