ASCVD Pooled Cohort Risk Calculator

The 2013 ACC/AHA pooled cohort equations, with all four race- and sex-specific coefficient sets, the validity range they refuse outside, and an honest account of the race term and of how far they over-estimate.

Last updated 8 Oct 2026

10-year ASCVD risk, pooled cohort equations

4 coefficient sets

Coefficient setThe equations were published as four separate models, each with its own coefficients, mean and baseline survival. There is no single equation with race and sex as terms inside it. The race term is the most criticised feature of this instrument; see the body of the page.

Age (years)Derived and validated in adults aged 40 to 79. No value is returned outside that range.

Total cholesterol (mg/dL)In mg/dL, the unit the equations were published in — multiply a mmol/L result by 38.67. Valid range 130 to 320 mg/dL, which is 3.36 to 8.28 mmol/L.

HDL cholesterol (mg/dL)In mg/dL. Valid range 20 to 100 mg/dL, which is 0.52 to 2.59 mmol/L.

Systolic blood pressure (mmHg)Valid range 90 to 200 mmHg. Diastolic pressure does not appear in the equations at all.

On treatment for blood pressureTreatment selects a different systolic coefficient rather than adding a constant, so a treated 140 carries more risk in the model than an untreated 140.

Current smokerCurrent smoking only. No term for former smoking, pack-years or time since quitting.

DiabetesA single yes or no — no term for duration, control, HbA1c or type, which is one of the model’s larger simplifications.

5.4% over 10 yearsExample

White or other man, 55 years, total cholesterol 213 mg/dL, HDL 50 mg/dL, untreated systolic 120 mmHg, non-smoker, non-diabetic

Formula

Risk = 1 − S₀^{exp(Σβx − mean)}
Σβx is built from ln(age), ln(age)², ln(total cholesterol), ln(HDL), ln(systolic), their interactions with ln(age), smoking and diabetes
S₀, the mean and every β differ between the four models

four models, not one: white or other women, African American women, white or other men and African American men each have their own coefficients, mean and baseline survival (0.9665, 0.9533, 0.9144, 0.8954). The sets are not even the same shape: only the white women’s model has a squared log-age term, only the African American women’s has an age-by-systolic interaction, and the African American men’s has no interactions at all
valid in, and what it predicts: adults 40 to 79 without established ASCVD, total cholesterol 130 to 320 mg/dL, HDL 20 to 100 mg/dL, systolic 90 to 200 mmHg. Outcome: a first non-fatal myocardial infarction, coronary death, or fatal or non-fatal stroke in 10 years — not angina, heart failure, revascularisation or peripheral arterial disease. Derived in pooled cohorts of white and African American adults, not in other ancestries
discrimination and calibration: C index 0.72 (95% CI 0.70 to 0.75) in REGARDS (10,997 adults), 0.67 (0.64 to 0.71) in a Medicare-linked subset of 3,333. Calibration is the weak point: REGARDS found observed and predicted five-year rates similar, while 307,591 Kaiser Permanente adults above a 5% threshold had 8.72% predicted against 1.85% observed
the race term: widely criticised, because race is social rather than biological and a coefficient set selected by it absorbs everything that tracks with it. PREVENT (2023) removed it and added estimated GFR, body mass index and a social deprivation index; on the reference profile it returns 2.8% for a white man where this returns 5.4%

Worked example

White or other man, 55 years, total cholesterol 213 mg/dL, HDL 50 mg/dL, untreated systolic 120 mmHg, non-smoker, non-diabetic
ln(55) = 4.00733, ln(213) = 5.36129, ln(50) = 3.91202, ln(120) = 4.78749
Σβx = 12.344 × 4.00733 + 11.853 × 5.36129 − 2.664 × (4.00733 × 5.36129) − 7.990 × 3.91202 + 1.769 × (4.00733 × 3.91202) + 1.764 × 4.78749 = 60.6995
Σβx − mean = 60.6995 − 61.18 = −0.4805, so risk = 1 − 0.9144^{exp(−0.4805)} = 5.4%
This is the guideline's own worked example, which prints 5.3%. The other three coefficient sets on identical inputs give 2.1% for a white woman, 3.0% for an African American woman and 6.1% for an African American man — all four reproduced exactly, which is the check that matters on a coefficient table this long
The same profile through the 2023 PREVENT equations returns 2.8%: a factor of nearly two on one unchanged patient, and the clearest single statement of why the newer equations exist
Change the age to 39 or 80 and the page returns no value — the equations were derived in 40- to 79-year-olds

The guideline’s reference profile through all four coefficient sets

A 55-year-old with total cholesterol 213 mg/dL, HDL 50 mg/dL, untreated systolic 120 mmHg, non-smoker, non-diabetic — the guideline’s own published example, which quotes 2.1%, 3.0%, 5.4% and 6.1%. All four fall out of these coefficients to the published figure, a far stronger check on four long tables than reading them twice. On the same profiles the 2023 PREVENT equations return 1.9%, 1.9%, 2.8% and 2.8% — no race term, and about half the estimate in the men.
Coefficient set	10-year risk	Baseline survival	Mean
White or other woman	2.1%	0.9665	-29.18
African American woman	3.0%	0.9533	86.61
White or other man	5.4%	0.9144	61.18
African American man	6.1%	0.8954	19.54

What the equations are for, and where they have been tested

The two calibration findings cannot be reconciled by picking one, and this page does not try: REGARDS and the Kaiser cohort differ in era, population and ascertainment. The honest reading is that these equations are well calibrated in some contemporary populations and substantially over-estimate in others. The coefficient tables were read from a faithful reproduction of the guideline’s appendix; ahajournals.org is not reachable from this environment.
Question	Answer
Population	Adults 40 to 79 without established atherosclerotic cardiovascular disease. Derived by pooling community cohorts of white and African American adults; not derived in other ancestries
Outcome	First hard ASCVD event in 10 years: non-fatal myocardial infarction, coronary death, or fatal or non-fatal stroke
Input limits	Total cholesterol 130 to 320 mg/dL, HDL 20 to 100 mg/dL, systolic 90 to 200 mmHg. This page returns no value outside them
Discrimination	C index 0.72 (0.70 to 0.75) in 10,997 REGARDS participants aged 45 to 79 without diabetes; 0.67 (0.64 to 0.71) in 3,333 Medicare-linked participants
Calibration	REGARDS found observed and predicted five-year rates similar. Kaiser Permanente Northern California, 307,591 adults without diabetes, found 8.72% predicted against 1.85% observed above a 5% threshold, and 13.38% against 5.5% in 4,242 with diabetes
Risk categories	2018 AHA/ACC cholesterol guideline: low under 5%, borderline 5% to 7.4%, intermediate 7.5% to 19.9%, high 20% or more
Successor	The 2023 AHA PREVENT equations: 30 to 79 years, 10- and 30-year risk, total cardiovascular disease including heart failure, no race term, with estimated GFR and body mass index added

Four coefficient sets, a race term, and a calibration problem

The pooled cohort equations answer a specific question: what proportion of people with this handful of measurements had a first heart attack, coronary death or stroke within ten years. They are not one equation. They are four Cox models — white or other women, African American women, white or other men, African American men — each with its own coefficients, population mean and baseline survival, and the four are not even the same shape. The white women’s model carries a squared log-age term and the African American men’s carries no interaction terms at all. That is what the pooling found, and it is why a calculator has to select a whole model rather than set a variable.

The race term is the most criticised feature of the instrument, and the criticism is not primarily that it produces the wrong numbers. It is that race is a social category rather than a biological one, so a coefficient set chosen by it absorbs the effects of everything that tracks with race — income, insurance, neighbourhood, access, discrimination — and attributes them to ancestry. The American Heart Association’s 2023 PREVENT equations removed race entirely, added estimated GFR, body mass index and a social deprivation index, extended the age range down to 30 and added heart failure to the outcome. On the identical reference profile they return 2.8% for a white man where these equations return 5.4%.

Which raises the calibration question. The REGARDS validation of 10,997 adults found observed and predicted five-year rates similar, with a C index of 0.72. A Kaiser Permanente cohort of 307,591 adults without diabetes found the opposite: among those with a predicted five-year risk above 5%, the predicted rate was 8.72% and the observed 1.85%, with a similar over-estimate in the diabetic subgroup. Both findings are real, from different eras, populations and ascertainment, and the reasonable conclusion is that these equations rank people by risk moderately well while over-estimating absolute risk substantially in some contemporary populations.

Two things follow for anyone reading a number off this page. A population risk is not a probability for an individual: two people with identical inputs and very different coronary anatomy get the same figure, and the risk-enhancing factors the 2018 cholesterol guideline lists — family history, lipoprotein(a), chronic kidney disease, chronic inflammatory disease, South Asian ancestry, pre-eclampsia — sit outside the equation altogether. The estimate is a starting point for a conversation rather than a rule, and what the categories are then used for belongs to the LDL target by risk category interpreter. This page computes the number, gives the strata the guideline publishes and says what each predicted in the cohorts studied. This supports a clinician’s judgement rather than replacing it.

Frequently asked questions

What are the pooled cohort equations?

Four race- and sex-specific Cox models from the 2013 ACC/AHA risk guideline, estimating the 10-year risk of a first hard ASCVD event — non-fatal myocardial infarction, coronary death, or fatal or non-fatal stroke — in adults aged 40 to 79 without established disease.

Why do the equations have a race term?

The derivation cohorts produced different coefficient sets for white and African American participants and the guideline published them separately. The criticism is that race is a social rather than a biological category, so the term absorbs everything that tracks with it. PREVENT removed it entirely.

Do the pooled cohort equations over-estimate risk?

In several contemporary cohorts, substantially. At Kaiser Permanente Northern California, among 307,591 adults without diabetes whose predicted five-year risk exceeded 5%, the predicted rate was 8.72% and the observed 1.85%. The REGARDS validation of 10,997 adults found observed and predicted rates similar. Both stand; the populations differ.

What is the valid age range?

40 to 79 years, and this page returns no value outside it — nor outside total cholesterol 130 to 320 mg/dL, HDL 20 to 100 mg/dL and systolic 90 to 200 mmHg. A Cox model extrapolated past its derivation range gives a confident number with no evidence behind it.

How do the PREVENT equations differ?

They cover ages 30 to 79, give 10- and 30-year risk, predict total cardiovascular disease including heart failure, drop the race term and add estimated GFR and body mass index. On the pooled cohort reference profile they return 2.8% for a white man against 5.4%.

Related calculators

References

Goff DC Jr, Lloyd-Jones DM, Bennett G, et al. 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2014;129(25 Suppl 2):S49–S73.
Rana JS, Tabada GH, Solomon MD, et al. Accuracy of the atherosclerotic cardiovascular risk equation in a large contemporary, multiethnic population. J Am Coll Cardiol. 2016;67(18):2118–2130.
Muntner P, Colantonio LD, Cushman M, et al. Validation of the atherosclerotic cardiovascular disease Pooled Cohort risk equations. JAMA. 2014;311(14):1406–1415.
Khan SS, Matsushita K, Sang Y, et al. Development and validation of the American Heart Association’s PREVENT equations. Circulation. 2024;149(6):430–449.
Grundy SM, Stone NJ, Bailey AL, et al. 2018 AHA/ACC/multisociety guideline on the management of blood cholesterol. J Am Coll Cardiol. 2019;73(24):e285–e350.

Not medical advice. For healthcare professionals and education. Reference intervals vary by laboratory and assay — always use your own laboratory's. Never base a dose or a treatment decision on this page alone. Full disclaimer at calcengines.com/disclaimer/