A risk score to predict the probability of postoperative vomiting in adults

Background: The aim of this study was to identify factors most relevant for postoperative vomiting (PV) and to develop a risk score to predict the probability of PV. Methods: Adult inpatients scheduled for elective ear, nose and throat (ENT) surgery under general anaesthesia were offered to participate in a prospective study for PV over 24 h. No prophylactic antiemetics were used. The data of 1137 patients were randomized and split into an evaluation set (n=553) and a validation set (n=584). The evaluation set was subjected to logistic regression analysis to quantify the relative impact of anaesthetic, surgical and individual factors and to develop a risk score. The score was then tested by applying it to the validation set. The area under a receiver operation characteristic (ROC) curve was calculated and the predicted and actual incidences of patients were correlated. Results: In the evaluation set, patient-related factors (female gender, young age, non-smoking, history of PV or motion sickness) and a high duration of anaesthesia were independent risk factors for PV. The probability of PV could be estimated from the equation: PV=1/(1+exp(-z)) where z=1.28 . (gender)-0.029 . (age)-0.74 . (smoking)+0.63 . (history of PV or motion sickness)+0.26 . (duration)-0.92. In the validation set this score achieved an area under the ROC-curve of 0.78 and the actual incidence correlated strongly with the predicted risks (R-2=0.93, P<0.001). Conclusion: The data suggest that the probability of PV following ENT surgery under inhalational anaesthesia with low-dose opioids can be predicted by a score mainly based upon patient-related risk factors.