The Veteran Affair Surgical Quality Improvement Program Calculator is a Poor Predictor of Morbidity and Mortality in Octogenarian and Nonagenarian Veterans Undergoing Major Lower Extremity Amputations

Background: Morbidity and mortality for major (above the ankle) lower extremity amputation (LEA) is high in veteran patients and age is a predictor of mortality. The Veteran Affair Surgical Quality Improvement Program (VASQIP) risk assessment tool has been validated for several operations but not for elderly patients undergoing LEA. The present study interrogated the accuracy for the VASQIP calculator for a medium/high-risk operation in a high-risk veteran population (octogenarians and nonagenarians). Methods: Variables required from input for the VASQIP calculator were retrospectively obtained for 57 octogenarians and 11 nonagenarians submitting to LEA at our institution from 2009 to 2021. The six-outcome variables provided by the VASQIP calculator (30-day mortality, 180-day mortality, 30-day morbidity, 30-day surgical site infection risk, probability of intensive care unit stay, and probability of hospital stay) were compared to observed morbidity and mortality. The accuracy of the calculator was assessed by area under the receiver operating characteristic curve and reported by the area under the curve (AUC) as previously described. Results: In the 68 patients included in this analysis, the time to death from the last index operation was 422.0 ± 604.9 days for octogenarians and 65.6 ± 89.3 days for nonagenarians. Predicted versus observed 30-day mortality for octogenarians and nonagenarians was 8.46 vs. 24.56 [AUC = 0.739; 95% confidence interval (0.581 to 0.898)] and 24.46 vs. 45.45 [AUC = 0.600 (0.171 to 1.000)], respectively. Predicted versus observed 180-day mortality for the same cohorts was 25.22 vs. 47.37 [AUC = 0.578 (0.427 to 0.728)] and 45.34 vs. 90.91 [AUC = 0.100 (0.000 to 0.286)], respectively. Thirty-day morbidity, 30-day surgical site infection, probability of intensive care unit, and probability of in-hospital stay produced an AUC less than 0.600 for all these outcomes. Conclusions: The VASQIP risk calculator is a poor predictor of short-term outcomes in octogenarians and nonagenarians undergoing major LEA. Most octogenarian and nonagenarian veterans died within 1 year, and the mean survival for nonagenarians was less than 3 months after LEA. The decision for major LEA in octogenarian and nonagenarian veterans warrants an informed discussion with the patient and family.