When a Good Flap Turns Bad: A Temporal Predictive Model for Free Flap Complications.

Abstract

Microsurgical cases are complex plastic surgery procedures with a significant risk of acute postoperative complications. In this study, we use a large-scale database to investigate the temporal progression of complications after microsurgical procedures and the risk imparted by acute postoperative complications on subsequent reconstructive outcomes. Microsurgery cases were extracted from the National Surgical Quality Improvement Program database by Current Procedural Terminology codes. Postoperative complications were collected for 30 days after surgery and stratified into four temporal periods (postoperative days [PODs] 0-6, 7-13, 14-20, 21-30). Postoperative complication occurrences were incorporated into a weighted multivariate logistic regression model to identify significant predictors of adverse outcomes (p < 0.05). Separately, a regression model was calculated for the time between index operation and reoperation and additional complications. The final cohort comprised 19,517 patients, 6,140 (31.5%) of which experienced at least one complication in the first 30 days after surgery. The occurrence of prior complications in the postoperative period was a significant predictor of future adverse outcomes following the initial week after surgery (p < 0.001). Upon predictive analysis, overall model performance was highest in PODs 7 to 13 (71.1% accuracy and the area under a receiver operating characteristic curve 0.684); 2,578 (13.2%) patients underwent at least one reoperation within the first 2 weeks after surgery. The indication for reoperation (p < 0.001) and number of days since surgery (p = 0.0038) were significant predictors of future complications after reoperation. Prior occurrence of complications in an earlier postoperative week, as well as timing and nature of reoperation, were shown to be significant predictors of future complications.