25 years of women being underrepresented in medical research, in charts

Ideally, women’s participation in research about a condition would reflect the proportion of people with that condition who are women.
Ideally, women’s participation in research about a condition would reflect the proportion of people with that condition who are women.
Image: Reuters/Yuriko Nakao
We may earn a commission from links on this page.

The issue of gender bias in research is not a new problem, but it is a persistent one.

Today (July 3) researchers from the Allen Institute for Artificial Intelligence in Seattle published work looking at women’s participation in medical research over 25 years, from 1993 to 2018. They analyzed over 43,000 research studies in PubMed, a searchable database of biomedical science, and 13,000 clinical trials registered on

Overall, women made up 49% of all participants across all studies—so far, so good. But when the researchers looked at studies condition by condition, they found that women were often underrepresented. For many disease types, the proportion of female participants didn’t match the gender breakdown of real-world patients.

Five conditions showed especially dramatic discrepancies: cardiovascular disease, hepatitis, HIV, chronic kidney disease, and digestive disease.

Clinical research on diabetes, mental health, non sex-specific cancers, and respiratory disease had roughly equal representation.

The only two categories in which women were somewhat overrepresented were neurological conditions and musculoskeletal conditions. Presumably, this means that men were underrepresented in this kind of research.

This analysis agrees with previous research into gender bias in research, though at an expanded scale. The researchers used a system called Semantic Scholar, which uses artificial intelligence to comb through thousands of scientific articles. Previous work, relying on manual data entry, has been limited to smaller sets of studies. But both approaches show how difficult gender bias has been to eliminate.

Historically, doctors and scientists used relatively thin men to stand in for most patients. Michelle Berlin, a physician and the director of Oregon Health and Science University’s Center for Women’s Health, recalls being taught to base physiological assumptions on a male patient in the early 1980s. And yet clearly, “children are not small men, and women are not small men, either,” she says. 

So in 1993—the year in which the oldest studies in this analysis were published—US Congress passed the National Institutes of Health Revitalization Act. It states that women and people of color have to be included in government-funded clinical research. The law recognized that gender bias in research existed. Yet it didn’t mandate real-world benchmarks for women and minority research participation.

Additionally, the Revitalization Act only applies to federally-funded research. The majority of clinical trials are funded by the pharmaceutical industry, says Melina Kibbe, a vascular surgeon at the University of North Carolina who has studied the underuse of female animal models in pre-clinical research. The US Food and Drug Administration, which approves pharmaceutical products, requires that women are included in research, but doesn’t require that their participation is proportional to the burden of disease.

Failing to study women can have serious health consequences. Consider heart disease: Women are historically underrepresented in cardiovascular research because they often have different symptoms than men. For women, a heart attack could feel similar to indigestion—not pain in the chest. This bias is at least partly responsible (pdf) for the fact that women are less likely to survive a heart attack, particularly when treated by a male doctor.

Women have also been the victims of unforeseen bad reactions to certain drugs. In a span of three years from 1997 to 2000, 8 of the 10 drugs for which the FDA withdrew its approval had harmful side effects for women. In 2013, the FDA recognized that the active ingredient in Ambien took longer for women to process than men, potentially leaving women dangerously groggy in the morning. Now the agency recommends that women take half-size doses.

To change the system, researchers need to address the social barriers that can discourage or prevent women from participating in clinical trials. Health care providers don’t always take women’s symptoms as seriously as they do men’s, meaning they may not be considered sick enough to participate in a clinical trial. Additionally, because women are often caregivers, they may have extra time and transportation constraints that limit their ability to return for follow up visits, Berlin says.

By continuing to point out gender bias in research, the hope is that researchers will start to change their practices, says Waleed Ammar, one of the co-authors of the paper, published in the journal JAMA Network Open. If researchers don’t, medical science won’t benefit everyone.