Dr. Stein and Colleagues Reply

To the Editor: We thank Drs. Jainer and Chawla for their interest in our work. We agree with the gist of their letter emphasizing the importance of detailed reporting of methods for randomized controlled trials. The CONSORT guidelines provide an excellent framework for publication of the results of randomized controlled trials, and we agree that each of the elements recommended therein should ideally be provided. Earlier drafts of our article did, indeed, include more of these methodological details, but several pieces of information were deleted as we worked to make our manuscript fit the Brief Report format. We support provision of additional space to permit more rigorous adherence to the CONSORT guidelines in all future reports in the Journal. It will be up to the Journal’s editors, of course, to determine if this is feasible.

We will use this opportunity (and space) to provide some of this additional information. The subjects included in this report were all 19 participants who met the entry criteria and agreed to participate; no other subjects were enrolled. We believe that we discussed the unique characteristics of our subjects (i.e., treatment-resistant male combat veterans) and the resultant probably limited generalizability of our results in sufficient detail in the report. Randomization was conducted by using a random-numbers table prepared by our research pharmacy. The code for randomization was also maintained in the pharmacy, where it could be broken in an emergency—which did not occur. Medications (active drug and identically appearing matching placebo tablets) were provided by the drug’s manufacturer and dispensed by the research pharmacy. We have no reason to believe that the double blind was compromised at any point during the study. We did not report effect sizes in the report, but these were easily calculated from the data provided (i.e., mean change/standard deviation). The effect size (≈1.0) was actually large for the drug-placebo difference in change on the Clinician-Administered PTSD Scale, as it would have to be in order to be statistically significant with such a small group. Response rates were clearly indicated in the article.