Introduction: The aim of the present study was to assess the predictive capacity of the POSSUM system in a Spanish university hospital, and to determine its behaviour in elective gastrointestinal surgery and compare it with emergency gastrointestinal surgery (operation < 24 hours). Patients and method: A total of 1,000 surgical episodes corresponding to 909 patients who required hospital admission, operated on under general or loco-regional anaesthesia, either in the elective (n=547 episodes) or the emergency setting (n=453), were included in the study. Results: The overall morbidity was 31.9% (32.8% in elective surgery; 30.7% in emergency surgery). The discriminatory capacity of the POSSUM scale, evaluated using receiver operating characteristic (ROC) curves, was higher for the Portsmouth variant of mortality (Area Under the Curve [AUC] = 0,92) than for morbidity (AUC=0,74). The goodness of fit between the expected values using the POSSUM scale and those observed was reduced for morbidity (Hosmer-Lemeshow [H-L] = 164.1; p<0.05). The POSSUM scale predicted a higher number of deaths than those observed, although the Portsmouth variant was better at predicting mortality. The goodness of fit for morbidity was better for elective gastrointestinal surgery (H-L=27.7) than emergency gastrointestinal surgery (H-L=177.3). The logistic regression analysis identified (besides the estimated risk using the POSSUM scale itself), surgical complexity, surgery type (elective, emergency), and age of patient, as significant predictive factors of morbidity and mortality. Conclusions: In a Spanish university hospital, the POSSUM system adequately predicts morbidity risk in elective gastrointestinal surgery, and over-estimates morbidity risk in emergency gastrointestinal surgery. (C) 2011 AEC. Published by Elsevier Espana, S.L. All rights reserved.
