Background
Main text
Defining sex and gender
Intersectionality
What is meant by sex- and gender-disaggregated research
Sex- and gender-disaggregated research is not only about women
Using the other sex as a comparator group
Example 1: Finding that a large percentage of women do not receive guideline-based care may be headline grabbing, but if men have a similarly low prevalence, the most crucial finding is that better care is required per se. This was the case in a survey of care given to people living with CHD that found only 6% of women were treated to target, for a cluster of risk factors [20]. This is an extremely poor result, which is worthy of attention, but cannot be used to show that women are disadvantaged since the equivalent result for men was 8%. The message here is to, whenever possible, include the other sex, perhaps only to serve as a comparator group, to produce meaningful findings even if the interest of the research is on a single sex |
Example 2: As an example of where including men as comparator group led to a different interpretation, consider the effect of increasing family size on cardiometabolic risk. Several studies showed that women with a higher number of pregnancies were at a higher risk of cardiometabolic diseases [21‐23]. While there are biological reasons to support this, even when ruling out the role of adverse pregnancy outcomes, having large families might also impose a burden on the cardiovascular system. Men cannot get pregnant, but they do get children. Men can therefore be used as a control group in determining whether it is childbearing or childrearing that explains the associations between the number of pregnancies and cardiovascular risk seen in women. In analyses in the UK Biobank and China Kadoorie Biobank, we demonstrated that the association between number of children and the risk of cardiometabolic diseases was similar in women and men [23‐25]. Hence, it may be mainly childrearing, and not childbearing, that underpins the association between the number of pregnancies and cardiovascular risk in women. Interestingly, in the UK Biobank, those with the lowest risk of CVD, had two children whereas having one child was associated with the lowest risk in the China Kadoorie Biobank. This might suggest that societal norms, structures, and policies on preferred family size might explain why those deviating from that preferred standard are at a higher risk of CVD |
Elements of sex and gender-disaggregated research methods
Phase 1: Exploration of sex and gender differences - Identify where sex and gender differences do (and do not) exist; - Always report sex-specific findings (with measure of variability); - Do not make conclusions on the presence (or absence) of sex differences based only on the sex-specific findings; - Quantify sex differences using a full interaction model that accounts for the possibility of sex-specific confounding |
Phase 2: Explanation of sex and gender differences - Exclude the artefactual explanation; - When evaluating sex differences in the associations of risk factors, consider both the absolute (risk difference) and relative (risk ratio) scales - Assess to what extent any sex or gender differences are due to differences in biology or due to different interactions with the healthcare system; - Use sex-specific Mendelian randomisation to strengthen sex-differentiated causal inferences; - Broaden the scope of research on the role of sex hormones |
Phase 3: Translation to policy and practice - Embed sex- and gender-inclusive medicine in the curriculum of health professionals; - Consider including sex-specific recommendations in guidelines; |
Systemic factors - Ensure that the participation of women and men in clinical trials, and medical research more broadly, is commensurate with the prevalence of the disease of interest in the population; - Funders and publishers of medical research should make the integration of sex and gender a requirement for funding or publishing; - Enhance the diversity in teams in research, policy, and practice, and address implicit biases against women |
Strengths | Limitations |
---|---|
The roadmap: - In three distinct phases, allows for a systematic evaluation of sex and gender differences in health and disease; - Provides practical guidance for researchers, policy makers, clinicians, and educators on how to explore and explain sex and gender differences in health and how to translate such findings to policy and practice; - Is generic and can be applied to a broad range health research areas; - Can be adopted to assess other aspects of intersectionality and gender identities | The roadmap: - Underscores that sex and gender exist along a continuum and are often intertwined, yet presents sex and gender as binary variables, to enhance coherence and accessibility; - Does not address the issue of how research into sex might differ from research into gender, or how the two might be researched together; - Has a quantitative focus without discussing the complex cultural and psychosocial concepts underpinning sex and gender; - Is a guiding document, which needs to be adapted to the research question and setting, or translational aim, at hand |
Phase 1: Exploration of sex and gender differences
Disease risk and prognosis
Disease presentation and diagnosis
Risk factor prevalence and associations
Diabetes is an important risk factor for a range of CVDs, regardless of sex. However, studies have consistently shown that the magnitude of that association in stronger in women [33]. Specifically, analyses in the UK Biobank showed that the adjusted hazard ratio for myocardial infarction associated with type 2 diabetes was 1.96 (1.60; 1.83) in women and 1.33 (1.18; 1.51) in men [34]. The corresponding women-to-men ratio of hazard ratios, as a measure of sex differences, was 1.47 (1.16; 1.87). In other words, the myocardial infarction conferred by diabetes is 47% greater in women than men. However, in absolute terms, the rates of myocardial infarction at a given age are lower in women than men, also in the presence of diabetes. Women lose some of their advantage, in terms of the risk of myocardial infarction, but do not surpass men |
Safety and efficacy of interventions
Provision and utilisation of healthcare services
Quantification of sex differences
Phase 2: Explanation of sex and gender differences
The artefactual explanation
Suppose that the 10-year disease risk in the absence of a risk factor (i.e. the reference group) is 1% in women and 3% in men. In other words, women have a third the risk of men, which — as mentioned already — broadly is the case for CVD (although attenuating with age). When the risk in those with the risk factor is 1% higher in both sexes, this results in a relative risk of 2/1 = 2 in women and of 4/3 = 1.33 in men. That is, women have a 2/1.33 = 1.5 times higher excess risk compared to men when they have the risk factor, even though the risk factor increases the risk by the same amount in both sexes. Thus, some would conclude that this implies that a finding of a higher relative risk in women is purely an artificial finding due to the lower background risk in women and the mathematical (statistical) metric used to compare the sexes |
The accessibility explanation
The biological explanation
A sex-specific Mendelian randomisation study based on data from the UK Biobank found no sex difference for the strength of the causal effect of genetic liability to type 2 diabetes on the risk of CHD [66]. This was in contrast with strong evidence from observational studies that consistently found evidence for a stronger association in women than men [34]. Another sex-specific Mendelian randomisation study showed that the genetically determined effect of BMI on the risk of type 2 diabetes was stronger in women than men [64]. It may therefore be that the sex differences in the association between diabetes and cardiovascular disease risk seen in observational studies actually occur before the actual diagnosis of diabetes. However, whether causal or otherwise, the higher excess risk seen in women with diabetes suggests a closer eye needs to be kept on them, and shows the importance of sex-specific risk scores |