Purpose: This study examined the accuracy of Portsmouth Physiologic and Operative Severity Score for enUmeration of Mortality and Morbidity system (P-POSSUM) in predicting the mortality of patients who underwent operations for obstructing colorectal cancer. It also is attempted to analyze the actual mortality and the predicted P-POSSUM mortality of different surgical options for obstructing left-sided cancer.
Methods: Data on patients who underwent surgery for obstructing colorectal cancer during 1998 to 2002 were collected. Mortality predicted by P-POSSUM was compared to the actual mortality with the method of linear analysis. The accuracy of using P-POSSUM to predict mortality in this group of patients was assessed by Hosmer and Lemeshow goodness of fit test and Receiver Operator Characteristic curve analysis. The predicted and actual mortality of patients who underwent different surgical options also were analyzed.
Results: A total of 160 patients were included in the study and 18 patients died postoperatively. The operative mortality was 11.3 percent. P-POSSUM predicted overall mortality of 15 percent. The observed and predicted mortality was found to have no significant lack of fit (chi-squared = 5.98; degree of freedom = 3; P = 0.11). The area under Receiver Operator Characteristic curve analysis was 0.75. For patients with left-sided tumors, P-POSSUM predicted mortality and actual mortality of patients who had resection without anastomosis were both significantly higher than patients with single-stage resection and primary anastomosis (P = 0.044 and 0.011, respectively).
Conclusions: P-POSSUM system is valid for prediction of overall mortality in patients with operations for obstructing colorectal cancer. Estimation of P-POSSUM predicted mortality during operation and its ability to correlate with choice of procedure is an area that is worth further study in emergency colorectal surgery.