PC3. A Nationally Validated Risk Assessment Score for Predicting Unplanned Reoperations in Hand Surgery

Abstract

PURPOSE: To present a validated model that reliably predicts unplanned reoperation after upper extremity/hand surgery. METHODS: A total of 83,409 hand surgery patients were identified using the 2012-2019 ACS-NSQIP databases. Thirty-day unplanned reoperation was defined as unexpected reoperation for a postoperative occurrence related to the principal hand surgery. Independent predictors of 30-day unplanned reoperation were identified using multivariable logistic regression with backward variable selection on the testing sample. Subsequently, the predictors were weighted according to β-coefficients to generate an integer-based Novel Risk Score (NRS) predictive of reoperation. The NRS was then validated 1000 bootstrapped replications of the original sample, and validated on NSQIP patients who had undergone hand surgery in 2020 (n = 15,247). The NRS was compared to two other widely used risk scores with receiver operating characteristics (ROC) analysis. RESULTS: The rate of 30-day unplanned reoperation in the 2020 cohort was 1.1%. Independent risk factors included male gender, inpatient status, smoking, dialysis dependence, transfusion within 72 hours of surgery, wound classification, ASA Class, diabetes mellitus, CHF, sepsis or septic shock, emergent case, and long initial operative time (all P < 0.05). ROC analysis of the testing (2020) cohort rendered an area under the curve of 0.730, which demonstrates the accuracy of this prediction model. The mFI-5 and mCCI rendered AUCs of 0.580 and 0.585, respectively. CONCLUSION: We present a validated risk stratification tool for unplanned reoperations following hand surgery. Future studies should determine if implementation of this NRS optimizes safety and reduces reoperation rates in hand surgery patients.