What the PREVENT equations are
PREVENT stands for Predicting Risk of Cardiovascular Disease EVENTs. The equations were developed by Khan and colleagues for the American Heart Association and published in Circulation in 2024. They were derived and validated in more than 6.6 million adults in the United States, drawn from a mix of research cohorts, health systems and nationally representative surveys, and they estimate the probability of a first cardiovascular event over 10 years and over 30 years. The headline output is total cardiovascular disease: the combination of atherosclerotic cardiovascular disease (ASCVD, meaning myocardial infarction or stroke) and heart failure. Separate estimates are also given for ASCVD, heart failure, coronary heart disease (CHD) and stroke, because the drivers of each outcome differ and because treatment conversations often focus on one outcome at a time.
The base model, which is what this calculator implements, uses ten inputs: age, sex, systolic blood pressure, whether blood pressure medication is being taken, total cholesterol, HDL cholesterol, whether a statin is being taken, diabetes, current smoking, kidney function as eGFR, and body mass index. Strictly, BMI only enters the heart failure equations; the other outcomes do not use it. Race is not an input. That is a deliberate design decision the authors made, replacing the race-based coefficients of older models with directly measured clinical factors such as kidney function. PREVENT also starts at age 30 rather than 40, which matters because the 30-year estimates are most useful in younger adults whose 10-year risk is almost always low even when their lifetime trajectory is poor.
The inputs, one by one
Age and sex. The equations are sex-specific and were developed for adults aged 30 to 79 years without existing cardiovascular disease. Outside that age range the calculator will not return a number, because extrapolating beyond the development range would produce a precise-looking estimate the model was never shown to support.
Systolic blood pressure and treatment status. Enter the systolic (upper) blood pressure in mmHg, and say whether blood pressure medication is being taken. The equations treat a given blood pressure level differently depending on whether it is achieved with medication, because in the derivation data treated hypertension carried a different risk signal than the same untreated pressure. The relationship is modelled with a flexible spline, so risk does not fall in a straight line at very low pressures.
Cholesterol values. Total cholesterol and HDL cholesterol can be entered in mg/dL or mmol/L; the selector converts for you. The equations work with HDL and non-HDL cholesterol (total minus HDL) internally. Whether a statin is being taken is also asked, again because the same cholesterol value carries a different meaning on treatment.
Diabetes and smoking. Both are yes or no inputs. Diabetes is one of the strongest single contributors in the models, and smoking raises every outcome, most sharply heart failure and total CVD.
eGFR. Kidney function enters as the estimated glomerular filtration rate in mL/min/1.73 m2. This is one of the genuine advances of PREVENT over older scores: reduced eGFR raises predicted risk across outcomes, so two people with identical blood pressure and cholesterol but different kidney function get different, more accurate estimates.
BMI. Body mass index is used only by the heart failure equations, where it carries substantial weight. The derivation cohorts for the heart failure model excluded people with BMI below 18.5 or at or above 40 kg/m2, so heart failure estimates near those extremes rest on thinner data.
How the calculation works
Each outcome and each time horizon has its own published equation. The inputs are first centred and rescaled around reference values (for example, age is expressed relative to 55 years, systolic pressure relative to spline knots at 110 and 130 mmHg, eGFR relative to knots at 60 and 90, and cholesterol in mmol/L relative to typical values). A weighted sum of these terms, including interactions between age and several risk factors, is passed through a logistic function to give a probability between 0 and 100 percent. This page uses the published base-model coefficients exactly as reported, and the results were verified against published worked examples, including the example printed in the original paper: a 50-year-old woman with treated systolic pressure of 160 mmHg, total cholesterol 240 mg/dL, HDL 55 mg/dL, no diabetes, BMI 35 and eGFR 90 has a 10-year total CVD risk of about 5.4 percent, ASCVD about 3.6 percent and heart failure about 2.5 percent, rising to about 9.3, 6.0 and 4.7 percent if she smokes. You can reproduce those numbers in the calculator above.
Two deliberate omissions are worth stating plainly. First, 30-year results are shown only for ages 30 to 59, matching the American Heart Association implementation; beyond that age a 30-year horizon leaves the validated range. Second, the 30-year coronary heart disease estimate for women is not displayed. Every other coefficient in the model set could be verified against two independent published reproductions of the equations, but the final interaction terms of that single equation could not be verified reliably, and this site does not publish medical numbers it cannot verify. The cell is left blank with a note rather than filled with an approximation.
Reading the result: categories and thresholds
A percentage is only useful if it connects to a decision. For 10-year PREVENT-ASCVD risk, the interpretation used with the 2026 ACC/AHA dyslipidemia guideline is: low risk below 3 percent, borderline risk 3 to below 5 percent, intermediate risk 5 to below 10 percent, and high risk 10 percent or above. Statin decisions in contemporary guidance are generally discussed from the borderline band upward, weighing risk enhancers, patient preference and the expected benefit. For blood pressure, the 2025 AHA/ACC hypertension guideline treats a 10-year PREVENT total CVD risk of 7.5 percent or more as a marker of increased risk when deciding on medication for stage 1 hypertension. Heart failure has no equivalent treatment threshold yet, but a high 30-year heart failure estimate in a younger adult is a strong argument for attention to blood pressure, weight, glucose and smoking long before symptoms appear.
Calibration deserves a sentence of its own, because it explains why PREVENT numbers often look lower than older estimates. In validation, PREVENT showed good calibration overall (observed to predicted ratios close to 1 across risk groups), meaning that among 100 people given a 10 percent estimate, roughly 10 actually had an event. The older Pooled Cohort Equations tended to overestimate risk in contemporary populations. Neither property makes one model "true", but for shared decision-making, a well-calibrated estimate is the more honest starting point.
PREVENT versus the Pooled Cohort Equations
The Pooled Cohort Equations, released in 2013, estimate 10-year ASCVD risk only, for ages 40 to 79, using race-specific coefficients. PREVENT differs in almost every design choice: it starts at 30, drops race, adds eGFR and (for heart failure) BMI, predicts heart failure and total CVD in addition to ASCVD, and adds 30-year horizons. Discrimination, the ability to rank higher-risk above lower-risk people, was similar between the two models in validation (C statistics roughly 0.79 in women and 0.76 in men for total CVD); the practical difference is calibration and scope. One consequence clinicians notice is reclassification downward: for a Pooled Cohort Equations estimate of 7.5 percent, the median PREVENT-ASCVD estimate was 4.9 percent in women and 3.7 percent in men. If you have an older estimate, do not average the two; use the current guideline-endorsed model and discuss the change with your clinician.
PREVENT is not the only modern option, and it is not automatically the right one outside the United States. In Europe, guidelines recommend SCORE2 and SCORE2-OP, which were derived and calibrated in European populations; in the United Kingdom, QRISK is embedded in primary care. PREVENT was developed in US data, and its authors and reviewers are explicit that transportability to other populations is still being studied. The American Heart Association presents PREVENT as its standard for US primary prevention, and that is the context in which this page offers it.
What PREVENT does not include: the add-on models
The PREVENT paper also published optional add-on models. One adds haemoglobin A1c; another adds the urine albumin-to-creatinine ratio (UACR); a further model adds a social deprivation index based on postcode; and a full model combines all of them with the base predictors. These add-ons modestly improved discrimination, mainly for heart failure, and they are intended for situations where the extra measurement is clinically indicated anyway, for example A1c and UACR in diabetes or chronic kidney disease. This calculator does not implement the add-on models, and it says so rather than guessing values for missing inputs. In the published framework, when an optional measure is unavailable, the base model is the intended fallback. If you have diabetes or kidney disease and your A1c and UACR are known, your clinician can run the add-on versions through the American Heart Association's own implementation and compare.
Limitations to keep in mind
Risk equations describe groups, not destinies. A 10-year estimate of 8 percent means that among 100 similar people, about 8 would be expected to have an event within 10 years; it cannot say which 8, and it does not account for everything that matters, including family history of premature disease, lipoprotein(a), chronic inflammatory conditions, pregnancy-related risk factors such as pre-eclampsia, or coronary artery calcium, all of which guidelines treat as risk modifiers to discuss separately. The models were built from observational data, so the inputs are associations fitted for prediction, not proof that changing one input by a given amount changes personal risk by exactly the predicted difference; treatment benefit estimates come from trials, not from re-running the score. Finally, performance was weakest for heart failure prediction in some external validations, and the derivation was US-based, so treat the output as one structured input to a clinical conversation, not as a verdict.
After the number
The point of estimating risk early is that the major drivers are modifiable. Smoking cessation, blood pressure control, lipid management, glucose control, physical activity, weight management and treatment of kidney disease all shift the trajectory these equations describe, and the 30-year view shows how much of the risk accumulates while a person still feels well. If your estimate is higher than you expected, the productive next step is a consultation where the number is checked against your full history, any risk enhancers are reviewed, and a plan is agreed. If it is lower than expected, the same conversation can confirm that current measures are sufficient. Either way, bring the actual laboratory values and blood pressure readings used here, because small input errors move the output more than any difference between modern risk models.
References
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. doi:10.1161/CIRCULATIONAHA.123.067626.
Aggarwal R. Review of Predicting Risk of Cardiovascular Disease EVENTs (PREVENT). MDCalc Journal. 2026. doi:10.65357/001c.164331. (Source of the published equation reproduction used for implementation and cross-checking.)
Related tools on this site: ASCVD risk (Pooled Cohort Equations), CKD-EPI eGFR 2021, Heart Failure Survival Score.