68-6

23/104

regression analysis. The intraoperative use of prophylactic antiemetics and the postoperative use of rescue opioid analgesics other than PCA fentanyl were associated with the incidence of PONV with univariate logistic regression analysis, but not with the multivariate logistic regression analysis. Reportedly, female sex, types of surgery (laparo-scopic, gynecologic), the use of inhalation anes-thesia versus TIVA, non-smoking status, longer duration of anesthesia, younger age (< 50 years), and the postoperative use of opioids are established risk factors of PONV1). In the present study, however, age, non-smoking status, or the method of general anesthesia did not affect the incidence of PONV, possibly due to too small numbers of patients aged 50 years or more (n = 31 [2.5%]), smokers (n = 85 [7.0%]), and patients anesthetized with TIVA (n = 25 [2.0%]) to detect their signifi-cant effects. The postoperative use of opioids other than PCA fentanyl in 208 patients (17%) was asso-ciate with the incidence of PONV with the univar-iate logistic regression analysis, but not with the multivariate logistic regression analysis, possibly reflecting complex interactions among the use of opioids other than PCA fentanyl, and the use or no use of fentanyl PCA with or without droperidol. BMI was not associated with the incidence of PONV, in agreement with previous reports1, 5).Reportedly, prophylactic uses of antiemetics, such as dopamine receptor antagonists including droperidol and metoclopramide, corticosteroids including dexamethasone, and serotonin receptor antagonists including ondansetron and granisetron, are effective in reducing the incidence of PONV1, 4). In the present study, however, the effect of prophy-lactic antiemetics given during anesthesia was detected with the univariate regression analysis, but not with the multivariate logistic regression analysis, possibly because prophylactic antiemetics were given only in much smaller number of patients (n = 136 patients [11.1%]), compared with our more recent study4), and also because metoclopra-mide, which is less effective than droperidol1), was more frequently used in the present study (meto-clopramide in 96 patients [7.8%] versus droperidol in 40 patients [3.3%]). On the other hand, longer duration of anesthesia was associated with the increased incidence of PONV in accordance with the guideline1). This finding could be related to the higher doses of potentially emetic drugs, such as inhalation anesthetics, administered during longer anesthesia. One of major findings of the present study was that the use of fentanyl PCA with droperidol was associated with the significantly lower incidence of PONV, compared with no use of PCA and the use of fentanyl PCA without droperidol. This finding is in good agreement with previous studies showing that addition of droperidol to PCA infusate reduces the incidence of PONV4, 6, 7). Conversely, the use of fentanyl PCA without droperidol tended to be asso-ciated with the higher incidence of PONV, compared with the no use of PCA. This finding agrees with the established finding showing that the postopera-tive use of opioids increases the incidence of PONV1). Paradoxically, however, PONV occurred less frequently in patients receiving fentanyl PCA with droperidol than those not receiving fentanyl PCA. This finding might be explained by the fact that postoperative rescue opioids, which would increase the incidence of PONV1), were given much more frequently in patients not receiving PCA (without co-administration of an antiemetic) than those receiving fentanyl PCA with an antiemetic droper-idol (42.5% and versus 6.8%, p < 0.0001). Interest-ingly, not only PONV occurred less frequently but also postoperative rescue opioid analgesics were used less frequently in patients receiving fentanyl PCA with droperidol than those receiving fentanyl PCA without droperidol (6.8% versus 14.3%, p = 0.0003), possibly reflecting the opioid sparing effect of droperidol8).Another major finding of the present study was that the use of the reversal agent neostigmine 2 mg co-administrated with atropine 1 mg was associ-ated with the lower incidence of PONV. A previous meta-analysis reported that neostigmine in doses ≥ 2.5 mg increases the incidence of PONV9). However, a more recent meta-analysis revealed that the use of neostigmine combined with anticho-linergic agents such as atropine did not increase the incidence of PONV10). While previous studies thus have demonstrated that neostigmine increases or does not increase the risk of PONV9, 10), no study has demonstrated that the use of a reversal agent neostigmine combined with anticholinergic agents decreases the incidence of PONV. Such discrepan-cies might result from the fact that all of previous 587