MESA heart risk calculator with a coronary calcium score

What the score predicts

The MESA score estimates the chance of a first coronary heart disease (CHD) event within 10 years: a heart attack, a cardiac arrest that was revived, death from CHD, or a stent or bypass for angina (McClelland 2015). It does not include stroke. The authors chose a CHD endpoint because CAC is much more strongly linked to CHD events than to stroke (McClelland 2015).

It was built in the Multi-Ethnic Study of Atherosclerosis (MESA), which recruited adults aged 45 to 84 who identified as White, African American, Hispanic or Chinese American from six US communities in 2000 to 2002, all free of known heart disease (McClelland 2015). Over about 10 years of follow-up, 422 of the 6,726 people in the analysis had a CHD event (McClelland 2015).

Why it needs a CAC score

A CAC score comes from a non-contrast heart CT scan and measures calcium in the walls of the coronary arteries, one part of atherosclerotic plaque (McClelland 2015). MESA scored it with the Agatston method, the number on most scan reports (McClelland 2015). The authors found CAC added information that the usual risk factors do not capture, and built this score so clinicians could fold a CAC result into a 10-year risk estimate (McClelland 2015).

If you have not had a CAC scan, this is not the calculator for you. Use the PREVENT heart risk calculator, the American Heart Association's newer equations for heart attack, stroke and heart failure, or the ASCVD risk calculator if your clinician uses the older Pooled Cohort Equations. Neither takes a CAC score. Whether a CAC scan makes sense for you is a question for your clinician.

How it is calculated

The paper prints two Cox models, one with risk factors only and one with risk factors plus CAC (McClelland 2015). Multiply each value by its coefficient and add them up to get A. Then the 10-year risk is 1 minus S(10) raised to the power exp(A), where S(10) is the printed baseline survival (McClelland 2015). CAC enters as the natural log of (CAC + 1) (McClelland 2015).

The paper's own example is a 70-year-old Hispanic man with treated mild high blood pressure and no other risk factors: 9.3% before CAC and 3.1% after a CAC score of zero (McClelland 2015). With the printed coefficients our tests reproduce the 3.1% exactly. For the risk-factor-only number they give 8.6%, and the MESA online calculator gives 8.7% for the same inputs today (MESA); the gap from 9.3% comes from the coefficients being rounded to four decimals in print. We also checked four more profiles against the MESA calculator and matched its with-CAC result each time (MESA).

Reading the medicine and family history terms

Taking blood pressure or cholesterol medicine raises the estimate. The authors explain this is not because the medicine causes harm: people on treatment tend to be a higher-risk group (McClelland 2015). They suggest entering "no" for medicines when the question is whether to start one (McClelland 2015). Family history in MESA meant a heart attack at any age in a parent, brother, sister or child (McClelland 2015), and current smoking meant any cigarette in the last 30 days (McClelland 2015).

How well it worked

Adding CAC raised the C-statistic in MESA from 0.75 to 0.80 (McClelland 2015). In the external checks, the with-CAC score had a C-statistic of 0.779 in the Heinz Nixdorf Recall Study and 0.816 in the Dallas Heart Study, and its average predicted risk was within half a percentage point of the observed rate in both (McClelland 2015). The risk-factor-only score was less well calibrated in those cohorts (McClelland 2015).

Limits

• It covers only the four MESA groups and ages 45 to 84 (McClelland 2015). The MESA calculator page says the same (MESA).
• It predicts coronary heart disease events only, not stroke or heart failure (McClelland 2015).
• People with very high CAC (over 400 or over 1,000) but otherwise normal risk factors were rare in MESA, so estimates for them are less certain (McClelland 2015).
• It uses single readings of cholesterol and blood pressure. A different day can give a different number.
• It does not tell you whether to take a statin or aspirin. That is a decision to make with your clinician, who knows your full history.

What to do with the number

Bring your result and your scan report to your clinician. The authors meant the score as part of the risk discussion between a clinician and patient after a CAC scan (McClelland 2015), for example to support lifestyle change or decisions about treatment (McClelland 2015).

General safety advice (not part of the source study): chest pain or pressure, shortness of breath, or pain spreading to the arm, jaw or back needs urgent care. Call 911.

Related tools: the PREVENT heart risk calculator, the ASCVD risk calculator, the Framingham risk calculator, the Reynolds risk score for men (with hsCRP) and the statin eligibility check.