Evaluation of POSSUM in patients with oesophageal cancer undergoing resection

Abstract Background The Physiogical and Operative Severity Score for the enUmeration of Mortality and morbidity (POSSUM) has been used to produce a numerical estimate of expected mortality and morbidity after a variety of general surgical procedures. The aim of this study was to evaluate the ability of POSSUM to predict mortality and morbidity in patients undergoing oesophagectomy. Methods POSSUM predictor equations for morbidity and mortality were applied retrospectively to 204 patients who had undergone oesophagectomy for cancer. Observed morbidity and mortality rates were compared with rates predicted by POSSUM using the Hosmer–Lemeshow goodness-of-fit test. Evaluation of the discriminative capability of POSSUM predictor equations was performed using receiver–operator characteristic (ROC) curve analysis. Results The observed and predicted mortality rates were 12·7 and 19·1 per cent respectively, and the respective morbidity rates were 53·4 and 62·3 per cent. However, the POSSUM model showed a poor fit with the data both for the observed 30-day mortality (χ2 = 16·26, P = 0·002) and morbidity (χ2 = 63·14, P < 0·001) using the Hosmer–Lemeshow test. ROC curve analysis revealed that POSSUM had poor predictive accuracy both for mortality (area under curve 0·62) and morbidity (area under curve 0·55). Conclusion These data suggest that POSSUM does not accurately predict mortality and morbidity in patients undergoing oesophagectomy and must be modified.