Validation of the CR-POSSUM risk-adjusted scoring system for major colorectal cancer surgery in a single center

PURPOSE: Scoring systems to predict mortality from surgery are important tools used to give information to the operator and patient and in the auditing of clinical practice. This study was designed to validate the recently developed the Colorectal Physiologic and Operative Severity Score for the Enumeration of Mortality and Morbidity (CR-POSSUM) scoring system in a single center for colorectal cancer surgery. We also analyzed whether albumin may have a role in the CR-POSSUM model. METHODS: We compared this model with two other scoring systems: POSSUM and Portsmouth-POSSUM (P-POSSUM) models. Inhospital mortality was used as the outcome, and Hosmer-Lemeshow statistic was used to determine goodness of fit. RESULTS: Complete data were collected prospectively from 304 patients from 1990 to the present. The overall operative mortality was 6.5 percent. Observed to expected ratios were used to compare the scoring systems at a given predicted mortality. The overall observed to expected ratio was 1.25 for CR-POSSUM, 1.59 for P-POSSUM, and 3.37 for POSSUM. The CR-POSSUM model showed a good fit with the data (Hosmer-Lemeshow statistic, 3.86; P = 0.795) and the area under the receiver operator curve was 0.74. After correcting for factors used in the CR-POSSUM, logistic regression showed a significant correlation between albumin and mortality (P=0.016). CONCLUSIONS: We have shown that the CR-POSSUM model is an accurate predictor of outcome for major colorectal surgery. The POSSUM and P-POSSUM models over-predicted mortality. Albumin, which is not a factor included in these three systems, may be an important addition in improving the accuracy of the CR-POSSUM model.