Background: /st> In assessing a patient's risk for postoperative nausea and vomiting (PONV), it is important to know which risk factors are independent predictors, and which factors are not relevant for predicting PONV.
Methods: /st> We conducted a systematic review of prospective studies (n>500 patients) that applied multivariate logistic regression analyses to identify independent predictors of PONV. Odds ratios (ORs) of individual studies were pooled to calculate a more accurate overall point estimate for each predictor.
Results: /st> We identified 22 studies (n=95 154). Female gender was the strongest patient-specific predictor (OR 2.57, 95% confidence interval 2.32-2.84), followed by the history of PONV/motion sickness (2.09, 1.90-2.29), non-smoking status (1.82, 1.68-1.98), history of motion sickness (1.77, 1.55-2.04), and age (0.88 per decade, 0.84-0.92). The use of volatile anaesthetics was the strongest anaesthesia-related predictor (1.82, 1.56-2.13), followed by the duration of anaesthesia (1.46 h(-1), 1.30-1.63), postoperative opioid use (1.39, 1.20-1.60), and nitrous oxide (1.45, 1.06-1.98). Evidence for the effect of type of surgery is conflicting as reference groups differed widely and funnel plots suggested significant publication bias. Evidence for other potential risk factors was insufficient (e.g. preoperative fasting) or negative (e.g. menstrual cycle).
Conclusions: /st> The most reliable independent predictors of PONV were female gender, history of PONV or motion sickness, non-smoker, younger age, duration of anaesthesia with volatile anaesthetics, and postoperative opioids. There is no or insufficient evidence for a number of commonly held factors, such as preoperative fasting, menstrual cycle, and surgery type, and using these factors may be counterproductive in assessing a patient's risk for PONV.