Menopause in the OMOP vocabularies

What this vocabulary review can establish

This worked example is for an OB/GYN planning research on perimenopause and menopause. It describes the concepts available in vocabulary release v5.0 29-AUG-26, their OMOP domains, mapping limitations, and decisions that need clinical review. A patient data set is not needed for this step. Vocabulary review can establish how an idea is represented; it cannot establish how often clinicians record it.

OMOP is the common data model. OHDSI is the community and tool ecosystem, pronounced “Odyssey.” ATHENA is the vocabulary browser; ATLAS supports vocabulary exploration, concept sets, and study design. A concept list is the starting point for a research definition.

Search across names and synonyms for menopause, perimenopause, climacteric, ovarian insufficiency, hot flashes/flushes, menstrual history, genitourinary symptoms, surgery, and treatments. A search for the word menopause alone misses clinically relevant concepts and retrieves irrelevant products or family-history records. Search results include standard concepts, nonstandard source codes, classification concepts, and retired terms.

The tables and downloads describe vocabulary release v5.0 29-AUG-26. They provide selected concepts and mappings for clinical review rather than an exhaustive inventory or a validated phenotype. The concept-reference download includes exact identifiers, domains, standard status, and validity metadata.

Domains and clinical concepts

Menopausal status is not confined to Observation, and menopausal symptoms are not confined to one domain. For a primary clinical event, use the standard concept's domain_id to identify the intended OMOP table. A source vocabulary's domain, the concept's name, and its concept_class_id are not substitutes for that field. A symptom can be a Condition; a concept class named “Procedure” can have the Observation or Measurement domain. [1]

Value-field exception: when a source code has a paired “Maps to” and “Maps to value,” the first mapping identifies the primary event and the second supplies its value. A value concept's domain does not by itself move that answer to another clinical table. See the HRT example below. [2]

Hot flashes and night sweats across domains

Menopausal flushing (4113666) is a valid standard Condition in v5.0 29-AUG-26. Medication-related and patient-reported hot-flash concepts can use Observation. Check the exact concept and its domain when defining symptoms and selecting OMOP tables.

Retired terms and mappings can change the apparent clinical meaning

Open clinical question: should a vasomotor study distinguish clinician-attributed menopausal symptoms, nonspecific flushing/sweating, medication-associated symptoms, and patient-reported burden? A useful review preserves all four categories until the question is settled. Frequency, severity, bother, and duration are different outcomes; a diagnosis label does not supply each of them.

Perimenopause, menopause, and ovarian insufficiency

Keep the menopausal transition, established postmenopause, premature ovarian insufficiency (POI), early menopause, and induced ovarian dysfunction distinct. The vocabulary includes useful labels, but those labels do not supply complete staging or onset histories.

Perimenopause is also represented by Abnormal perimenopausal bleeding (4201098), Perimenopausal disorder (4141640), Perimenopausal atrophic vaginitis (4255393) and Symptom severity level: perimenopause (4229863). Finding one general perimenopausal-state concept is not evidence that only one relevant concept exists.

Clinical definitions need a named framework

Natural menopause is generally established retrospectively after 12 months without menstruation when another explanation does not apply. Hormonal contraception, hysterectomy, and other causes of absent bleeding require different interpretation. A missing menstrual record is not evidence of amenorrhea. [4]

The ESHRE/ASRM POI guideline distinguishes POI before age 40 from early menopause at ages 40 through 44. Its POI criteria combine at least four months of disordered cycles with FSH above 25 IU/L; repeat FSH is recommended if uncertainty remains. NICE NG23 retains a two-sample POI recommendation. Name the guideline and version chosen for a study instead of assuming one universal laboratory rule. These are clinical definitions to discuss, not validated OMOP algorithms. [5] [4]

Menstrual history, laboratory results, and patient reports

Clinically similar questions can be represented in different domains. LOINC does not mean “laboratory test only,” and SNOMED does not mean “Condition only.” Question concepts need answers, numbers, units, or dates. The event's recording date is not necessarily the date reported in its value.

What remains to be specified

• Final menstrual period: distinguish an LMP date, an age-at-menopause response, and a retrospectively established final menstrual period. A dedicated final-menstrual-period concept was not identified in this vocabulary review; this is not proof of absence from every vocabulary.
• Hormone tests: FSH, LH, estradiol, AMH, and inhibin B test concepts are available. Specimen, units, assay, result, timing, and hormone exposure determine interpretation. A test order or a single abnormal value is not a universal menopause definition; routine staging in otherwise healthy people aged 45 or over is often clinical. [4]
• Patient-reported outcomes: Greene, FACIT, and PhenX-related entries were found. Dedicated standard concepts for MENQOL, the Menopause Rating Scale, and the Kupperman index were not identified by this search. Confirm instrument version, language, scoring, and permissions separately; a vocabulary concept does not grant instrument-use rights.
• Missing concepts versus incomplete search: search synonyms, component items, panels, answer concepts, and source vocabularies before declaring a gap. A symptom item in a questionnaire is not automatically menopause-specific.

Surgery, induced menopause, and mapping specificity

Separate procedures performed, acquired-absence status, historical questions, and diagnosis labels. Hysterectomy does not establish removal of both ovaries; endometrial ablation does not establish loss of ovarian function. Removal of one ovary, both ovaries, or the last remaining ovary has different implications. Bilateral removal after natural menopause does not date the onset of menopause.

These are specific mappings or clinical ambiguities for review. They do not justify globally deleting a valid target concept or silently changing the OHDSI vocabulary. Record the source and target IDs, release, relationship, proposed interpretation, and evidence when raising a vocabulary issue.

Keep chemotherapy, radiation, GnRH treatment, and endocrine therapy as separate context. Drug-induced suppression may be temporary, and a product concept does not encode whether ovarian function recovered. Pregnancy, lactation, thyroid/pituitary disorders, hypothalamic amenorrhea, and other explanations should be distinguished rather than pooled into menopause.

Coverage limitation: ICD10PCS content was unavailable for this review, and CPT4 coverage was incomplete. The surgical examples therefore do not establish comprehensive coverage of procedure codes. Historical gravidity counts also do not establish a current pregnancy episode.

Hormone therapy, contraception, and other treatments

Treatment representation spans Drug, Observation, Procedure, and sometimes Device. Keep product, formulation, route, strength, combination ingredients, and intended use separate. A prescription can support a research description of treatment but does not independently establish menopause or its cause.

Contraceptive vaginal rings: ethinyl estradiol/etonogestrel and ethinyl estradiol/segesterone products, including NuvaRing and Annovera, represent hormonal contraception. Classify these products using their ingredients and indications. Vaginal administration alone does not establish local menopausal hormone therapy. [6]

A value-field example: ICD9CM V07.4 maps to 441482 Administrative reason for encounter, with 37110086 Postmenopausal hormone replacement therapy as “Maps to value.” The primary record is an Observation with a therapy value, even though 37110086 has Procedure domain when used as a primary event. A query of only procedure_concept_id would miss that representation. [2]

Organizing concepts for clinical review

A concept set can contain more than one domain. Separate modules often make review easier, but the important rule is that cohort criteria must reach the relevant event tables and, where needed, require the associated value. A Condition criterion does not automatically search Observation. [3]

Descendants need inspection

Do not switch on descendants for every parent. Under 4172857 Menopause finding, the reviewed hierarchy includes positive menopause, menopause absent, and perimenopause. Under broad menstruation/menopause parents, descendants can extend outside the intended topic. Review the resolved list, not only the selected parent names.

Family-history concepts such as FH: Early menopause (4050823) describe relatives; they may be useful covariates but not the person's own menopausal status. A negative concept can remain useful in a separate interpretation module. Excluding it from a positive set is different from excluding a person forever because a premenopausal record exists.

For each proposed inclusion, record the clinical purpose, domain, standard status, validity, descendant decision, source mappings, required value, and reviewer's rationale. Preserve the vocabulary release and distinguish a semantic concern from a coding or grouping error.

Using ATHENA and ATLAS without a clinical data set

1. Search ATHENA for the clinical term and its synonyms, then inspect the exact concept page.
2. Record concept_id, source vocabulary/code, domain, class, standard flag, validity, relationships, and descendants.
3. In ATLAS, compare the same ID in the vocabulary loaded by that instance. Record the release actually in use; do not assume it matches ATHENA.
4. Build draft concept sets and inspect their resolved concepts and mapped source codes. Keep unreviewed mappings visible.
5. Export the reviewed definitions and open questions. Record-count availability is optional at this stage.

ATLAS count columns depend on the configured sources and available analyses. They are not universal counts for menopause. Concept metadata can be reviewed without those counts; source-specific recording, table placement, and clinical completeness require a later data review.

Clinical questions to settle before a study

The immediate deliverable is a clinical interpretation guide and a list of unresolved decisions. The following questions can be answered by reviewing the vocabulary with an OB/GYN, before selecting a data source.

Two candidate designs to discuss later

Documented menopausal status: begin with reviewed positive Condition concepts, then consider validated Observation question-answer pairs as a separate route. State that the index is documentation, not biological onset. Prior-observation rules describe visibility in a future data source; they cannot prove that menopause started within that window.

POI or early menopause: use separate definitions with onset-age evidence and distinguish spontaneous from induced cases. Generic ovarian failure and the age at the first coded encounter are insufficient by themselves. A one-year surgical-history exclusion can miss remote surgery; the relevant history interval needs clinical justification. These are design sketches, not executable or validated phenotypes.

Drafting a research question

Use this form to draft a question after the clinical review. Its data-dependent fields start as unknown, and it does not calculate a feasibility score in this vocabulary-only example. A future data source would be needed to assess recording, cohort size, follow-up, and access. The downloaded JSON is a planning draft.

Possible directions include characterizing documentation of the transition, comparing symptom representations, describing systemic versus local therapy, or examining how surgical history changes the interpretation of menopausal status. Treatment-effect questions additionally require an appropriate comparator, timing, and confounding strategy; vocabulary coverage alone cannot establish those.

Download the two unscored research-question drafts.

Downloads and sources

Downloads: reviewed concept reference; source-mapping examples; vocabulary-only review prompt.

Reference release: v5.0 29-AUG-26. Concept metadata, mappings, and hierarchy examples were checked against vocabulary extracts for this release. The downloads distinguish vocabulary fields from clinical interpretation. Coverage is limited to the concepts and source vocabularies reviewed.

• [1] OMOP CDM specification and The Book of OHDSI, standardized vocabularies: domains, fields, and vocabulary interpretation.
• [2] OHDSI mapping documentation: Maps to, Maps to value, and changes in specificity.
• [3] ATLAS concept-set documentation and cohort-definition guidance.
• [4] NICE NG23 recommendations: clinical identification, hormone-treatment context, and POI recommendations. Choose and record the version relevant to the protocol.
• [5] ESHRE/ASRM and collaborators, evidence-based POI guideline: age definitions, clinical and biochemical criteria, and intermittent ovarian activity.
• [6] DailyMed prescribing information for NuvaRing and Annovera.
• [7] EMA Agreal referral for veralipride; elinzanetant prescribing information.
• [8] Progesterone monotherapy trials: postmenopausal trial and perimenopausal trial. Evidence varies; these studies do not make a progesterone prescription proof of menopause.
• [9] DailyMed prescribing information for Femring and Estring: product-specific indications and systemic absorption.
• ATHENA, ATLAS demo, and OHDSI Vocabulary Users forum.

Using this reference in a study

1. Record the vocabulary release and source vocabularies used by the study.
2. Verify the selected concepts, source mappings, standard status, and validity in the study environment.
3. Review descendants and any required answers, units, or dates with the clinician.
4. Document inclusion decisions, mapping ambiguities, and clinical definitions in the protocol.
5. Assess recording and phenotype performance when an appropriate clinical data source becomes available.

Questions or suggestions: danielle@boycedatascience.com