A newly devised scoring system for prediction of mortality in patients with colorectal cancer: a prospective study

Abstract

Postoperative morbidity and mortality from colorectal cancer varies widely across hospitals in the UK. We aimed to assess whether a newly developed score from the Association of Coloproctology of Great Britain and Ireland (ACPGBI) could predict mortality from colorectal cancer surgery as accurately as the Physiology and Operative Severity Score for enUmeration of Mortality and morbidity (POSSUM), Portsmouth POSSUM (P-POSSUM), or the ColoRectal POSSUM (CR-POSSUM). We analysed prospectively 618 patients with histologically confirmed colorectal cancer who had surgery to remove primary tumours done by colorectal surgeons or non-colorectal surgeons in a 3-year period. We compared observed mortality with those predicted by the ACPGBI, POSSUM, P-POSSUM, and CR-POSSUM scoring systems using the Hosmer-Lemeshow test and Receiver Operating Characteristic (ROC) curve analysis. Between April 1, 2002, and May 31, 2005, 618 consecutive patients with colorectal cancer had surgery to remove primary tumours. Overall observed 30-day mortality over the 3 years was 10.2% (95% CI 8.0-12.9). Overall predicted mortality (mean score) by use of POSSUM was 12.7% (11.7-13.7), by use of P-POSSUM was 4.4% (3.4-5.4), by use of CR-POSSUM was 9.6% (8.6-10.6), and by use of ACPGBI score was 8.1% (7.3-8.8). POSSUM overpredicted mortality, whereas P-POSSUM underpredicted mortality from colorectal-cancer surgery. CR-POSSUM was a more-accurate predictor of mortality in most analyses than was POSSUM and P-POSSUM. Although CR-POSSUM gave the closest prediction of overall mortality, analyses of subgroups of patients showed that ACPGBI score predicted overall mortality most accurately.