ASCVD Risk Calculator: PREVENT and Pooled Cohort

ASCVD and PREVENT Heart Risk Calculator

An ASCVD risk calculator estimates the chance that a person who does not already have heart disease will have a cardiovascular event, such as a heart attack or a stroke, within a set number of years. This page runs two published United States models: the American Heart Association's PREVENT equations and the older Pooled Cohort Equations, known to most people as the ASCVD risk score. It reports a percentage and shows how each risk factor moved that number. It gives no medical advice.

What the calculator does

The page asks for age, sex, total and HDL cholesterol, systolic blood pressure, whether that pressure is treated, smoking, and diabetes. PREVENT also asks for body mass index and eGFR, a blood-test measure of kidney function, and whether the person takes a statin. The Pooled Cohort Equations instead ask for a race group, because they were published as four separate models split by sex and race. PREVENT has no race input.

PREVENT returns six numbers: the 10-year risk of total cardiovascular disease, of ASCVD, and of heart failure, and the same three over 30 years. The Pooled Cohort Equations return one number, the 10-year risk of a first hard ASCVD event, meaning a non-fatal heart attack, death from coronary heart disease, or a fatal or non-fatal stroke.

How PREVENT calculates risk

PREVENT is a logistic model. It adds up weighted risk factors, then squeezes the total into a probability:

$R = \frac{1}{1 + e^{-x}}$

where R is the estimated risk, x the person's score and e the base of the natural logarithm.

The score x is an intercept plus one weighted term per risk factor. Each input is first centred, which means it is measured as a distance from a reference point, so a person sitting at every reference point scores zero and only the intercept is left. Age is centred at 55 and counted per 10 years. Non-HDL cholesterol, which is total cholesterol minus HDL, is centred at 3.5 mmol/L. HDL is centred at 1.3 mmol/L and counted per 0.3 mmol/L.

Three inputs are split at a hinge, so they carry a different weight either side of it: blood pressure at 110 mmHg, body mass index at 30 kg/m², and eGFR at 60 mL/min/1.73m². The eGFR terms count units lower, so worse kidney function raises the score. Most factors also carry an age interaction, and the 30-year equations add an age-squared term. The coefficients come from Khan SS, Coresh J, Pencina MJ, et al., "Development and Validation of the American Heart Association's PREVENT Equations", Circulation 2024;149(6):430-449, Supplemental Table S12.

How the Pooled Cohort Equations calculate risk

The Pooled Cohort Equations are a Cox survival model:

$R = 1 - S(10)^{\exp(u - m)}$

where R is the estimated risk, S(10) the group's baseline 10-year survival, u the person's weighted total and m the published average total for their group. Age, cholesterol and blood pressure enter as natural logarithms, and several terms multiply ln(age) by another logarithm. Treated and untreated blood pressure use different coefficient rows; exactly one applies. The four coefficient sets come from Goff DC Jr, Lloyd-Jones DM, Bennett G, et al., "2013 ACC/AHA Guideline on the Assessment of Cardiovascular Risk", Circulation 2014;129(25 Suppl 2):S49-S73, Table 4 of the Full Work Group Report.

Worked example: PREVENT

A 50-year-old woman, total cholesterol 200 mg/dL, HDL 45 mg/dL, untreated systolic pressure 140 mmHg, current smoker, no diabetes, body mass index 28, eGFR 85, no statin.

Centred values: age (50 − 55) / 10 = −0.5. Non-HDL is 155 mg/dL, which is 4.0083 mmol/L, so 4.0083 − 3.5 = 0.5083. HDL is 1.1637 mmol/L, so (1.1637 − 1.3) / 0.3 = −0.4543. Pressure above the hinge is (140 − 130) / 20 = 0.5. eGFR above 60 is (85 − 90) / −15 = 0.3333. Smoking is 1.

Multiplying by the 10-year total-cardiovascular-disease coefficients for women gives these contributions, shown to three decimals as the page shows them:

Adding the unrounded contributions to the intercept of −3.307728 gives x = −2.7948. Then risk = 1 / (1 + e^2.7948) = 5.76%, which the page reports as 5.8%. The same profile gives 3.6% for 10-year ASCVD, 2.0% for 10-year heart failure, and 29.2%, 17.9% and 13.5% over 30 years.

Worked example: Pooled Cohort Equations

Table 4 of the 2013 Full Work Group Report publishes one profile: 55 years old, total cholesterol 213 mg/dL, HDL 50 mg/dL, untreated systolic pressure 120 mmHg, non-smoker, no diabetes. For a White woman the calculator produces a weighted sum of −29.6767. The group mean is −29.18 and baseline survival is 0.9665, so:

exp(−29.6767 − (−29.18)) = 0.6086, and 1 − 0.9665^0.6086 = 0.02052, or 2.1%.

That matches the 2.1% printed in Table 4. The same profile gives 3.0% for an African American woman and 6.1% for an African American man, both matching the table. For a White man the table prints 5.3% and the calculator gives 5.4%. The gap is the table's own arithmetic: it rounds every coefficient-by-value product to two decimals before adding them, and feeding its rounded sum of 60.69 back into the survival step reproduces 5.3%.

Risk bands

A Pooled Cohort Equations result is labelled low under 5%, borderline from 5% to under 7.5%, intermediate from 7.5% to under 20%, and high at 20% or more. Those four bands are not in the 2013 report the equations come from, which publishes no such table. They are the categories in Grundy SM, Stone NJ, Bailey AL, et al., "2018 AHA/ACC/Multisociety Guideline on the Management of Blood Cholesterol", Circulation 2019;139(25):e1082-e1143, repeated in Arnett DK, Blumenthal RS, Albert MA, et al., "2019 ACC/AHA Guideline on the Primary Prevention of Cardiovascular Disease", Circulation 2019;140(11):e596-e646. Those guidelines use the bands when weighing treatment. This page uses only the labels, and says where a number falls. It sets no threshold of its own. PREVENT results are shown without bands.

The result is rounded to one decimal place before it is classified, so the band shown can never disagree with the number printed beside it.

Cholesterol units

Both models are written in mg/dL, so a value entered in mmol/L is multiplied by 38.67 once on the way in. That factor is the rounded clinical one from the NCBI units conversion table (Bookshelf NBK83,505); the exact molar figure is 38.665, and neither model paper prints it. The PREVENT equations then apply their own published factor of 0.02586 mmol/L per mg/dL internally, because that is the number the model was fitted with. It is not the exact reciprocal, and the round trip gains about six parts in a million, far below the one decimal place anything is reported to.

What the models cover, and what they do not

The tool calculates and stops there. It sets no treatment threshold, recommends no statin or blood pressure medicine, gives no doses, and diagnoses nothing. A risk estimate is a population average for people with similar measurements. It is not a prediction about one person and not a diagnosis.

PREVENT was fitted for ages 30 to 79. Its 30-year figures exist only for ages 30 to 59, because a 30-year horizon has to end inside the model's own age range. The Pooled Cohort Equations were fitted for ages 40 to 79 and have no 30-year version. Both are meant for people who do not already have cardiovascular disease. Neither covers pregnancy, inherited very high cholesterol, or people already being treated for an existing heart condition.

Inputs are checked in code, not only by the form: total cholesterol 130 to 320 mg/dL, HDL 20 to 100 mg/dL, systolic pressure 90 to 200 mmHg, and for PREVENT body mass index from 18.5 to under 40 and eGFR 15 to 150. A value outside those ranges is refused rather than estimated. Body mass index of exactly 40 is refused, matching the reference implementations, whose published input window stops at 39.9.

The coefficients are fixed on the page and cannot be edited. Their dates are the versions in use: PREVENT from the 2024 Circulation paper for the model the American Heart Association released in 2023, the Pooled Cohort Equations from the 2013 guideline, and the risk bands from the 2018 cholesterol guideline. What a user does control is which model runs, the sex and race group where a model has one, and the cholesterol unit.

Frequently asked questions

What is the difference between PREVENT and the ASCVD risk score? The ASCVD risk score usually means the 2013 Pooled Cohort Equations. PREVENT is newer, covers ages 30 to 79 rather than 40 to 79, uses no race input, adds kidney function and body mass index, and predicts heart failure and 30-year risk as well.

Why do the two models give different numbers for the same person? They were fitted on different populations and predict different outcomes. In the cohort used to validate PREVENT, the Pooled Cohort Equations predicted about twice the number of events that actually occurred, so they usually read higher.

Why does PREVENT not ask about race? Its authors dropped race as an input and added kidney function instead. The 2013 equations were published as four models split by sex and race, so this page asks for a race group only when those equations are selected.

Why is there no 30-year result for an older adult? PREVENT publishes 30-year equations only for ages 30 to 59. Above that age the horizon would run past the range the model was fitted on, so the page shows nothing rather than an unsupported number.

What does the breakdown table mean? Each row is that factor's contribution to the model's internal score, with its age interaction folded in. Positive numbers raise the estimated risk and negative numbers lower it. The rows are not percentages.

Can this replace a check-up? No. It reproduces two published equations from numbers a person already has. Interpreting a result, and deciding anything about it, is a matter for a clinician.

References

1. Khan SS, Coresh J, Pencina MJ, et al. "Development and Validation of the American Heart Association's PREVENT Equations." Circulation. 2024;149(6):430-449. doi:10.1161/CIRCULATIONAHA.123.067626.
2. Goff DC Jr, Lloyd-Jones DM, Bennett G, et al. "2013 ACC/AHA Guideline on the Assessment of Cardiovascular Risk." Circulation. 2014;129(25 Suppl 2):S49-S73.
3. Grundy SM, Stone NJ, Bailey AL, et al. "2018 AHA/ACC/Multisociety Guideline on the Management of Blood Cholesterol." Circulation. 2019;139(25):e1082-e1143.
4. Arnett DK, Blumenthal RS, Albert MA, et al. "2019 ACC/AHA Guideline on the Primary Prevention of Cardiovascular Disease." Circulation. 2019;140(11):e596-e646.
5. "Units Conversion Table." NCBI Bookshelf NBK83,505, https://www.ncbi.nlm.nih.gov/books/NBK83505/.