Cherry-Picked Testosterone: The Dangerous Game of Social Media Science

Testosterone, Dementia, and the Abuse of Observational Studies

Uwe Krejci/Getty Images

I am regularly sent videos of medical menopause influencers talking about some important new study (or older study, for that matter). The video is often the influencer in front of a screenshot of the abstract emphatically explaining the findings as if this study is the game changer. But then, when I read the study and check the state of the literature, I am often disappointed. The studies are often low quality, or not designed to reach that bold conclusion. Cherry picking is also all too common, meaning one study is promoted, but the others that refute it are ignored. I've even seen influencers report findings that weren’t in the study. And possibly the most jaw-dropping (and not in a good way) was seeing research that was misrepresented, and then when one of the authors of the paper challenged the influencer explaining that paper was a small exploratory study and the results did not held up in a subsequent, larger study, the influencer doubled down! I’m left wondering how many influencers just read abstracts and when they find one that they can use to suit their needs, they hit record on their social media platform of choice?

For example, my friend Amanda Thebe, who is an expert in women’s fitness, addressed some bold claims from a couple of menopause influencers about weighted vests supposedly improving bone density. These influencers apparently used a study of 18 women, nine of whom wore weighted vests AND did an exercise program that included jumping. (I am sure you are thinking right about now that a study of 18 people, nine who did the intervention as well as a special exercise program is exploratory at best, and nothing that should be used to direct medical care, and you would be right!). This was an observational study, and so the women were not randomized. The control women were active, but not assigned a specific exercise program. Also, some of the women were taking menopause hormone therapy, but the doses and types were not mentioned. The weighted vest/exercise group lost 0.84% of the bone mass at their hip over five years versus 3.8% for the control group that did not exercise. The weighted vest plus exercise group did gain a small amount of bone mass (1.54%) in one part of the hip (femoral neck). It is not possible to draw any conclusion about meaningful benefit for weighted vests from this study. There are too few participants, it is observational, and the women wearing the weighed vests also participated in a jumping program, which can improve bone density! However, the study seems legit when people with thousands or even millions of followers promote weighted vests using the study as “proof.” I can see how the average person who doesn’t know the literature would think this was something validated by science instead of what it really is, a way to get attention online and probably drive Amazon affiliate sales. Also, if you are not following Amanda on Instagram, you should. She is very evidence-based and loads of fun (here is a link to her account).

I joked on Amanda‘s post that I should start addressing studies that get misused and abused on Instagram, and many people replied, ‘Yes, please.” I didn’t think much of it…until the very next day when I was sent a link to a video with with a menopause influencer claiming a new study adds more support to the idea that women in menopause should be taking testosterone to support brain health and prevent dementia.

I was intrigued…and as soon as I saw the study, I sprained my eyes after rolling them back into my head (this seems to be a growing workplace injury for me). Look, it’s hard not to be snarky here, because as you will soon see, making broad claims about anything for women based on this study is wrong. It’s barely a hypothesis-generating pamphlet at best, but how would the public know this?

Not all studies are useful in helping us with patient care, as we’ve just seen with the 18 person weighted vest study. Some studies are interesting, but essentially useless for the here and now, but they may be of value to help direct future research. But even good studies about clinical care often have limitations. And of course, it’s always important to put every study in context with the body of literature. But explaining these complexities and nuances don’t help a video go viral. With that in mind, I think this testosterone study is a good selection for a mini-journal club, so you can see why this annoys me so much, and it may help you be protected from future misinformation, especially about testosterone.

The study is called, “Sex Hormones and Risk of Incident Dementia Risk in Men and Post Menopausal Women” (May 2025). It is a study from the UK Biobank, an extensive repository of information and biological samples. Researchers looked at testosterone levels and levels of sex hormone binding globulin (SHBG) for men and women, and then calculated free testosterone, which is the amount of testosterone that is not attached to carrier proteins. The investigators also evaluated the records to see who developed dementia and who did not and then they looked for a connection between testosterone, free testosterone, SHBG and dementia. We won’t discuss the results for the men, because I only have a personal interest in one man, not a professional interest in men in general.

The results for the women indicated that higher levels of SHBG and lower levels of free testosterone were associated with a higher incidence of dementia. There was no association between total testosterone levels and subsequent dementia. In their conclusions, the authors state, “but further studies must be done to determine causality.” Meaning we should not draw a line between cause and effect. After all, this is an observational study of hormone levels produced by the body, not a study that looks at testosterone therapy and the risk of dementia.

Meaning, the findings are really hypothesis-generating at best.

Except there are quite a few issues with this paper, and when you understand them, I think you’ll be pretty shocked that anyone would use this paper to support any hypothesis. So let’s get into it!

The Method of Measuring Testosterone

This study used a chemiluminescent immunoassay, which is not the most accurate way to measure testosterone levels for women, because it does not perform well at the low levels that are typical for women. The authors even address this point as a study limitation! They state:

“…the chemiluminescent immunoassay method may not be accurate at lower serum hormone levels.”

It seems to me that having the proper test seems…important. One might say critical, even.

This is why I rolled my eyes, because I know the U.K. Biobank data can’t really be used to draw conclusions about testosterone levels for women.

But wait, there’s more!

Free Testosterone Is Not Be the Best Measure

Assuming, for the sake of argument, that the testosterone levels from the test are accurate. The biological significance of free testosterone, which is the testosterone not on carrier proteins, has not been determined, in part due to the complex intracrinology of testosterone (you can read more about that here). Total testosterone is now believed to be a better measurement, so I care much less about any link with free testosterone, what I care about is total testosterone, and in this study, total testosterone levels were not associated with dementia.

We Can Not Draw Sweeping Conclusions from Observational Studies

Observational studies are rarely, if ever, used to change clinical decisions with medications. They can be useful for generating hypotheses, which means the conclusions require further testing. For example, hypothetically, if several studies using a more accurate method of detecting testosterone were to find that women with higher total testosterone levels were less likely to develop dementia, that would support investing in the next step of research. This study does not support that next step.

We also cannot assume that levels produced normally in the body reflect what happens with a medication, which is another issue with these kinds of observational studies. What if a study using the right test for testosterone found that lower total testosterone levels are associated with a higher risk of dementia? One conclusion could be that low levels of testosterone cause dementia. However, an equally plausible conclusion is that the reverse is true and dementia causes low levels of testosterone. It is also possible that low levels of testosterone are a marker for a health behavior or another factor that is associated with dementia, such as alcohol use or lack of exercise. Observational studies of blood tests cannot inform us about cause and effect with a medication.

Another issue with the study is that women self-reported if they were on menopause hormone therapy with a yes or no. But there are different doses and formulations and that might affect risk of dementia. Oral estrogen can raise SHBG, and not only was SHBG one of the measurements in the study, SHBG is used in the calculation for free testosterone.

A Better Study Looking at Testosterone Levels Was Also Just Published

As we do not pick cherries in this Substack, let’s discuss another study that examines testosterone levels, cognition, and dementia. It was published two months before the study we’ve just been discussing.

This study, “Associations between blood sex hormones, cognitive decline and incident dementia in community-dwelling older Australian women: a prospective cohort study,” (March 2025) looked at the levels of DHEA, estrone, and testosterone in blood using a gold standard test: liquid chromatography–tandem mass spectrometry (LCMS). This study enrolled thousands of women aged 70 and older, measured hormone levels, and performed cognitive assessments at baseline, year 3, and year 5. Only women with no evidence of cognitive decline at enrollment were included. Unlike the UK Biobank, this study excluded women taking medications that could potentially affect hormone levels.

The findings? The highest testosterone levels, when compared with the lowest, were associated with a greater decline in executive function and verbal fluency, although this was not statistically significant. Testosterone levels were not associated positively or negatively with any other cognitive measure or with dementia. The authors appropriately conclude, “Any explanation for our findings is highly speculative.”

This paper generates some hypotheses that need more exploration, but it would be inappropriate to conclude anything about pharmaceutical testosterone and the brain based on this study.

As we don’t have good data to show benefit for testosterone therapy for cognition or brain health, and we’re just learning what is typical for levels, using testosterone outside of its indications is not recommended. When we don’t know, we need to be careful, because there are so many hypotheses to consider. For example, we now know that testosterone levels rise for women in their seventies. What if giving testosterone therapy blunts that normal rise, but that rise in hormone was beneficial and the pharmaceutical product doesn’t achieve the same benefit? This is simply a hypotheses, so I’m not saying that is a concern, but it’s the type of thing that needs to be considered in studies. We need to be careful making sweeping conclusions about testosterone considering we’re really at the early stages of understanding levels and what they may or may not mean.

I know hyping testosterone outside of guidelines is popular among a certain menopause influencer niche on social media, and it’s really a shame to see. Certainly posting about testosterone gets lots of attention, and perhaps that matters to some, but good data matters more to me.

Five Simple Steps to Assess Social Media Claims About Testosterone Studies

When people wave study abstracts around, making bold claims, it can be easy to mistake noise for nuance. Remember, not every study deserves a headline or a post on Instagram. Also, science is not a single paper; it’s a process of discovery and a body of knowledge that is constantly growing, and some studies contribute more to that growth than others, and anyone discussing a study should really explain how it fits with the body of knowledge.

To help you decide if the next study on testosterone deserves your attention, or an eye roll, consider these five steps:

1. Is the study also being covered by the traditional media? This requires lateral searching, meaning getting off the social media platform to look. If a study about testosterone were truly ground breaking, it would be covered by the press.

2. How were the testosterone levels measured? If it is with an immunoassay, the study is not of much value and really isn’t worth discussing.

3. Is the study making claims about total testosterone or free testosterone? We believe total testosterone is the important measurement.

4. Is this an observational study or a clinical trial? Only clinical trials can inform us if a medication achieves a desired outcome.

5. What did the authors actually conclude? Did they say this study can inform clinical care (how we prescribe) or, did they caution about conclusions? If they cautioned about conclusions, we should take them at their word!

Critical thinking isn’t just for scientists—it’s self-defense in the age of Instagram medicine.

References

Snow CM, Shaw JM, Winters KM, Witzke KA. Long-term exercise using weighted vests prevents hip bone loss in postmenopausal women. J Gerontol A Biol Sci Med Sci. 2000 Sep;55(9):M489-91. doi: 10.1093/gerona/55.9.m489. PMID: 10995045.

Zhao Y, Wang Q, Fu C, Li M, Hao W, Wang X, Song Q, Zhu D. Sex Hormones and Risk of Incident Dementia in Men and Postmenopausal Women. Clin Endocrinol (Oxf). 2025 May 11. doi: 10.1111/cen.15271. Epub ahead of print. PMID: 40350794.

Sultana, F., Davis, S. R., Wolfe, R. S., McNeil, J. J., & Islam, R. M. (2025). Associations between blood sex hormones, cognitive decline and incident dementia in community-dwelling older Australian women: a prospective cohort study. Climacteric, 1–10. https://doi.org/10.1080/13697137.2025.2470458

Goldman AL, Bhasin S, Wu FCW, Krishna M, Matsumoto AM, Jasuja R. A Reappraisal of Testosterone's Binding in Circulation: Physiological and Clinical Implications. Endocr Rev. 2017 Aug 1;38(4):302-324. doi: 10.1210/er.2017-00025.

Davis SR, Barber R, Panay N, et al. Global Consensus Position Statement on the Use of Testosterone Therapy for Women. J. Clin Endocrinol Metab 2019;104:4660.

Davidson SL, Bell, R, Donath JG, Davis SR. Androgen Levels in Adult Females: Changes with Age, Menopause, and Oophorectomy. J Clin Endocroinol Metab 2005; 90:3847–3853

Somboonporn W, Davis S, Seif MW, Beel R. Testosterone for peri- and postmenopausal women. Cochrane Database Syst Rev 2005;CD004509.