Testosterone in perimenopause: what the data shows

Testosterone has become the midlife hormone of the moment, offered for energy, focus, motivation and libido. The research behind it is narrower than the marketing, and on one point it runs in the opposite direction: across the menopause transition, total testosterone stays broadly flat while estradiol falls away. The decline people describe when they say their testosterone crashed largely happened earlier, in their thirties. And nothing worn on the wrist measures any of it.

Why “my testosterone crashed” doesn’t fit the cohort data

Two pieces of evidence sit awkwardly beside the popular story.

The first is old and small. In 1995, researchers measured 24-hour mean plasma testosterone in 33 healthy, regularly cycling women between 21 and 51. Total testosterone fell steeply with age — the expected value for a woman of 40 was about half that of a woman of 21 — and because the free fraction did not change much with age, free testosterone tracked the same slope. Whatever decline there is, most of it is spent before perimenopause begins.

The second is large and longitudinal. In SWAN, 1,862 premenopausal and perimenopausal women were followed through 6,296 woman-years, and the paper opens by stating the position plainly: during the menopause transition total testosterone remains unchanged, while estrogen decreases markedly, creating a state of relative androgen excess. The age-adjusted testosterone-to-estradiol ratio rose by about 10% per year over five years of follow-up.

So the transition does change something — the balance between two hormones — but the moving part is mostly the estradiol. The consensus statement written by eleven professional societies says the same thing in one line: testosterone concentrations decline during the reproductive years, and appear to be maintained in women beyond 65.

This is baseline-not-average in an unusually literal form. A level drawn at 47 gets compared against a population range, when the only comparison that would mean anything is you at 30 — a measurement almost nobody has.

There is no blood test that can call you low

This is the part that surprises people, and it comes straight from the guideline rather than from us.

The consensus statement notes that direct assays for total and free testosterone are highly unreliable in the female range. Mass spectrometry methods are accurate, but they are not what most laboratories run for a routine request. Then it goes further: no cut-off blood level for any circulating androgen can be used to separate women with and without sexual difficulties.

That conclusion has a source behind it. A community-based study of 1,021 Australian women aged 18 to 75 looked for exactly this link and did not find one — no clinically meaningful relationship between low self-reported sexual function and low total or free testosterone, and the authors concluded that no single androgen level predicts it.

Measurement still has a job in this picture, but it is the opposite job from the one people expect. The guideline uses testing to exclude high levels before treatment and to keep concentrations from running above the female range during it. That is a safety check, not a diagnosis — the same number doing entirely different work.

What the trials found, and what they did not

The meta-analysis underpinning the guideline pooled 36 randomised controlled trials and 8,480 participants. In postmenopausal women, testosterone increased satisfying sexual event frequency by an average of about 0.85 events per month (95% CI 0.52 to 1.18), along with desire, arousal, orgasm, pleasure and self-image, and reduced sexual distress. That is a real, replicated, specific effect.

The null results are rarely quoted next to it. Pooled data showed no effect of testosterone therapy on general wellbeing, and no effect on depressed mood. Evidence was judged insufficient to support using it for cognitive performance. No effect was demonstrated on bone density at the spine or hip at 12 months, or on lean body mass, total body fat or muscle strength. Acne and mild body or facial hair growth were more common; serious adverse events were not seen in trials at female physiological levels, and safety data do not extend beyond 24 months.

None of that says nothing happens for anyone. It says the studies capable of showing an energy or focus effect have either looked and found nothing, or have not been done. Those are different situations, and both are worth naming honestly before deciding anything.

The evidence stops just short of perimenopause

Every trial in that meta-analysis recruited postmenopausal women. The consensus statement is explicit about the other side of the line: there are insufficient data to make any recommendation about testosterone in premenopausal women, for sexual function or for any other outcome. Its single evidence-based indication is postmenopausal low sexual desire causing distress, identified through a full biopsychosocial assessment.

Most of perimenopause sits on the under-studied side of that boundary. That is not a verdict on anyone’s decision — it is a description of where the map ends, and useful to have in mind when trial numbers are quoted to someone who is still cycling.

The practical backdrop matters too. In most countries no testosterone formulation is approved for women, so prescribing is off-label from male products, and the guideline does not recommend compounded preparations where an approved equivalent exists. If you are weighing this alongside estrogen or progesterone, our notes on what changes when HRT starts and on stopping cover the same measurement problem from other angles.

How Perigee reads it

No wrist sensor measures testosterone, or any androgen. Not heart rate variability, not wrist temperature, not sleep. Any app implying otherwise is making a claim about hormones that consumer hardware cannot support, and we would rather say so plainly — you can read what we do and do not claim and the evidence behind each signal.

What a dated timeline does add is the comparison that is otherwise lost. Guidance to clinicians includes reviewing at around six months, which is long enough for memory to smooth everything out. The two to four weeks before anything starts is the window most people skip and later wish they had.

One honest caveat: testosterone is usually added alongside estrogen therapy, and most trials included women already taking it. A single before-and-after cannot separate two changes made in the same week. Staggering them can, and that is a conversation with your clinician rather than something your data can settle. Meanwhile sleep, resting heart rate and temperature keep moving for their own reasons — a night sweat you slept through leaves a mark too.

One small thing

Before an appointment about testosterone, write down the one or two things you would actually want to be different, in your own words. Then rate each of them 1 to 5 every day for two weeks, before you look at anything on your watch.

The outcome with the strongest evidence behind it is something you report, not something a sensor captures — which makes that list the measurement, and the dates on it the part worth bringing to an appointment.