Health Records Have No Standard Field for Menopause Stage

A woman in her early fifties is standing in a kitchen at eleven at night, three weeks after a change of insurance and therefore a change of clinician. The night sweats started around eighteen months ago; the sleep went first, then the concentration at work. She has explained all of it twice now, to two different people. What she is reading is a headline saying menopause is missing from most electronic health records, and the sentence lands because it explains something she had put down to bad luck: every new clinician starts from nothing, and the chart that is supposed to carry her history forward appears not to know that the last two years happened.

The study behind that coverage is a scoping review published in Menopause, the journal of The Menopause Society, on 1 September 2026, by Robin R. Austin, Malin B. Lalich, Meijia Song, Jeana M. Holt and Carolyn Gibson. The lagging question — the one still standing after the news cycle moves on — is not whether the gap exists. It is narrower: what exactly is missing, which part of the record it is missing from, and what a review of this kind can and cannot establish about it.

A scoping review is a map, not a verdict. Where a meta-analysis pools results from comparable studies into a single estimate, a scoping review asks what the literature contains at all: which studies exist, what they measured, and where the blank spaces sit. It is the right instrument when a field is too heterogeneous to pool and the useful question is the shape of the evidence rather than its average. It is also the reason no single percentage comes out of the other end of this paper.

The authors searched four databases — Ovid MEDLINE, Scopus, CINAHL and IEEE Xplore — for work published between 2004 and March 2026, using terms covering menopause, climacteric, perimenopause and EHR documentation, and added further studies by following citations backward out of the papers they already had. Two reviewers screened independently and settled disagreements by consensus, which is the standard guard against one reader's judgment quietly setting the boundary of a field. Quality was appraised with the Johns Hopkins Evidence-based Practice framework. The presence of IEEE Xplore in that list is a small tell about the subject: part of this literature is not clinical research at all but informatics, written by the people who design the fields rather than the people who fill them in.

Nineteen studies met the inclusion criteria. Most were retrospective cohort or cross-sectional analyses conducted in US health systems between 2013 and 2026, and their sample sizes ranged from 45 women to 307,512 — a spread of four orders of magnitude, which is itself a statement about how unsettled the area is. A field with a mature method does not produce studies that differ in size by a factor of several thousand.

Thirteen of the nineteen identified menopause-related experience through ICD-9 or ICD-10 diagnosis codes. Only nine reported which codes they used. That second number is the one worth holding onto: in four studies, the definition of who counted as menopausal cannot be recovered from the paper, so no one can check whether two studies that appear to measure the same thing actually did.

Across the nineteen, the authors describe three interlocking themes. Documentation limitations, including the absence of structured reproductive stage fields and a resulting reliance on unstructured text — a sentence inside a visit note, which a database query looking for a coded value will not find. Undertreatment of symptoms that were documented. And underreporting of menopause-related symptoms in the EHR data itself. The three compound rather than merely coexist: a symptom that is not asked about is not recorded, a symptom recorded only in prose is not counted, and what is not counted does not appear in the evidence that would justify building a field for it.

The authors conclude that menopause remains inconsistently represented in EHRs, and that the gaps are structural rather than a matter of individual diligence. They name three things those gaps constrain: longitudinal tracking, clinical decision-making, and research reproducibility. Those are less three separate complaints than one loop. A record that cannot carry reproductive stage forward cannot show the next clinician where a patient stands in the transition; decision support built on coded values cannot act on a fact that lives in a paragraph; and research that has to reconstruct menopause status from diagnosis codes inherits every inconsistency in how those codes were applied in the first place.

What the review calls for is standardized menopause common data elements — agreed definitions, with structured fields for stage and for symptoms — and it is candid that this will require coordinated, multidisciplinary effort rather than one vendor's update. That is a recommendation about infrastructure, and infrastructure recommendations are slow. Nothing in the review reports a health system that adopted such fields and then measured what changed, because the review found no such study to include.

The temptation with a paper like this is to read it as an indictment of clinicians who did not write things down, and that is the one reading the evidence does not support: a clinician cannot enter a fact into a field that does not exist. What makes the finding serious is the loop it describes rather than any single number inside it — the absence of a structured stage field pushes menopause into free text, free text does not survive a database query, and a condition that cannot be counted never generates the evidence base that would argue for building the field. That gap is self-sealing, and self-sealing gaps do not close on their own timetable. The call for common data elements is the right shape of answer and also the slowest one available, needing agreement across specialties, vendors and coding bodies before anything appears on a screen. The most quietly damning detail in the paper is not its conclusion but those four studies that never reported their codes: until the same fact can be named the same way twice, the size of this problem stays unmeasurable, and unmeasurable problems lose to measurable ones every time a budget is set.