Abstract
Surgery for upper gastrointestinal malignancy carries a high postoperative mortality and morbidity risk. The importance of preoperative physiological reserve and intraoperative events in determining clinical outcomes is recognised in the Physiological and Operative Severity Score for the enUmeration of Mortality and Morbidity (POSSUM) score that comprises variables relevant to both phases. Whether adding variables linked to ICU admission characteristics improves the predictive capacity of POSSUM is unclear, especially in an Australian/New Zealand healthcare context. This study aimed to evaluate the predictive capacity of the POSSUM score for 30-day mortality and in-hospital morbidity in 80 patients undergoing resection of oesophageal (28%), gastric (26%) or pancreatic (46%) malignancies and admitted to ICU. The 30-day mortality was 8.8% and 65% of patients developed some postoperative complication. Receiver operating characteristics generated an area under the curve (95% CI) to predict mortality by Portsmouth POSSUM of 0.87 (0.77 to 0.93) and morbidity by POSSUM of 0.67 (0.55 to 0.77). Multiple regression analysis including biochemical variables and vital signs on admission to ICU identified renal function parameters, fluid balance and need for cardiorespiratory support beyond the first postoperative day as independent factors associated with mortality and morbidity (in addition to the POSSUM score) but the inclusion of these variables in a logistic regression model did not significantly improve the predictive capacity for mortality (to area under the curve 0.93 [0.85 to 0.97]) or morbidity (to area under the curve 0.67 [0.55 to 0.78]). In conclusion, the POSSUM score provides clinically useful predictive capacity in patients undergoing surgery for upper gastrointestinal malignancies. The incorporation of ICU admission variables to the pre- and intraoperative POSSUM variables did not significantly enhance the precision.