OBJECTIVETo investigate the value of Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity (POSSUM) and a modification of the POSSUM system (P-P0SSUM) scoring system in predicting the surgical operative risk of pancreaticoduodenectomy for periampullary tumors.

METHODSPOSSUM and P-POSSUM scoring systems were used to retrospectively evaluate the clinical data of 432 patients with periampullar tumors who underwent pancreaticoduodenectomy in the Department of Abdominal Surgery, Cancer Hospital, Chinese Academy of Medical Sciences from January 1985 to December 2010. The predictive occurrence of postoperative complications and mortality rate were calculated according to the formula. ROC curve analysis and different group of risk factors were used to determine the discrimination ability of the two score systems, and to determine their predictive efficacy by comparing the actual and predictive complications and mortality rates, using Hosmer-Lemeshow test to determine the goodness of fit of the two scoring systems.

RESULTSThe average physiological score of the 432 patients was 16.1 ± 3.5, and the average surgical severity score was 19.6 ± 2.7. ROC curve analysis showed that the area under ROC curve for mortality predicted by POSSUM and P-POSSUM were 0.893 and 0.888, showing a non-significant difference (P > 0.05) between them. The area under ROC curve for operative complications predicted by POSSUM scoring system was 0.575. The POSSUM score system was most accurate for the prediction of complication rates of 20%-40%, showing the O/E value of 0.81. Compared with the POSSUM score system, P-POSSUM had better ability in the prediction of postoperative mortality, when the predicted value of mortality was greater than 15%, the predictive result was more accurate, and the O/E value was 1.00.

CONCLUSIONSPOSSUM and P-POSSUM scoring system have good value in predicting the mortality of patients with periampullary tumors undergoing pancreaticoduodenectomy, but a poorer value of POSSUM score system in prediction of complications. We can establish a more suitable scoring system for pancreaticoduodenectomy by modifying the score constant and weight, to better predict surgical risk and reduce the operative complications and mortality.

Ampulla of Vater ; Common Bile Duct Neoplasms ; mortality ; surgery ; Humans ; Morbidity ; Pancreaticoduodenectomy ; adverse effects ; mortality ; Postoperative Complications ; diagnosis ; mortality ; Postoperative Period ; Predictive Value of Tests ; ROC Curve ; Retrospective Studies ; Risk Assessment ; Risk Factors