Individualized risk of surgical-site infection: an application of the breast reconstruction risk assessment score.

BACKGROUND: Risk factors for surgical-site infection following beast reconstruction have been thoroughly investigated at a population level. However, traditional population-based measures may not always capture the nuances of individual patients. The authors aimed to develop a validated breast reconstruction risk assessment calculator for surgical-site infection that informs risk at an individual level. METHODS: Mastectomies with immediate reconstruction (n = 16,069) from 2005 to 2011 were identified from the National Surgical Quality Improvement Program database. A multiple logistic regression model was created for postoperative surgical-site infection. Hosmer-Lemeshow, C statistic, and Brier score were computed to assess model performance. Bootstrap analysis validated the model. RESULTS: A robust, validated risk model for surgical-site infection was developed using 11 covariates. The model Hosmer-Lemeshow p value was 0.371, the Brier score was 0.0357, and the C statistic was 0.682 (optimism-corrected C statistic, 0.678). The distribution of individual risks demonstrated a positive skew. Population-derived risk underestimated or overestimated individual risk by at least 1.5-fold in nearly one-fifth of all patients. CONCLUSIONS: The breast reconstruction risk assessment score risk calculator for surgical-site infection mitigates the potentially inaccurate interpolation of population-based risk to individual patients. The authors concomitantly developed an online interface-accessible by patients and surgeons alike-to quantify a patient's risk for surgical-site infection, better informing evidence-based decisions and managing patient expectations. CLINICAL QUESTION/LEVEL OF EVIDENCE: Risk, III.