Reporting Sex Differences in Research Publications
Journal policies lay out what is required of manuscripts that are submitted for publication, Workshop Cochair Rae Silver noted as she opened the session on reporting sex differences in research publications. If journal editors believe that knowing the sex of origin of the cell type discussed in the report is important, or likewise the sex of the animals or human participants in a study, then investigators will have to include that information when describing strains, species, and participants in manuscripts. In this session, two panelists representing peer-reviewed, professional neuroscience journals provided perspectives on the status of reporting sex differences in neurological health and disease.
JOURNAL OF NEUROCHEMISTRY
Sean Murphy, professor in the Department of Neurological Surgery at the University of Washington Medical School, is midway through an 8-year appointment as editor of the Journal of Neurochemistry (JNC). Established 50 years ago, JNC was the first neuroscience journal. JNC receives about 2,000 manuscripts each year and publishes about 30 percent of those, in a variety of areas from molecular biology through animal studies (JNC does not publish clinical studies).
In conducting his own informal survey, Murphy found that out of 30 journals that focus on or report neuroscience studies, only one, the European Journal of Neuroscience, stated directly in the author instructions a requirement for reporting the species, strain, sex, age, supplier, and
numbers of experimental animals used. He was surprised to find that his own publication, JNC, did not require the sex of animals to be reported.
A recently published survey of 271 randomly selected articles that reported the results of animal studies found less than 60 percent stated clearly the hypothesis, and the number and characteristics of the animals used, including sex (Kilkenny et al., 2009). More than 85 percent did not report any attempt at randomization or blinding to reduce bias in assessing outcomes, and 30 percent did not report statistical methods. This lack of information sharing may well contribute to why animal models are not optimized.
In response to this, guidelines are currently being developed that will ask editors of journals to require authors to address a checklist of 20 items that are the minimum information that should be included in all scientific publications reporting research using animals (e.g., number and specific characteristics of animals used, including species, strain, sex, genetic background; details of housing and husbandry; and experimental, statistical, and analytical methods, including methods to reduce bias).
Murphy noted that although these guidelines will call for the reporting of the sex of animals used, it does not ask authors submitting an article to give the rationale for studying either males or females, or to describe what the potential implications are for not studying the other sex. Including these requirements would be an educational step, Murphy said, and as editors are currently sensitive to these issues, the time could be right to implement such requirements.
Marie-Francoise Chesselet, chair of the Department of Neurobiology at the University of California–Los Angeles and associate editor of Experimental Neurology, is also treasurer of the Society for Neuroscience (SFN) and an ex-officio member of the publications committee for the SFN publication, the Journal of Neuroscience. Money is often a primary reason why many studies do not use both sexes, she said. But a valid scientific explanation and justification should be given for studying only one sex. Chesselet supported Murphy’s suggestion that authors be required to state the potential implications of studying only one sex.
In many papers, Chesselet observed, authors indicate that they balance experimental groups by sex, but do not disclose the exact balance. In some cases that “balance” turns out to be, for example, two males and seven females in one group, and one female and eight males in the other group. This information should be explicitly disclosed, she said.
Proper analysis of data by sex should also be required because, as mentioned in the earlier discussions, even National Institutes of Health–
sponsored studies that do include both male and female animals do not always analyze the two sexes separately.
Journals have an educational role to play and can lead the way by providing, for example, checklists for authors and reviewers to consult when preparing and reviewing manuscripts. Reviewers should increase their awareness of the potential importance of sex-related effects, and they should not consider a finding that applies to only one sex as a shortcoming. Journal editors can also write editorials to increase awareness about the need for reporting sex differences. Researchers who serve on editorial boards should raise these points with their colleagues.
Silver suggested that for journals to be effective in changing the way researchers report sex and sex differences, a concerted and coordinated effort is needed to revise and enforce policies; otherwise authors may simply choose to submit their manuscripts to journals with less stringent requirements.
Many participants agreed that having journal editors require specification of sex in publications would not create an undue burden on scientists. One participant cited a 1994 statement by the New York Academy of Sciences that recommended journal editors and reviewers require specification of the numbers and proportions of males and females studied, and that generalization from single-sex studies should be restricted to the sex investigated. In 1992, the University of California–Berkeley similarly encouraged reviewers to ensure that authors state the sex and reproductive state of the species studied, and that editors adopt sex specification as part of journal policy. These recommendations had no way to be enforced, and fell on deaf ears, he said.
Decisions to invest in studying sex differences during drug development are based, in part, on the presence of symptomatic or epidemiologic differences in disease characteristics between males and females. On the other hand, it was also noted that studies of the underlying pathophysiology of diseases between sexes suggest large differences that may, in essence, cancel each other out, and lead to a common symptomatic or epidemiologic presentation. Therefore, dissemination of all information obtained regarding sex differences is very important.