2005 Abstracts: Risk Adjusted Prediction of Operative Morbidity in Patients Undergoing Pancreatoduodenectomy with the Use of Possum
Back to 2005 PostersBack to 2005 Program and Abstracts
Risk Adjusted Prediction of Operative Morbidity in Patients Undergoing Pancreatoduodenectomy with the Use of Possum
S.M.M. de Castro, Academic Medical Center, Amsterdam, Noord Holland, Netherlands; J.T. Houwert, Academical Medical Center, Amsterdam, Noord Holland, Netherlands; K.F.D. Kuhlmann, Academica Medical Center, Amsterdam, Noord Holland, Netherlands; O.R.C. Busch, Academic Medical Center, Amsterdam, Noord Holland, Netherlands; T.M. van Gulik, Academica Medical Center, Amsterdam, Noord Holland, Netherlands; H. Obertop, D.J. Gouma, Academic Medical Center, Amsterdam, Noord Holland, Netherlands
Comparison of operative morbidity ratesbetween units may be misleading because it does not account for the physiological variation of the condition of the patient at the time of surgery. Especially since the awareness of the volume-outcome effect of pancreatic surgery results in an tendency to refer high risk patients to high volume centers. The aim of the study was to evaluate the applicability of the Physiological and Operative Severity Score for the enUration of Mortality and Morbidity (POSSUM) for patients who undergo pancreatoduodenectomy and to look for specific risk factors associated with morbidity in a high volume center.
Between January 1993 and December 2003, 511 patients underwent a pancreatoduodenectomy of which 424 (83%) for malignant disease. POSSUM was calculated according to the generally accepted criteria. The performance of POSSUM was evaluated by assessing the “goodness of fit” using the exponential analysis method. The ratio of observed to predicted morbidity (O:P ratio) was calculated. A ratio greater than 1 indicates that the model under-predicts morbidity and a ratio less than 1 indicates over-prediction of morbidity. Secondly, predictive factors of interest associated with morbidity were analyzed using Univariate an Multivariate analysis. Overall, 285 of 511 patients (51%) had one or more complication after pancreatoduodenectomy and 7 patients (1.4%) died during initial hospital stay . The O:P ratio for POSSUM was 1.09. In Multivariate analysis, four statistically significant factors associated with an increased morbidity (p<0.05) were identified: hypertension (Odds ratio (OR) = 1.86, 95% Confidence interval (CI): 1.10-3.14), advanced age >76 years (4th quartile) (OR = 1.73, 95% CI: 1.01-2.95), male gender (OR = 1.56, 95%CI 1.05-2.94) and ampulla of Vater adenocarcinoma (OR = 1.66, 95% CI: 1.01-2.75). The first three factors are already incorporated in POSSUM. Therefore, POSSUM was subsequently analyzed for Ampulla of Vater (n=123) and compared with the remainder of the group (n=388). The O:P ratio was 1.24 and 1.05, respectively. Overall, POSSUM performed well and may serve as a useful comparative audit tool for patients who undergo pancreatoduodenectomy. A dedicated PAN-POSSUM model can be made by adjusting for tumor pathology since this was the only factor which was not incorporated in POSSUM
Back to 2005 Posters
Back to 2005 Program and Abstracts
