Validation of a risk score for prediction of vomiting in the postoperativeperiod

Background: A risk score to predict postoperative vomiting was presented in a recent issue of this journal. In the present study this score was evalua ted at another hospital under different surgical and anaesthetic conditions . Furthermore, we examined whether the score, which was originally designed to predict the occurrence of postoperative vomiting (POV) only, is also us eful for prediction of postoperative nausea and vomiting (PONV). Methods:The risk score was applied to 226 patients undergoing in surgery un der standardised general anaesthesia (propofol, desflurane in N2O/O-2, fent anyl, vecuronium, postoperative opioid analgesia). For 24 hours postoperati vely,the patients were followed up for the occurrence of nausea, retching, and vomiting. Perioperatively, risk factors for POV were recorded (gender, age,smoking habits, history of previous PONV or motion sickness, duration o f anaesthesia). Using these risk factors the individual al risk for sufferi ng POV was calculated for each patient. With these data two ROC-curves (for prediction of POV and PONV respectively) were constructed and the area und er the ROC-curve (AUC) as a means of the prediction probabilities of the sc ore was calculated. Results:The incidence of POV as predicted by the score (22,8%) fits well to the actual incidence of this event (19,5%). The score predicts the occurre nce of POV significantly better than can be expected by a random estimation . ln spite of different surgical and anaesthetic conditions, the accuracy o f the prediction in the present dataset was not significantly different fro m that reported by the authors of the scores in their validation set. Furth ermore, the prediction properties for POV (AUC: 0,73) were not different fr om the prediction of PONV (AUC: 0,72). Conclusion: The present risk score provides valid prognostic results even u nder modified surgical and anaesthetic conditions, and, thus, may obviously be applied to other institutions. Furthermore our results support the hypo thesis, that individual risk factors rather than the type of surgery or ana esthetic management have a major impact on the occurrence of POV and PONV.