The revised ACPGBI model is a simple and accurate predictor of operative mortality after potentially curative resection of colorectal cancer.

Abstract

The Association of Coloproctology of Great Britain and Ireland (ACPGBI) risk-adjustment model for colorectal cancer surgery has been recently revised. The aim of the present study was to compare the performance of the revised ACPGBI model, the original ACPGBI model, P-POSSUM, and CR-POSSUM, in the prediction of operative mortality after resection of colorectal cancer. A total of 423 patients who underwent potentially curative resection of colorectal cancer at a single institution (1997-2007) were included. Data used in the construction of the ACPGBI model was collected prospectively. The models were compared by examining observed to expected (O:E) ratios, the Hosmer-Lemeshow (H-L) goodness-of-fit test, and area under the receiver operator characteristic curve (AUC) analysis. The 30-day mortality rate was 4%. The performance of the models was as follows: revised ACPGBI model (O:E ratio = 1.05, AUC = 0.73, H-L = 11.02), original ACPGBI model (O:E ratio = 0.58, AUC = 0.76, H-L = 14.23), P-POSSUM (O:E ratio = 0.87, AUC = 0.79, H-L = 10.63), and CR-POSSUM (O:E ratio = 0.63, AUC = 0.84, H-L = 15.84). In subgroup analysis, the revised ACPGBI model performed well in both elective cases (O:E ratio = 1.06) and emergency cases (O:E ratio = 0.91). The revised ACPGBI model is simple to construct and accurately predicts operative mortality after potentially curative resection of colorectal cancer.