Existing literature has emphasized the utility of the five-item modified frailty index (mFI-5) in predicting postoperative outcomes after surgical procedures. However, in breast reconstruction, a primarily elective post-oncologic procedure for otherwise relatively healthy patients, several components of the index may be sparse and not strongly contribute to predictive value. Breast reconstruction cases were identified in the 2012-2022 National Surgical Quality Improvement Program. Three metrics were compared in this cohort: 1) the mFI-5, 2) a simplified two-item index comprising diabetes and/or hypertension, and 3) American Society of Anesthesiologists (ASA) class. Each metric was incorporated into three multivariate logistic regression models for occurrence of at least one postoperative complication. Predictive performance among metrics was compared over fifty iterations (p<0.05). Performance was compared between autologous and implant-based modalities in a supplementary analysis. 134,983 breast reconstruction cases were identified in the final cohort. In multivariate regression, both MFI and the two-item index of diabetes and hypertension were significant predictors of postoperative complications (p<0.001). The mFI-5 (average AUC: 0.6106) and two-item index (average AUC: 0.6105) performed without significant difference (p=0.93). Discriminatory performance of ASA class (average AUC: 0.6115), was not significantly different from the other metrics (p>0.05). In the context of a population of mainly elective breast reconstruction procedures, the five-item modified frailty index does not outperform an index of diabetes and hypertension alone. Such findings may motivate the selection of other variables that may be more useful for prediction of postoperative outcomes.
Read full abstract