SRS107 - Predictive accuracy of the american college of surgeons National Surgical Quality Improvement Program (NSQIP) surgical risk calculator for patients undergoing infra-inguinal bypass for chronic limb-threatening ischaemia

Abstract Background Patients with chronic limb threatening ischaemia (CLTI) are often at high peri-operative risk from associated co-morbidities yet require expedient revascularisation. Several risk-prediction models exist, with the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) Surgical Risk Calculator widely used. However, it lacks external validation specific to CLTI, questioning its reliability for this population. Methods A retrospective review consecutive patients undergoing infra-inguinal bypass for CLTI between December 2023 and September 2024 at a single institution. NSQIP calculated risk scores were evaluated using Receiver Operating Characteristic curves, Brier scores, and correlation coefficients to assess predictive accuracy and discriminative ability. Results A total of 120 infra-inguinal bypass procedures were analysed. The NSQIP calculator demonstrated strong discriminative ability for predicting cardiac complications (Area Under the Curve [AUC] = 0.92; Brier score = 0.02) and moderate performance for sepsis prediction (AUC = 0.73; Brier score = 0.05). However, the tool showed poor predictive accuracy for 11 out of the 13 operation-specific outcomes assessed. For predicted length of hospital stay, the mean absolute error was 10.4 days, and the Pearson correlation coefficient was low (R = 0.213), indicating a weak association between predicted and actual durations. Discussion While the NSQIP Surgical Risk Calculator demonstrated value in predicting cardiac and sepsis complications in CLTI patients, its limited performance across most outcome domains raises concerns about its broader applicability. Caution is advised when using this tool for informed consent and shared decision-making in this patient population. Alternative or supplementary risk assessment tools should be considered.