Primary outcomes for perioperative adverse events included the occurrence of minor adverse events, severe adverse events, infectious adverse events, any adverse event, extended length of hospital stay, and discharge to higher-level care. Each of the indices was assessed for discriminative ability in predicting perioperative adverse outcomes.
Using a population of 5,166 patients who met the inclusion criteria, the most predictive comorbidity index was ASA (AUC 0.54) and demographic factor was BMI (AUC 0.53). However, all predictive indices performed poorly to predict “any adverse event”(AUC 0.53-0.54). ASA demonstrated the highest AUC of the indices for predicting serious (0.55), minor (0.53), infectious events (0.54), and discharge to higher level care (0.58) – though none were very impressive. A combination of the most predictive demographic factor (BMI) and comorbidity index (ASA) resulted in incremental improvements in discriminative ability over the individual components for all outcome variables, but still not very high.
Based on this, the authors note that ASA and BMI, both of which are simply calculated and obtained, have overall better discriminative abilities for perioperative adverse outcomes than the other indices (more difficult to calculate). However, we still have room to improve as the AUCs were still not very impressive (<0.70).
Presented by: Xiaosong Meng
Co-Authors: Audrey Renson, James Wysock, William C. Huang, Samir S. Taneja, Marc Bjurlin, NYU Langone Health, Brooklyn
Written by: Thenappan Chandrasekar, MD, Clinical Fellow, University of Toronto, twitter: @tchandra_uromd at the 18th Annual Meeting of the Society of Urologic Oncology, November 20-December 1, 2017 – Washington, DC