October 21, 2025

expert reaction to conference abstract on the association between hormone therapy and autoimmune disease risk

A conference abstract presented at the 2025 Annual Meeting of The Menopause Society looks at an association between hormone therapy and the risk of autoimmune diseases.

Prof Kevin McConway, Emeritus Professor of Applied Statistics, Open University, said:

“It’s always frustrating for me to comment on a conference abstract like this, from my professional point of view as a statistician interested in research methods and quality. All we have so far is a one-page Abstract (summary of the research and its findings) and a fairly brief press release, which in this case does correspond pretty well to the abstract. Together, these just don’t tell us anywhere near everything I need to evaluate the research properly.

“If and when a full research paper is published about this study, it will be much longer, and contain much fuller details on the research methods that the scientists used, as well as more detail on the findings and more on the background to the research. It would also have gone through a peer review process. Peer review does not always pick up all the problems with a piece of research, but it does often lead to a final paper that covers the research properly, making clearer its strengths and limitations.

“I need most of that detail to make a proper assessment of whether the top-line research findings are actually supported by what the researchers really did.

“But so far I have none of that. The abstract has been through a selection process for the conference, but that couldn’t be like a proper peer review, because the selectors would have seen no more than the abstract itself. I’m not trying to run down the researchers who did the study here – just to point out that it’s impossible to know, from just an abstract, how good the research actually is, and therefore what its impact is.

“On the basis of the limited information I have, I can say that on the face of it, the researchers do seem to have done the right kind of thing to evaluate whether the use of hormone therapy by postmenopausal women is associated with a change in their risk of autoimmune diseases. But first impressions on the basis of incomplete information can be wrong.

“I think the researchers are correct to point out that this issue is, so far, poorly understood. There has been some research on possible effects of hormone therapy on women who already had an autoimmune disease, and at least one much broader study covering a wider age range and looking at several different types of hormone supplementation, not just for menopausal symptoms. But those studies have a different focus from this new one.

“There is one important issue that limits seriously what we can conclude from this new study. However competent the study turns out to be when the full paper is available, we just can’t conclude with any degree of certainty that differences in the risk of autoimmune diseases between women who use and don’t use hormone therapy are actually caused by whether they use hormone therapy or not. They may be, but it remains possible that they aren’t, or at least not fully.

“If it eventually turns out that the risk differences aren’t caused by whether women used hormone therapy, then the study can’t tell us anything about the decision as to whether to use hormone therapy. The issue is that the study is observational. I’ll explain more below.

“That’s why the conclusion in the abstract, and the quote from Dr Jiang in the press release, both call for more research, using different research methods. Some of that research would still be observational, but prospective, following women forward from the time of their menopause, rather than looking backward at records after they may have been diagnosed with an autoimmune disease. As Dr Jiang states, that could throw more light on what happens when in different women, and possibly on the processes inside the body by which hormone therapy could be involved in increasing the risk of autoimmune diseases.

“But this current piece of research can’t tell us about that – that’s all for the future, if and when the proposed new research happens. It’s unlikely to be quick or cheap.

“For now, it’s important to note Dr Jiang’s statement in the press release that, even if the increase in risk eventually turns out to be caused by the hormone therapy, “the actual increase in absolute risk is relatively small” and varies between different autoimmune diseases. For individual women, there will be a trade-off between the benefits of hormone therapy during and after menopause and any increase in autoimmune disease risk. But because we can’t yet be sure about what’s causing the risk differences, it’s hard at this stage to evaluate exactly how that trade-off might work in individual people

“The reason for the doubt about what causes what is that this is an observational study. It’s not like a randomised drug trial, where some women are chosen at random by the researchers to have hormone therapy, while other women do not have it. The women, and their health professionals, just did what they would have done anyway in terms of the therapy, and that’s recorded in the TriNetX databases that the researchers used.

“Inevitably there would have been differences between women who used hormone therapy, and women who didn’t, in terms of other factors than their use of hormone therapy. If some of those other factors are also correlated with differences in the risk of autoimmune diseases, it could be differences in these other factors that are causing the differences in autoimmune disease risk, rather than whether the women used hormone therapy or not.

“This situation, where one can’t be sure whether a risk difference is caused by the factor under investigation, hormone therapy, or by some other factor, is known as confounding, because the possible effects of hormone therapy and of other factors are inevitably mixed up (or confounded) together

“The researchers on this study were, as you’d imagine, well aware of this issue, and so they used a statistical technique called propensity score matching. This involves setting up a statistical model that predicts the chance that a woman will use hormone therapy, on the basis of a lot of other factors that might possibly confound what’s going on and lead to biases. Then the women were divided into pairs. In each pair, both women had very similar predictions of the chance that they would use hormone therapy, but in fact one of them did use it and the other did not.

“The procedure can take some of the sting out of possible confounding – but it can’t remove entirely the possibility that the risk differences are caused by something other than whether they used hormone therapy or not. The statistical modelling that produces the propensity scores can only be based on data that the researchers had available and considered to be relevant.

“The abstract only gives brief details on how the propensity score matching was done, so at this stage I can’t tell whether all the potentially relevant factors were included in appropriate ways. In some circumstances, if important factors aren’t included in the propensity scoring, perhaps because the data just is not available, the process could even increase bias rather than reducing it.

“Actually, even when the full research paper is available, some doubts about exactly what causes what will remain. That’s an inevitable problem with an observational study like this, particularly a retrospective one. It could be that there are differences in the risk of autoimmune disease that are actually caused, in whole or in part, by whether a women used hormone therapy or not. But it still could be that the risk differences are actually caused by other factors.”

Dr Hugo Pedder, Senior Research Associate in Statistical Modelling at the Bristol Medical School, University of Bristol, said:

“The study provides an interesting observation that postmenopausal women who use hormone therapy may have a higher recorded incidence of autoimmune diseases. However, the findings are based on a large observational dataset reported in a conference abstract, and the study has not yet been peer-reviewed. This means the results should be interpreted with caution.

“Because this is not a randomized trial, we can’t assume a causal link between hormone therapy and autoimmune disease. There are likely many other factors that could influence both hormone therapy use and the likelihood of receiving an autoimmune diagnosis. For instance, women who seek treatment for menopausal symptoms may be more engaged with healthcare and therefore more likely to have other conditions detected.

“Additionally, hormone therapy is more likely in women with greater menopausal symptom severity, but it could be possible that the severity of these symptoms is impacted by an underlying/nascent autoimmune disorder. It may be that underlying disorders influence both the use of hormone therapy and the eventual autoimmune diagnosis.

“The apparent increase in risk is statistically significant but relatively modest in size. Large sample sizes can yield precise estimates, but that does not necessarily make the result robust or clinically meaningful, and the risk of bias from such a study design remains high. Hormone therapy remains an important and effective treatment for many women suffering from menopausal symptoms, and these findings should not deter appropriate use pending further, well-controlled research.”

The conference abstract ‘Association of Hormone Therapy with Autoimmune Disease Risk in Postmenopausal Women: A TriNetX-Based Analysis’ by Ameera Syed et al. was presented at the 2025 Annual Meeting of The Menopause Society in Orlando, October 21-25. The embargo on the abstract lifted at 05:01 UK time on Tuesday 21^st October.

Declared interests

Mr Hugo Pedder: I am co-director of the NICE Guidelines Technical Support Unit, and have worked on the NICE guideline for menopause. Until 2023 I sat on the Medical Advisory Board of the British Menopause Society. Within this role I published papers on the risks and benefits of hormone therapy for menopause. I am an employee of ConnectHEOR, and in this role have worked with Roche and MSD within the last 12 months.

Prof Kevin McConway: I have no interests to declare about this.