Evaluating Unplanned Returns to the Operating Room in Head and Neck Free Flap Patients

Abstract

Background Head and neck oncologic surgery with reconstruction represents one of the most complex operations in otolaryngology. Unplanned return to the operating room represents an objective measure of postoperative complications. The purpose of this study was to identify reasons and risk factors for unplanned return to the operating room in patients undergoing head and neck surgery with reconstruction. Methods This retrospective cohort study of 467 patients undergoing head and neck surgery with free flap reconstruction used a previously-developed Head and Neck-Reconstructive Surgery-specific National Surgical Quality Improvement Program. Disease and site-specific preoperative, intraoperative, and postoperative data were gathered. Comparisons between those with and without an unexpected return to the operating room were completed with univariate and multiple logistic regression models. Results The rate of unexpected return to the operating room was 18.8% (88 patients). Most common reasons for URTOR were flap compromise (24 patients, 5.1%), postoperative infection (21 patients, 4.5%), and hematoma (20 patients, 4.3%). Two risk factors were identified by multivariate analysis: coagulopathy (OR adjusted \u2009=\u20092.83, 95% CI\u2009=\u20091.24–6.19, P \u2009=\u20090.010), and use of alcohol (OR adjusted \u2009=\u20091.9, 95% CI\u2009=\u20091.14–3.33, P \u2009=\u20090.025). Conclusions Preexisting coagulopathy and increased alcohol consumption were associated with increased risk of unexpected return to the operating room. These findings can aid physicians in preoperative patient counseling and medical optimization and can inform more precise risk stratification of patients undergoing head and neck surgery with reconstruction. Strategies to prevent and mitigate unexpected returns to the operating room will improve patient outcomes, decrease resource utilization, and facilitate successful integration into alternative payment models.