One Hormone Draw Cannot Tell a Woman She Is in Perimenopause

She is forty-seven, and the last four cycles arrived at twenty-three days, forty-one days, twenty-six days, and then not at all. She wakes at three most mornings with the sheets damp. By eleven at night, after the house has gone quiet, she is on her phone trying to find the name of what is happening to her, and the search returns an offer almost immediately: a panel of hormones, a finger-prick or a morning blood draw, results in a week, clarity for a few hundred dollars. The appeal is not hard to understand. The transition announces itself as a mess of symptoms with no clear edges, and a number looks like the opposite of a mess.

Whether that number is worth buying has a settled answer in the clinical literature, and it is not the answer the market implies. An editorial in The BMJ in August 2025 stated it plainly: commercial hormone testing is unnecessary and does not improve care. What makes that verdict more interesting than a brush-off is the endocrine evidence underneath it, which explains not only that such tests disappoint but why they have to.

This is not one trial with a headline effect size, and it would be dishonest to present it as one. The claim rests on two separate bodies of work. The first is guideline consensus. Christakis and colleagues, writing in The BMJ, noted that guidance from the British Menopause Society, the National Institute for Health and Care Excellence and the American College of Obstetricians and Gynecologists, among others, agrees that in women over 45 presenting with validated symptoms of menopause, menopause is a clinical diagnosis and hormone testing is unnecessary. NICE and the British Menopause Society set practice in the United Kingdom rather than the United States; the weight of the point lies in the convergence across all three bodies, not in any one of them governing American care.

The second body of work is a set of endocrine cohort studies that sampled women repeatedly across the transition rather than once. Those studies are small, dozens of women rather than thousands, and old enough that assay methods have moved on since. They remain the ones that measured the quantity a consumer panel assumes away: how much a woman's hormones move while she is standing still.

Hale and colleagues drew blood three times a week for a little over a full cycle from 77 women aged 21 to 55, sorted by Stages of Reproductive Aging Workshop stage from mid-reproductive age through the late transition. In the transition cycles, what they found was not a decline. In 11 of 29 ovulatory cycles, which is 37%, estradiol rose a second time during the mid and late luteal phase and kept climbing into the following period, a surge with the shape of a normal follicular phase superimposed on a cycle that already had one. The authors named these luteal out-of-phase events.

Those cycles were not subtly different from the rest. They carried lower luteal-phase progesterone, higher early-cycle follicle-stimulating hormone and lower early-cycle inhibin B than the typical ovulatory cycles, and they were the cycles that came abnormally short, under 21 days, or abnormally long, over 40. The cycles most likely to send a woman looking for a test, in other words, are the cycles in which a single measurement is least interpretable. A draw taken during one of these events reads like a different woman than the same draw two weeks later.

Anti-Mullerian hormone is often presented as the stabler marker, on the reasoning that it reflects the remaining follicle pool rather than the month's fluctuations. Robertson and colleagues tested that directly, measuring it across ovulatory cycles in 18 women of mid-reproductive age and 43 in late reproductive life. In the younger group, and in 33% of the late-reproductive cycles, fourteen of them, the level held steady across the cycle as advertised. In the rest it did not: eleven cycles showed a two-fold fall between the follicular and luteal phases and ten showed a 4.2-fold rise, both at P less than 0.01, and in eight further cycles the level sat below the assay's limit of detection entirely. The authors concluded that precisely when anti-Mullerian hormone is low, which is the situation of a woman near menopause, it becomes less reliable as a marker.

The most serious attempt to turn a hormone level into a forecast is the Study of Women's Health Across the Nation. Finkelstein and colleagues measured anti-Mullerian hormone and follicle-stimulating hormone in 1,537 pre- or early perimenopausal women, mean age 47.5 years with a standard deviation of 2.6 at baseline, then serially until each had gone twelve months without a period, using a two-site ELISA with a detection limit of 1.85 pg/mL. The newer marker won: areas under the receiver operating curve for predicting a final menstrual period within 24 months were significantly greater for the anti-Mullerian models than the follicle-stimulating hormone ones.

Better, though, is not the same as decisive. A woman with a level below 10 pg/mL had a 51% probability of her final period within the next twelve months if she was under 48, rising to 79% if she was 51 or older. The cleaner end of the finding is the negative one: above 100 pg/mL, the probability of not reaching a final period within the year was 97% under 48 and 90% at 51 and over. The test is therefore informative at the extremes, and its answer in the middle, for a woman in her late forties with an ambiguous level, is a probability band in which her age is doing much of the work. Her age is something she already knows.

The clinical alternative to a panel is not a better panel; it is a history. The Stages of Reproductive Aging Workshop + 10, a consensus of scientists from five countries that reviewed the cohort evidence in 2011, stages reproductive ageing chiefly on bleeding patterns, with endocrine and ovarian markers in a supporting role. It simplified the bleeding criteria for the early and late transition, set out how long the late transition and early postmenopause last, and recommended that the staging apply regardless of a woman's age, ethnicity, body size or lifestyle. Its authors also named seven research priorities still outstanding, which is a more candid account of the remaining uncertainty than a results page tends to carry.

The BMJ editorial adds the practical reason a number cannot stand in for that history: there is no clearly defined therapeutic window for menopausal hormone therapy. A level does not identify a dose, because no target level has been established to aim at. Hence the authors' own formulation, that symptoms should be prioritised ahead of testing, and that effective treatment begins not with numbers but with listening. The tests, they noted, can cost hundreds of dollars.

None of this makes hormone measurement useless. It makes it specific. The question changes entirely when the woman is young: in suspected primary ovarian insufficiency, ovarian function ending years ahead of schedule, testing becomes diagnostic rather than confirmatory. A 2025 review in Endocrine reports that an anti-Mullerian hormone level below 8 pmol/L has shown 85% sensitivity and 100% specificity for diagnosing the condition in women presenting with secondary oligomenorrhoea. That is the profile of a test doing real work.

The same review is careful about the other direction. Anti-Mullerian hormone's reliability for identifying menopause itself is inconsistent, it notes, particularly in younger women and when the exact onset is the question, and tracking the marker over time predicts better than any single value. Two readings of one marker, two different jobs. The panel advertised to a forty-seven-year-old with erratic cycles is being asked to do the job it does worst.

The testing market has found an unusually effective seam: it sells certainty about the one process in medicine defined by refusing to hold still. The evidence does not say these panels are fraudulent, and it does not say hormones are irrelevant to midlife care. It says something narrower and more damaging to the product, which is that a single draw measures a quantity that can halve or quadruple inside one cycle in exactly the women most likely to buy the test, and that even the better-performing marker resolves to something near a coin flip for a woman in her late forties. Set against that, what the guidelines ask for looks almost old-fashioned: the pattern of bleeding across months, the symptoms as described, the age. It is slower than a result in a week and much harder to package, and on the present evidence it is also the thing that works. The honest summary of a perimenopause hormone panel is that it most often charges a few hundred dollars to report what a woman's own calendar and her account of the year had already established.