A pre-operative nomogram for decision making in oncological surgical emergencies

Background The purpose of the study was to propose a clinical decision-making tool for predicting mortality in patients undergoing emergency abdominal surgery with a palliative intent in the oncology setting. Methods Identification of all emergency surgical procedures performed in a Department of Oncologic Surgery in a Comprehensive Cancer Center between January 2008 and January 2013. Multivariate logistic and Cox regression models were used to identify factors predicitve of mortality at 3 months and survival probabilities. Models were internally validated using bootstrapping and calibration. Results The mortality rates were 30% at 1 month, 46.7% at 3 months and 83.3% at the end of the study. One model based on the albumin level and the P-POSSUM score (AUC: 0.725) adequately predicted mortality at 3 months. A survival nomogram predicted mortality with a concordance index (CI) of 0.718, using the following factors: WHO performance status (P = 0.02), albumin level (P < 0.01) and P-POSSUM score (P < 0.01). The origin or the extent of the carcinoma did not own sufficient pronostic impact to be selected in this model. Conclusions Pre-operative mortality risk scores can be developed in a palliative context. Physicians counselling and surgical decision making should be based on the use of these tools.