a) American Journal of Epidemiology (AJE) publishes 2 issues/month, for 12 months (24 issues/year).
b) American Journal of Public Health (AJPH) publishes 1 issue/month, for 12 months (12 issues/year).
c) Epidemiology publishes 1 issue every 2 months (6 issues/year).
d) Nature Genetics publishes 1 issue/month, for 12 months (12 issues/year).
For each article, we determined whether sex or gender was used in the title or abstract, and then coded for all of the following terms: sex, gender, male, female, men, women, maternal, paternal, and X chromosome throughout the article. We ascertained whether the analysis focused on a single sex or gender, and if so, whether the health outcome under study was a reproductive or sex-specific health outcome. Among those articles that were not single sex or gender studies, we reviewed the context in which all of the terms were used, and developed codes to summarize how sex or gender was used in analysis. Findings, which are presented for all journals as well as for the public health journals only, are shown in Table 1, and discussed further in the following.

TABLE 1—

Summary of Incorporation of Sex or Gender in Research Articles on Genetics and Health
VariableAll Journals, a No. (Proportion)Public Health Journals, b No. (Proportion)
Total number of articles198139
Single sex/gender analysis sample
Total3532
Reproductive or sex-specific outcomes12 (0.34)10 (0.31)
Non–sex-specific health outcome23 (0.66)22 (0.69)
Multiple sex/gender analysis sample
Total163107
Level of inclusion
 Sex/gender in title2 (0.01)0 (0.00)
 Sex/gender in abstract19 (0.12)18 (0.13)
Type of inclusion
 Match on sex in sampling20 (0.12)14 (0.10)
 Control/adjust for sex in analysis93 (0.57)62 (0.45)
 Incorporate sex with twin design16 (0.10)15 (0.11)
 Stratify analysis by sex34 (0.21)22 (0.16)
 Interact sex with genetic variant12 (0.07)6 (0.04)
 No mention of treatment of sex/gender in analysis28 (0.17)10 (0.07)
Note. All articles were published between 2007 and 2011. Ten studies did not clearly indicate which sex(es) or gender(s) they included and are presumed not to be single sex or gender studies.
a) All Journals includes American Journal of Epidemiology, American Journal of Public Health, Epidemiology and Nature Genetics.
b) Public Health Journals excludes Nature Genetics.
c) Breast cancer was coded as non–sex-specific health outcome.
Of the 198 articles, 35 included empirical analyses based on data specific to either men or women. Of these 35, 12 featured reproductive or sex-specific outcomes (e.g., endometrial cancer). Another 23 articles limited analyses to either men or women, but studied a health outcome that was not specific to a single sex (e.g., atherosclerosis). Of the articles that included both men and women, very few featured the terms sex or gender prominently in the title or abstract, which is suggestive of the relative (un)importance of these 2 concepts in the analyses themselves. One study of nongenital cancers published in the American Journal of Epidemiology was motivated by sex differences in health, and examined, but did not find support for, X-chromosome linked factors.27 Another American Journal of Epidemiology study of blood pressure and serum insulin investigated sex-specific distributions of the outcomes.28 Overall, the general motivation for taking sex into account, if stated, included both sex-specific variation in physiological differences and sex or gender specific differences in health outcomes. Notably, in 20 articles (16 of which were published in public health journals), discussion of sex or gender was absent because of the case-control design approach, which led researchers only to remark in their data section that cases and controls were matched on sex and other factors. About 10% of studies including men and women relied on a twin design and often same-sex pairs.
By and large, the treatment of sex took the form of a control variable, or a term included in regression analyses to account for—but not necessarily explain—differences between men and women. Of all articles based on data from males and females, 57% adjusted for sex. Alternatively, some 21% of all articles not limited to a single sex, stratified by sex. In research that entailed investigation of genome-wide genetic associations, analyses sometimes included the X chromosome. One study, an association analysis of 249 796 individuals in search of loci associated with body mass index, was an example of an article with detailed attention to sex; analyses were adjusted for sex, stratified by sex, and included testing for sex-by-gene interactions.29 Still others tested for substantively plausible interactions by sex, and notably, half of the articles that did so were published in Nature Genetics. Finally, in roughly 15% of articles that included men and women, sex and gender were not discussed and were not used in analysis.
In reviewing these articles, we offer 3 observations. First, established sex and gender differences in health and longevity motivate very little research on genetics and health in these journals. Second, sex, when incorporated, is most often treated as a confounding factor, or a source of variation that needs to be controlled, to make room for substantive conclusions about other factors. Third, the conceptual reasoning behind the incorporation of sex as a control is rarely articulated. Does it derive from observed differences in gender and health, observed sex or gender differences in relevant genetic or biological pathways, assumptions about relevant sex difference in physiological “environment,” or some combination? The lack of systematic attention to sex differences in health is curious, given that such differences are persistent, largely unexplained, and presumed to be, to some extent, a function of biology.