Perimenopause as a data problem.

The thing I want to say about perimenopause, and the thing I want primary care to start saying about perimenopause, is that the transition is not subtle. The vasomotor flush is not subtle. The 03:14 a.m. wake-up is not subtle. The cognitive disruption that members describe as "I forgot how to read for a week" is not subtle. The mood lability that members describe as "I am angry at my own kitchen" is not subtle. None of it is subtle. The thing that has been subtle, for a long time, is what primary care has had to say about it.

I trained at the bedside of obstetric nursing for two decades before I came to family practice, and what I learned in obstetrics was that the volume of practical, evidence-based, plain-language information given to a 32-year-old in pregnancy is roughly two orders of magnitude larger than the volume of information given to the same person twenty years later in perimenopause. Pregnancy has a 700-page book, a class series, a midwife, an OB, a doula option, an app, a hospital tour, a sibling-orientation visit. Perimenopause, as recently as the year I left obstetrics, had a printed pamphlet from the manufacturer of an SSRI and a referral to a gynecologist who had a four-month wait. Same person. Different decade. Different attention.

The reason for the asymmetry is not that perimenopause is biologically unimportant. The reason is, in part, that the data primary care has on perimenopause is messier than the data on pregnancy, and the messy data has been used as a reason to say less. I want, in this transcript, to make the opposite case: the messy data is a reason to say more, more often, and with explicit acknowledgment of where the messiness lives.

What we know with high confidence

The Menopause Society 2022 hormone therapy position statement^[1] is, in my reading, the most readable and most clinically useful summary of the high-confidence data we have. The summary, compressed: menopausal hormone therapy initiated within ten years of the final menstrual period and under age 60 — in members without a contraindication — is highly effective for vasomotor symptoms, is the most effective treatment for genitourinary syndrome, reduces fracture risk, and carries small, characterizable absolute risks of venous thromboembolism, breast cancer, and stroke that vary by formulation, route, dose, and the patient's underlying risk profile. The Statement spells these out, with numbers, in tabular form.

What we also know with high confidence: the Women's Health Initiative did not say what most primary-care clinicians of my generation believe it said^[2]. The 2002 first publication produced a public-health response — including a 70% drop in hormone-therapy prescribing within two years — that was wildly out of proportion to what the underlying numbers supported, especially in the under-60 cohort that our perimenopausal patients fall into. The long-term WHI follow-up has reframed those findings considerably. Most clinicians have not updated.

What we know with moderate confidence

SSRIs and SNRIs reduce vasomotor symptom frequency and severity, modestly, in members for whom MHT is not appropriate or not desired^[3]. Gabapentin works, especially for nighttime symptoms. Cognitive behavioral therapy for hot flashes (CBT-HF, distinct from CBT for insomnia) has a real evidence base — modest effect on symptom bother independent of frequency, more durable than the comparison conditions, and useful for members who want a non-pharmacologic option. Fezolinetant, the first FDA-approved selective NK3-receptor antagonist, is genuinely new and genuinely useful^[4]; it is also expensive and underinsured.

What we don't know — and where we should say so

We do not have great data on the relative weight of perimenopausal mood lability versus a major depressive episode that happens to be co-occurring with the transition. The two share clinical features. The treatment overlaps. The prognosis differs. We do not have a clean diagnostic instrument; we have clinical judgment and PHQ-9 scores and a willingness to revise.

We do not have great data on the relationship between hormone therapy timing and dementia risk in a way that yields a primary-care-actionable answer. The hypotheses are interesting; the data are mixed; the long-term follow-up of MHT cohorts initiated within ten years of FMP is suggestive of neutrality or possibly modest benefit, and longer-time-distant initiation may be the opposite^[5]. We do not know enough to give an unambiguous recommendation, and I will say so to a member's face.

We do not have great data on the comparative effectiveness of fezolinetant versus standard-dose transdermal estradiol, head to head, in members with vasomotor symptoms. The trials we have were against placebo. The decision is currently driven by a member's preference for or against hormone therapy, not by relative efficacy, and that's an honest stance.

We do not have great data on perimenopausal cognitive symptoms. We have member-reported outcomes that converge — "I cannot remember names anymore" is a near-universal complaint that I encounter at a rate that defies coincidence — and we have small studies. We do not have a treatment with strong evidence beyond MHT-secondary improvement. I will tell a member that, too.

The data the member can collect

The instrument I recommend, more often than any other, in this lane is a paper notebook. Six weeks of cycle, sleep, vasomotor count, mood. Not for the chart, though we'll add it; primarily for the member, who, after six weeks, has a much sharper picture of what is happening to them than a recall-the-month conversation in the office can produce. The clinical decision in the second visit is dramatically better-informed for it. So is the member's. The most consequential thing primary care can do in this lane is hand a member a tracker and not ask for results before week six.

A modest stance

The transition is not subtle. The information primary care has historically given members about it has been. The data we have now is good enough to support a more informative conversation. The data we don't have is real, and we should say so. If the member's question is "is hormone therapy safe for me," the right answer is some flavor of "in the under-60, within-ten-years-of-FMP, no-contraindications population, the absolute risks are small and well-characterized; here are the numbers; here is what you would notice; here is what you would call us about." If the member's question is "will I get my brain back," the right answer, in 2026, is "I don't know, but I would like to track it with you."

I would like primary care to be in the habit of saying both kinds of sentences out loud.

— JA