Background: The American College of Surgeons National Surgical Quality Improvement Program's (ACS-NSQIP) calculator has been endorsed to counsel patients regarding complications. The aim of this study was to assess its ability to predict outcomes after hepatectomy.
Methods: Outcomes generated by the ACS-NSQIP were recorded in a consecutive cohort of patients. By using established classifications of complications, post-hepatectomy insufficiency and bile leak, the calculator was tested by the comparison of expected versus observed rates of events. The performance of the calculator was tested by using c-statistic and Brier score.
Results: 950 patients who underwent hepatectomy between January 2014 and June 2019 were included. Predicted rates were significantly lower than actual rates: the mean ACS-NSQIP morbidity was 17.97% ± 8.4 vs. actual 37.01% ± 0.56 (P < 0.001); the mean ACS-NSQIP mortality was 0.91% ± 1.48 vs. actual 1.76% ± 0.11 (P < 0.001). Predicted length of stay (LOS) was significantly shorter: mean ACS-NSQIP was 5.81 ± 1.66 days vs. actual 10.91 ± 4.6 days (P < 0.001). Post-hepatectomy liver insufficiency and bile leak were recorded in 6.8% and 11.9% of patients, respectively. These events were not expressed by the calculator. C-statistic and Brier scores showed low performance of the calculator.
Conclusion: The calculator underestimates the risks of complications, mortality and LOS after hepatectomy. Refinements of the ACS-NSQIP model that account for organ-specific risks should be considered.
Keywords: ACS-NSQIP calculator; Bile leak; Complications; Hepatectomy; Liver failure; Mortality.
Copyright © 2020. Published by Elsevier Ltd.