Research case · Independent candidate 0.5 · October 26, 2026

Keep the history.
Test its added value.

The goal is a useful, pregnancy-informed assessment that is earned through evidence. It is not simply a larger heart-age number.

Research only · Clinical coefficients remain unfitted

What established tools use.
Where the research starts.

Approach	What it uses	What it leaves out or cannot establish
U.S. PREVENT	Age, sex, blood pressure, cholesterol, BMI, kidney function, diabetes, smoking and treatments; optional UACR, HbA1c and social deprivation.	No separate inputs for preeclampsia, historical gestational diabetes or preterm delivery. Clinical guidance can consider pregnancy history alongside the score. PREVENT-Age is already available.
UK QRISK3	Age, sex, ethnicity, deprivation, blood pressure, lipid ratio, smoking, diabetes, family history and selected diagnoses and medicines.	No direct preeclampsia, historical gestational diabetes or preterm-delivery input. It is not a full pregnancy-history assessment.
UK QR4	Extends cardiovascular prediction with additional conditions, including preeclampsia and postnatal depression.	Important progress, but not a complete episode-by-episode reproductive history or a U.S.-validated pregnancy-adjusted age. Publication does not establish universal NHS adoption.
HEART RESERVE research candidate	Would fit its own coefficients from current measurements, detailed pregnancy episodes, reproductive history and selected additional markers.	Clinical weights and the reference-age curve are not fitted. Longitudinal outcomes, external validation, useful clinical decisions and safe communication must still be established.

These tools have different populations, endpoints and time horizons. Their numerical scores are not interchangeable. Sources: AHA PREVENT and its 2026 Quickstart Guide; QRISK3; Hippisley-Cox et al., Nature Medicine 2024 (QR4); NICE NG238. Read sources →

Absence from an equation is not absence from clinical care. Some effects of pregnancy history may already be reflected in current hypertension, diabetes or kidney health. The research must measure additional information, not count the same risk twice.

What HEART RESERVE would study

A more complete history,
with weights learned from outcomes.

The independent model would estimate its own coefficients and age reference. It does not take a PREVENT score as its input or borrow PREVENT’s age coefficient. The final feature set must be small enough for the available events and data quality.

Current measurements

Blood pressure, lipids, BMI, kidney function, diabetes, smoking and treatments; selected optional biomarkers when measured.

Pregnancy episodes

Preeclampsia, gestational hypertension, gestational diabetes, gestational age at delivery, fetal-growth context and pregnancy loss; study timing, recurrence and severity without duplicate penalties.

Reproductive life course

Age at first period, cycle history, diagnosed PCOS, menopause timing and cause. These are candidate features with differing evidence, not automatic added risk years.

Selected additional markers

Family history, measured Lp(a), ApoB, triglycerides and selected inflammation measures. Fit a feasible subset and test added value beyond correlated routine measurements.

TTTS, twin pregnancy and fetal interventions provide obstetric context, not an automatic numerical penalty. Period patterns, mental health, sleep, stress and other exploratory fields must not be assigned unsupported “extra years.” Unknown history must remain unknown, rather than silently becoming “no.”

What has actually been done

Public-data feasibility.
Not clinical validation.

The current package audited 11 NHANES 2017–2018 files. It counted 2,128 examined women ages 30–79, including 1,775 reporting prior pregnancy and 157 reporting diabetes during pregnancy.

These are unweighted sample counts, not national rates. The audit generated no clinical risk predictions or risk ages.

What is missing?

The audit does not supply the required prospective first-CVD outcomes, complete pregnancy episode phenotypes, or a validated reference-age curve. Public cross-sectional data cannot establish that the model predicts future events better.

CARDIA metadata mapping covers 17 domains and 80 variable references, but restricted participant records have not been received. The proposed year-20 landmark has a narrower age range than the eventual 30–79 goal. WHI and clinical-cohort discussions remain conditional on feasibility and access.

Lisa confirms that Dr. Quintero has offered his cohort as a potential research starting point. Her Rutgers discussion is scheduled for Wednesday, October 28, 2026. The next step is to define the cohort, maternal data, study question, research team and access approvals. No hospital partner, institutional study or validated model is secured.

The next study should answer three questions

Can we measure it?
Does it improve prediction?
Does it help care?

1. Feasibility

Confirm eligible participants, dates, complete predictors, adjudicated outcomes, follow-up, competing deaths and sufficient events. Agree on permitted use and independent validation data before fitting.

2. Prediction

Compare current measurements alone, measurements plus pregnancy, and carefully selected fuller models. Assess calibration, discrimination, prediction error, clinical decision benefit and uncertainty, including relevant subgroups.

3. Care and communication

Only after adequate validation, test whether the information improves understanding and appropriate care without false reassurance, unnecessary anxiety, testing or workload. An age display is optional.

How would the research age be calculated?

First estimate risk with the fitted HEART RESERVE model. Then find the age with the same risk on a fixed reference profile: AHR = inverse(Rref)(pHR). For a positive, linear reference log-odds age relationship: AHR = 50 + 10 × [logit(pHR) − eta50ref] / betaAgeRef. All reference parameters must come from the fitted independent model.

Comparing the full model with a model refitted without pregnancy information estimates a model-based age difference. It does not measure biological aging or causal years added by pregnancy. No number should be displayed outside a validated reference range.

The initial research population is proposed to be adults with prior pregnancy, ages 30–79, at least 12 months after pregnancy and without established cardiovascular disease. This research restriction is not advice to delay postpartum care. Current pregnancy, recent postpartum concerns and known heart failure need their own clinical pathways.

A separate scientific request.

Ask an eligible academic or health-system investigator to lead the cohort-feasibility study, with clinical and biostatistical expertise. Seek an appropriate research funding pathway separately from the national education budget.

The education campaign can carry pregnancy history into care while that research proceeds. No claim of superiority to PREVENT or QR4 should be made without a valid direct comparison.

Detailed collaborator and cohort pathway →

Internal source: HEART_RESERVE_Independent_Model_Cohort_Research_v05.zip, current model manifest, independent specification, public-data report and cohort-feasibility review, October 26, 2026. Historical PREVENT-extension formulas in that archive do not govern the current independent model.