Preventing postoperative nausea and vomiting

Editor—Metoclopramide is one of the drugs that we reviewed in our recent Cochrane meta-analysis.¹ We also found that metoclopramide was effective. We found that, compared with placebo, the average relative risk for postoperative nausea and vomiting was 0.76 for all doses of metoclopramide, which compares to 0.89, 0.75, and 0.63 for 10 mg, 25 mg, and 50 mg of metoclopramide in the study reported by Wallenborn et al.² We would like to comment on the useful information that Wallenborn's study provides in this context.

Wallenborn et al provided the most information linking antiemetic dose with effect. We found a clear pattern of increasing effect in our review for droperidol, but less convincing patterns for other antiemetics, including metoclopramide.¹ Wallenborn et al did not measure a significant difference between 10 mg of metoclopramide and placebo. From our systematic review, we think that 10 mg of intravenous metoclopramide does have an effect, but not sufficient to be detected at the P = 0.05 level in 788 participants with a control risk of 0.23.

We included 737 studies involving 103 237 people.¹ We found convincing evidence that nine drugs were effective. But the results for all nine were skewed, probably by publication bias. The results for metoclopramide were skewed the least, as illustrated by a funnel plot that was only mildly asymmetric. As Sweeney comments in his editorial,³ many clinicians either dismiss metoclopramide or consider it to be a weak antiemetic. This conservative expectation has probably resulted in the least distortion of effect for metoclopramide, perhaps by “allowing” the publication of studies that show little or no antiemetic effect.

We found that the funnel plots for newer agents, either in comparison with placebo or older antiemetics, were markedly skewed. We think that clinicians should be cautious, in both extolling the virtues of new antiemetics and in discarding older drugs, particularly when rare adverse reactions are detected only after a large number are exposed to a new drug.

In the United Kingdom 10 mg of intravenous metoclopramide costs about £0.27. The metoclopramide cost of prophylaxis for 1000 people would be £270 (€396; $510), £675, and £1350 at the three doses in this study. This compares to about £5990 for 4 mg of ondansetron and £8600 for 1 mg of granisetron.

Finally we disagree with Sweeney's suggestion that a head to head trial of metoclopramide and dexamethasone versus a 5-HT3 antagonist combined with dexamethasone would be the next logical step.³ This proposed study would add little to all the other studies of metoclopramide and 5-HT3 antagonists. If clinicians remain set on more “primary” research, despite the incomplete synthesis of the research already published, they should concentrate on the detection of rare serious side effects.