The gender data gap in health research

Until the 1990s, most major clinical trials either excluded women outright or enrolled so few that the results could not meaningfully be broken down by sex, meaning much of the medical evidence still in everyday use was built almost entirely from male bodies (AlRahimi, 2026). The exclusion had a specific origin, the thalidomide tragedy of the early 1960s, after which regulators moved to protect women of childbearing age from experimental drugs by simply removing them from early-phase trials altogether. The intention was protective. The effect, compounded over decades, was a medical evidence base with a structural blind spot at its centre.

The scale of that blind spot is easy to underestimate because it does not announce itself. A trial that excludes women does not come with a warning label attached to the resulting guidance, the resulting dosing chart, or the resulting textbook description of what a disease is supposed to look like. It simply becomes the default, cited and re-cited until its origins in a narrower population are no longer visible to the people relying on it decades later.

Medicine's default research subject has long been male

A 2022 analysis spanning a century of biomedical publishing quantified how deep that blind spot runs. Studies including both sexes made up only around fifteen per cent of the literature across the entire period studied, and even as explicit reporting of subjects' sex increased in more recent decades, this partly came from more studies openly stating they used male subjects only, rather than a straightforward move toward inclusion (Zucker, Prendergast & Beery, 2022). Among the studies that did include both sexes, fewer than half went on to actually analyse their results by sex, meaning the data existed to ask the question and mostly was not asked. Some fields showed particularly stark imbalances: as recently as 2009, studies in neuroscience and pharmacology were around five times more likely to report on only male subjects than only female subjects. The same analysis found that decades of accumulated meta-analyses have thoroughly disproven the old justification that female biology is simply too variable, too complicated by hormonal cycling, to study efficiently, a claim that turns out not to hold up once it is actually tested.

The consequences are not abstract. The same 2022 review found a 96 per cent concordance between drugs known to have female-biased pharmacokinetics, meaning women process them differently, and drugs known to produce female-biased adverse reactions. Women, in other words, appear to experience more harm from medications tested and dosed largely on male physiology, a pattern that follows directly and predictably from the research gap rather than sitting apart from it.

The funding gap is not just historical

If the exclusion itself was largely a story of the twentieth century, the funding gap that followed it remains a live, current problem. A 2025 review by the National Academies of Sciences, Engineering, and Medicine found that women's health research received 8.8 per cent of all US National Institutes of Health grant spending between 2013 and 2023, a share that had fallen further, to 7.9 per cent, by the final year of that period, despite women making up roughly half the population the agency's research is meant to serve (National Academies of Sciences, Engineering, and Medicine, 2025). The report was explicit that conditions disproportionately affecting women, several of them named directly, including endometriosis and polycystic ovary syndrome, remain funded well below the scale of the burden they cause, and argued this shows a longstanding institutional default in which "male bodies are the standard and treated female bodies as an exception."

This matters well beyond the specific conditions named in the report. A funding gap of this size shapes which questions ever get asked in the first place, and a question that is never funded is a question that never generates the evidence that would let clinicians, researchers or women themselves understand what is actually happening in a female body across a normal midlife decade. Funding decisions made years or decades ago are still shaping the textbooks, guidelines and default assumptions clinicians work from today, which means a gap that opened in the 1970s or 1980s continues to have consequences for a woman sitting in a consultation room now, long after the original funding decision has been forgotten by everyone except the evidence itself.

A cardiovascular case study shows what the gap costs in practice

Heart disease offers one of the clearer illustrations of what an evidence gap costs once it reaches everyday clinical care. Women still comprise only twenty to forty per cent of participants in cardiovascular clinical trials, despite heart disease being a leading cause of death in women as well as men, and a 2026 review found that women continue to be underdiagnosed, undertreated and underrepresented in the research base that should be informing their care (AlRahimi, 2026). Part of the explanation is that women more often present with less classic symptoms, fatigue, breathlessness or discomfort in the upper abdomen rather than the crushing chest pain more commonly described in male-derived clinical descriptions, which increases the risk that a genuine cardiac event is initially read as something else. This is not a story about individual clinicians failing individual patients so much as a story about a recognition framework built substantially from one sex's presentation of a disease that affects both.

The gap is not only clinical, it runs through everyday life too

The Observatory's interest in this subject extends past the clinical trial and into the texture of an ordinary week, because measurement gaps of this kind are not confined to hospitals and laboratories. Analysis of English and Welsh data found that 57.7 per cent of unpaid carers were women, with the gap between women and men providing unpaid care widest in the fifty-to-sixty-four age band (Office for National Statistics, 2013), a form of daily, cumulative load that sits almost entirely outside conventional health statistics. A separate systematic review of the "mental load" of household planning and organising, drawn from thirty-one peer-reviewed studies, found the same pattern in the invisible, cognitive dimension of domestic work: consistently higher in women, narrower only where both partners work full time, and linked to stress and lower life satisfaction, yet rarely captured by any dataset built to measure health or the economy in the conventional sense (Reich-Stiebert, Froehlich & Voltmer, 2023). Even burnout research, a field that does explicitly study both sexes, has struggled to settle a clear picture of sex differences, a large meta-analysis finding only small, mixed effects across 409 comparisons, a result its own authors read as evidence against, not for, simple assumptions about who burns out more (Purvanova & Muros, 2010).

Put together, these gaps describe a consistent shape rather than a coincidence: the clinical evidence base was built mostly around male physiology, the funding that might narrow this remains disproportionately small, and the everyday, non-clinical work that fills much of women's time is barely measured by anyone at all. Each gap alone would be a limitation worth naming on its own terms. Together, spanning the laboratory, the funding committee and the kitchen table alike, they describe a research and measurement system with women's ordinary lives sitting largely outside its field of view, not through any single decision but through a long accumulation of smaller ones.

What the Observatory observes

None of this is a case for distrusting medicine or research broadly; it is a case for being precise about where the evidence is genuinely strong, where it is thinner than commonly assumed, and where it barely exists at all. The Observatory's own method, inviting members to record simple, longitudinal observations of their own sleep, energy, stress and everyday context, is a direct response to this specific gap, an attempt to build at least one small, consented, well-governed dataset that centres the kind of daily female experience the wider research system has been slow to prioritise.

This is described honestly as a contribution rather than a correction; one membership organisation's dataset cannot resolve a funding gap measured in tens of billions of dollars, and the Observatory does not claim otherwise. What a longitudinal, consented record can do is generate real, if modest, evidence about patterns in real lives, gathered on terms the people living those lives actually agreed to. For a closer look at where the everyday version of this gap shows up most clearly, see the female data gap in everyday life; for the case for longitudinal evidence specifically, see why women need longitudinal evidence.

The seven ceramic faces on a wood-panelled wall, each wearing a different expression, are a reasonable image for what good research on women's health should look like and mostly has not: attentive to difference, unwilling to flatten one experience into a stand-in for all of them. Building a dataset that can actually see that difference, over years rather than a single sitting, starts with the same small act this whole publication keeps returning to: observing one's own pattern, honestly, and adding it to a record built to be read with care.