Perioperative Risk Stratification Model for Readmission after Panniculectomy.

BACKGROUND: Readmission is an important metric for surgical quality of care. This study aimed to develop a validated risk model that reliably predicts readmission after panniculectomy using the American College of Surgeons National Surgical Quality Improvement Program database. METHODS: The American College of Surgeons National Surgical Quality Improvement Program database was queried to identify all patients who had undergone panniculectomy from 2005 to 2018. The outcome of interest was 30-day readmission. The cohort was divided randomly into 70 percent development and 30 percent validation groups. Independent predictors of 30-day readmission were identified using multivariable logistic regression on the development group. The predictors were weighted according to beta coefficients to generate an integer-based clinical risk score predictive of readmission, which was validated against the validation group. RESULTS: For the model selection, 22 variables were identified based on criteria of p < 0.05 percent and complete data availability. Variables included in the development model included inpatient surgery, hypertension, obesity, functional dependence, chronic obstructive pulmonary disease, wound class greater than or equal to 3, American Society of Anesthesiologists class greater than 3, and liposuction. Receiver operating characteristic curve analysis of the validation group rendered an area under the curve of 0.710, which demonstrates the accuracy of this prediction model. The predicted incidence within each risk stratum was statistically similar to the observed incidence in the validation group ( p < 0.01), further highlighting the accuracy of the model. CONCLUSIONS: The authors present a validated risk stratification model for readmission following panniculectomy. Prospective studies are needed to determine whether the implementation of the authors' clinical risk score optimizes safety and reduces readmission rates. CLINICAL QUESTION/LEVEL OF EVIDENCE: Risk, III.