A novel scoring system to predict postoperative mortality after colorectal cancer surgery: A retrospective cohort study

Background: Many scoring systems have been developed to predict outcomes after surgery, but it has limitations due to differences in population, comorbidity, type of surgery, or hospital effects. External validation from these scoring systems sometimes failed to achieve good discriminatory power consistently. This study aims to develop a novel scoring system for predicting postoperative mortality and comparing its performance with the AFC, CR-POSSUM, IRCS, and ACS-NSQIP SRC model. Methods: Data were collected retrospectively from all consecutive patients (n=1,294) undergoing colorectal cancer surgery in Dr. Soetomo Hospital between 2011 and 2020. After excluding missing data and 215 patients who did not satisfy the inclusion criteria, multivariate logistic regression analysis was performed in 1,079 patients to estimate odds ratios (ORs) and 95% confidence intervals (CIs) linking the explanatory variable postoperative mortality, and a Surabaya scoring system was constructed. Data were analyzed using SPSS version 23 for Windows. Results: Variables identified as the strongest predictors based on Odds Ratio (OR) postoperative mortaliy were albumin < 3.4 g/dL (6.93; 95%CI: 4.37-10.99; p< 0.001), pulse > 120 times per-minute (5.49; 95%CI: 2.11-14.29; p< 0.001), totally dependent functional status (4.43; 95%CI 2.06-9.49; p< 0.001), ascites (3.58; 95%CI: 1.84-6.94; p=0.001), major procedure (2.48; 95%CI: 1.38-4.33; p=0.009), dyspnea (2.40; 95%CI: 1.19- 4.84; p=0.014), and haemoglobin < 10 g/dL (1.85; 95%CI: 1.12-3.04; p=0.016). The Surabaya model predicted postoperative mortality with a predictive performance (0.831; 95%CI: 0.790-0.871) in the validation population. In this population the predictive performance of the AFC score was 0.630 (95%CI: 0.498-0.762), CR-POSSUM 0.698 (95%CI: 0.563-0.833), IRCS 0.564 (95%CI: 0.426-0.702), and ACS-NSQIP SRC 0.674 (95%CI: 0.541-0.806). Conclusion: The Surabaya score has been shown as a good predictor of postoperative mortality after colorectal cancer surgery despite the relatively low number of risk factors.