Q&A: Is it time for scientists to rethink how they study sex in research?

Our understanding of sex has come a long way. Aristotle theorized that heat during conception determined whether someone was born male or female. But while researchers have long moved on from that idea, scientists in a new issue of Cell published this week argue that our ideas about sex — and how to best measure it in research — remain crude and outdated.

STAT spoke with Madeleine Pape, a sociologist of gender at the University of Lausanne in Switzerland and lead author of a perspective about the need to more clearly define sex in research. She and her co-authors argue that “sex” isn’t the most accurate or useful variable in research; it’s a category made up of other characteristics. And it’s these more measurable related factors — such as the presence of a uterus, hormone levels, and genetics — that researchers should track. Exactly which factors they track should be based on the context of a given study.

“We encourage researchers and policymakers alike to strive for new standards of rigor and precision in the consideration of sex in biomedical research,” the article’s authors wrote.

The conversation has been edited for length and clarity.

In your Cell piece, you write that sex is much more complex than we currently imagine. How is sex thought of in research and medicine?

It’s not only in research, actually, but even in everyday usage. We often talk about sex as though it’s an actual part of the body that can be located, that can then have an effect on that individual’s health. Actually, sex is something that we assign as a set of categories. We think it’s a good proxy for a range of biological characteristics or mechanisms, but sex in and of itself isn’t something that has a physical form. So then there is an added step that needs to happen if we want to start talking about sex, or sex-related factors, having an effect on some part of the body or health or illness.

When we start talking about sex being complex, people assume that we’re talking about sex and gender minorities like trans people, gender-diverse people, people with intersex variations. And, yes, we are talking about those people. But we’re also talking about cisgender bodies. Women aren’t a one-size-fits-all. There’s a great deal of variation among women for all kinds of reasons. And that variation is worth understanding and worth engaging with if we want to actually address gender disparities in health.

One example of this in the piece is that using the category ‘‘women’’ as a proxy for the presence of a uterus actually underestimates the incidence of uterine cancer by 23%-53% and racial disparities in cervical cancer by 44%. Those are big numbers. How exactly does that happen?

In that example, someone has taken the gender identity category assigned to individuals in a study as a proxy for the presence or absence of a uterus, rather than actually use the presence or absence itself as the factor to identify risk for that individual.

In clinical settings, there’s a strong tendency to rely on categories to quickly make judgments about an individual and what their particular risk factors, biological characteristics, or other characteristics might be. But it’s not necessarily going to be precise. So if we want to aim for precision, we need to go further than relying on proxies, toward thinking more carefully about, what are the actual traits that could be relevant in this instance?

Where does this misuse of sex in research come from?

For a long time, when it came to clinical trials, there was a tendency or widespread perception that research was really relying on white male, middle-class, straight subjects. And it took the NIH Revitalization Act in 1993 to enforce more of a gender balance onto clinical research. That tendency to treat the male model as the universal model was also seen in many areas of animal research.

It’s a very interesting and tough conundrum in some ways, because the problem is, once we start to enforce the addition of female models in animal research, there’s a tendency to assume that the point of including the females is to understand differences between the groups. What we’re really trying to do is correct this male-centric tendency and ensure that we have full diversity in our samples, and that if we’re going to look for sex differences, that we’re doing so in a way that is hypothesis-driven and that the sample has been chosen to reflect a particular research interest.

What would it look like practically for researchers to move beyond the current system of using sex as a proxy for more specific variables?

Researchers should use their own expertise and knowledge about the factors that could be important to the phenomenon that they’re studying to determine how they’re going to operationalize sex in that context.

Take the uterine cancer example, for instance. It could be that it’s something related to the mechanism of hormones, in which case, we can define that. That’s something we can measure in research animals or in human participants. And there we can see considerable variation across not just women and men, but also within women and within men. That’s one of our key points that gets lost. There is a tendency to flatten the two categories. Our way of analyzing and drawing conclusions about sex-related variation tends to focus on comparison of the means of the two groups and seeing if there’s a significant difference. That’s only one way of describing the distributions that you see. Wouldn’t it be more interesting to not only analyze means, but actually be able to say something meaningful about the distributions that occur in the data, which will include some overlap very often between the female group and the male group?

Is it realistic to expect all clinicians and researchers to start being more precise about referencing sex-related factors rather than sex in this way?

It is a practical challenge because people who are working in a clinical setting face a whole different variety of constraints compared to researchers that are working in basic science. Nevertheless, we have potentially much to gain in clinical settings from being able to bring more precision to the ways that we’re making judgments on the basis of either sex categories or sex-related factors.

People tend to think in terms of black and white. As soon as you put something on the table that sounds more complex, they nevertheless try to sort you into either black or white. So in this case, what we’ve written could very easily be categorized as being anti-sex differences research, for example, rather than in favor of a more complex way of studying sex-related variation, which is what I think we’re advocating for.