Accuracy of American College of Surgeons National Surgical Quality Improvement Program Universal Surgical Risk Calculator in Predicting Complications Following Robot-Assisted Radical Cystectomy at a National Comprehensive Cancer Center.

Introduction: There is paucity of literature about the validation of the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP® ) surgical risk calculator for prediction of outcomes after robot-assisted radical cystectomy (RARC). We sought to evaluate the accuracy of the ACS NSQIP surgical risk calculator in the patients who underwent RARC at our institute. Methods: We retrospectively reviewed our prospectively maintained database for patients who underwent RARC between 2005 and 2017. Accuracy of the ACS NSQIP surgical risk calculator was assessed, by comparing the rate of actual complication events after surgery with the receiver operating characteristics curve analysis by calculating the fractional area under the curve (AUC) and the Brier score (BS). We utilized the code number 51595 and 51596 in the ACS NSQIP calculator for the patients undergoing radical cystectomy and reconstructed with the ileal conduit and neobladder, respectively. Results: A total of 462 patients were included in this study: 99 (22%) had diabetes, 302 (66%) had hypertension requiring medication, and 241 (52%) were classified as high American Society of Anesthesiologists (≥3) class. The actual observed rates of any complication and serious complications were 48% and 11%, vs 29% and 25% predicted by the ACS NSQIP, respectively. The actual mean length of hospital stay (10.6 ± 7.8 days) was longer compared with the predicted length (8.5 ± 1.6 days). AUC values were low and the BSs were high for any complication (AUC: 0.50 and BS: 0.29), serious complication (AUC: 0.53 and BS: 0.12), urinary tract infection (AUC: 0.61 and BS: 0.14), renal insufficiency (AUC: 0.64 and BS: 0.08), return to operation room (AUC: 0.58 and BS: 0.07), and early readmission (AUC: 0.55 and BS: 0.11, respectively). Conclusions: The ACS NSQIP calculator demonstrated low accuracy in predicting postoperative outcomes after RARC. These findings highlight the need for development of procedure- and technique-specific RARC calculators.