Comment on Ekinci et al.: A randomized comparison of droperidol, metoclopramide, tropisetron, and ondansetron for the prevention of postoperative nausea and vomiting.

Abstract

type I error level of 0.01 for the detection, e.g. of a difference between rates of 50 and 31% (this setting would match the assumptions of the authors), we found 244 patients per group to be included. We therefore kindly propose to publish the details of the calculations presented by Ekinci et al. since we (and certainly also other investigators) would like to learn and benefit from innovative sample size sparing methods. The intention of the authors to document side effects in a larger population by improving the sample size from 15 to 20 per group is nice, but they have only a 20% chance of detecting a common side effect. Furthermore, we are concerned about the dose of metoclopramide used in this study. It was already shown in a large multicenter trial that 25 or 50 mg of metoclopramide was efficacious while 10 mg was not [4] . Even if in that trial metoclopramide was given on top of dexamethasone (the latter was applied in all patients), the results of a trial with n 1 750 per arm somehow suggest that 10 mg of metoclopramide should no longer be used as an active comparator. In the discussion the authors stated that Mitsunari et al. [5] found ondansetron to be superior to metoclopramide and droperidol. How should these investigators come to this result when they only investigated droperidol in the prevention of PONV? What does the term ‘Time (min)’ on the y-axis of figure 2 mean? Last but not Ekinci et al. [1] present a five-armed trial comparing four drugs versus placebo, aimed at preventing postoperative nausea and vomiting (PONV), including 20 patients per group, and draw brave conclusions. How to study PONV is well described in recommendations by expert opinions and consensus guidelines [2, 3] . According to these, PONV as a whole is the primary outcome and other outcome variables; mainly nausea, vomiting, rescue treatment and the author’s construct of ‘severe emetic sequelae’ should be assessed independently. In contrast to the title of their paper, Ekinci et al. did not report about PONV. The commonly accepted outcome of vomiting (yes or no) yielded non-significant results. Yet, conclusions are based on significances regarding ‘severe vomiting’ (versus non-severe or no vomiting) which is not defined anywhere, and hence not really in agreement with the CONSORT recommendations on reporting endpoints in clinical trials. Nausea is mentioned in the prosaic part of the paper, but no data are presented. The sample size calculation is indeed a mathematical masterpiece. Using a conventional sample size software (PASS 2002, www.ncss.com) we found a maximum power of 37% to detect a 19% difference in event rates with 20 patients per group even at a type I error level of 0.05. To ensure a power of 95% at a two-sided Received: April 7, 2011 Accepted: April 10, 2011 Published online: August 27, 2011

DOI: 10.1159/000329303

Cite this paper

@article{Gelbrich2011CommentOE, title={Comment on Ekinci et al.: A randomized comparison of droperidol, metoclopramide, tropisetron, and ondansetron for the prevention of postoperative nausea and vomiting.}, author={Goetz Gelbrich and Jan Wallenborn}, journal={Gynecologic and obstetric investigation}, year={2011}, volume={72 3}, pages={215; author reply 216} }