To the Editor
The article by Garland et al1 recently published in Menopause reports outcomes that are different than those initially registered.2
The prespecified primary outcome (Pain as measured by Brief Pain Inventory [BPI] pain severity score) was not reported. A secondary outcome was converted in primary (Pittsburgh Sleep Quality Index [PSQI]), and a total of three other secondary outcomes were not reported. One additional non-prespecified secondary outcome (hot flash composite score [HFCS] as the weekly average of hot flashes reported in their daily hot flash diary) was reported in the article. Some of those outcomes were previously reported by the authors,3 but with a design that involved three arms (acupuncture, sham acupuncture, and no treatment). No reference to treatment with gabapentin was initially prespecified. If there were changing circumstances, it is essential that they are fully reported to help the reader interpret the results.
Outcomes related to hot flashes reported by the participants were previously reported by the authors in a paper cited in the methods section that describes the protocol for the study.4 But, such a design that effectively includes a gabapentin arm, belongs to another trial5 (ClinicalTrials.gov Identifier NCT01005108) and not to the one reported. At this point, it is difficult to understand what design/cohort the authors analyzed: the one that belongs to the reported trial that does not include a gabapentin arm and is declared in their article in Menopause, or the one that belongs to the ClinicalTrials.gov Identifier NCT01005108 that is reported in the “Methods” section of the paper and that includes two prespecified timepoints (12 and 24 weeks) that were not analyzed in this paper, plus a number of outcomes that were not reported. The last update of the ClinicalTrials.gov record NCT01013337 was in July 15, 2016. The authors have not reported any desirable change in their proposed protocol since that date.
Menopause endorses the recommendations of the International Committee of Medical Journal Editors, “Uniform Requirements for Manuscripts Submitted to Biomedical Journals” (http://www.icmje.org/update.html) that supports the CONSORT (Consolidated Standards of Reporting Trials) guidelines6 on best practice in clinical trials reporting. Any changes to trial outcomes after the trial commenced should be declared (with reasons) in the methods section to avoid selective outcome reporting7 that may produce an overestimation of benefit.