10-Year Risk Calculator

One screen, nine inputs, one number

The Pooled Cohort Equations, published with the 2013 ACC/AHA cholesterol guideline (Goff et al., Circulation 2014), estimate the chance of a first hard ASCVD event within 10 years: a nonfatal heart attack, death from coronary heart disease, or a fatal or nonfatal stroke. The estimate comes from sex-specific and race-specific equations fitted on large US cohort studies, and the NHLBI describes the 40-to-79 age range as the window where the 10-year estimate applies.

This page asks for all nine inputs on one screen: age, sex, race group (African American or white/other), total cholesterol, HDL cholesterol, systolic blood pressure, whether blood pressure medicine is taken, current smoking, and diabetes. The output is one percentage, the band it falls in, and a sensitivity table that re-runs the equation with one factor changed at a time. Nothing is stored and no account is needed.

What the number is not: it is not a verdict on your arteries and not a lipid panel interpretation. Other pages on this site read the panel itself; this one treats the panel, the pressure, the smoking status and the diabetes status as inputs and concentrates on the size of the result and what moves it.

From percent to band: the 2018 guideline categories

A raw percentage invites false precision, so the 2018 AHA/ACC Multi-Society cholesterol guideline (Grundy et al., Circulation 2019) sorts 10-year risk into four bands and attaches a decision to each. The guideline document is the source for the cutoffs used by the widget below.

The bands matter because the treatment question is binary while the equation is continuous. A result of 7.4% and one of 7.6% are nearly identical numbers in the same neighborhood of risk, yet they sit on opposite sides of the intermediate cutoff. That is exactly why the sensitivity table exists: it shows how far your result sits from the nearest line, and which change would carry it across.

The arithmetic behind the percentage has three steps, and knowing them explains why the bands sit where they do. Each input is logged and multiplied by its stratum coefficient, the products are summed, and the sum is compared with the mean sum of the cohort that produced the equation. That difference is exponentiated and used as the power of the stratum's baseline survival, the share of average-risk people who stayed event-free for ten years. One minus that figure is the percentage on screen, which is why a small change in the sum can move the result by several points near the top of the scale and by a fraction of a point near the bottom.

A worked profile: 55-year-old male smoker

Take a 55-year-old white man who smokes, with total cholesterol 210 mg/dL, HDL 48 mg/dL, systolic pressure 132 mm Hg untreated, and no diabetes. The white-male equation runs age, both cholesterol values and the pressure through natural logs, applies the treated or untreated blood pressure coefficient, then adds fixed terms for smoking and diabetes. The result is 11.9%, which lands in the intermediate band, 7.5% to 19.9%.

Read as a count, 11.9% x 1,000 = 119: about 119 of every 1,000 men with exactly this profile would have a first heart attack or stroke within ten years. Read against the guideline, the number sits comfortably above the 7.5% line and far below the 20% line, so the statin conversation is appropriate but not compulsory, and the details of his other risk enhancers would settle it with his clinician.

Notice what the equation did not ask: family history, weight, exercise, diet, kidney function, sleep. Those are risk enhancers a clinician weighs around the score, not inputs to it. The nine inputs define what this tool can and cannot see.

The sensitivity table: one change at a time

The most useful question about a risk score is not the score but the distance to the nearest band line, and which single change closes it. The widget re-runs the equation four times from the base profile, changing one factor each time and everything else held fixed.

Smoking dominates. The equation carries a smoking term of 7.837 for a white man, reduced by 1.795 x 4.0073 (the natural log of 55), so smoking alone contributes about 0.64 to the internal sum, and its removal cuts the risk from 11.9% to 6.5%, a drop of 5.5 points and a full band. A 10-point systolic drop is worth 1.5 points of risk here even though 132 is untreated; on a treated pressure the coefficient is larger and so is the gain from lowering it.

The three single changes add to 5.5 + 1.5 + 1.0 = 8.0 points on paper, yet taken together they remove 6.8. The equation is nonlinear: each factor's weight depends on the others through log terms and interactions. Plan on the combined number, not the sum of the parts. The chart shows each lever against the base.

Age moves the number when nothing else does

Hold the same male smoker profile and change only the age. At 45 the equation returns 6.5%, borderline. At 55 it is 11.9%, intermediate. At 65 it is 19.4%, brushing the high cutoff. At 75 it is 28.7%, high. Nothing about his behavior changed; the equations weight age heavily because the underlying cohorts did.

Two practical points follow. First, a rising score with age is not evidence that your health declined, and comparing your score at 55 with your score at 65 mostly measures time. Second, the validated range ends at 79. The CDC's heart disease risk factor guidance applies at every age, but this particular equation was built and calibrated on people aged 40 to 79, and results outside that window are extrapolations.

The widget prints a boundary check beside every result: inside the range it is quiet, outside it states plainly that the output is an extrapolation and should be read with a clinician, not instead of one.

Who the equation does not fit

The Pooled Cohort Equations estimate risk of a first event for people who have not had one. Four groups fall outside their design. Anyone with established ASCVD, meaning a prior heart attack, stroke or diagnosed coronary disease, is already treated as high risk by default and does not need a 10-year estimate to open the statin discussion. Anyone with LDL cholesterol of 190 mg/dL or higher is likewise treated on the lipid level alone.

The third group is age: below 40 the equation was never fitted, above 79 it stops being calibrated, and the NHLBI reserves the 10-year estimator for the 40-to-79 window. The fourth is anyone whose numbers the nine inputs cannot express: severe kidney disease, inflammatory conditions, or a family history of early heart disease change the clinical picture without changing the score, which is why the guideline lists them as risk enhancers to be weighed by a clinician rather than fed into the arithmetic.

Pregnancy, recent illness and certain medicines can also distort cholesterol and pressure readings. If any of those apply, the score from this page is a rough sketch at best, and the right move is to have the measurement repeated when the picture is stable.

One more boundary is worth stating plainly: the equations predict first events, so a person already on a statin after a heart attack cannot use the score to decide whether to stop. Treatment decisions after an event follow secondary prevention guidance, and the same applies to anyone with a stent, a bypass or a documented stroke.

Two profiles at the edges of the scale

The equations are stratum-specific, so the same raw numbers do not transfer between groups. Two further runs of the widget show the spread the tool is built to expose.

The third profile is the one that deserves attention: 36.8% x 1,000 = 368, more than one in three within ten years, high band, statin therapy indicated under the guideline together with aggressive control of pressure, glucose and smoking. The African American female equation uses its own coefficients, including an age-times-pressure interaction for treated blood pressure, which is why running her numbers through a generic calculator would misstate the result.

Between 0.5% and 36.8% sits the whole purpose of the tool: the same nine questions, answered in one screen, sorted into a band, and then probed factor by factor to see which lever carries the most weight for you.