PREVENT Equations for Assessing Cardiovascular Risk

Using demographic and laboratory data, what is the best way to assess the likelihood that a patient will experience a cardiovascular event in the next 10 or 30 years?

Recommendations regarding the use of statins, antihypertensives, and aspirin have increasingly relied on an initial assessment of cardiovascular risk.^1–4 The Framingham Risk Score was developed using data from the 1960s and 1970s in a relatively homogenous population in Massachusetts and was one of the first risk scores to quantify lifetime cardiovascular risk.⁵ It is limited by the homogeneity of the population and that type 2 diabetes mellitus was less common, tobacco use was more common, and treatment of lipid disorders and hypertension was less widespread.⁶

Since 2013, guidelines from the National Heart, Lung, and Blood Institute and the American Heart Association have recommended the use of the Pooled Cohort Equations.¹ These were developed from several cohorts that had greater diversity and more closely reflected contemporary practices regarding management of cardiovascular risk factors. However, studies have shown that the Pooled Cohort Equations tend to systematically overestimate cardiovascular risk by as much as 40%.⁷ This could result in inappropriately intensive treatment for many patients, leading to higher costs and a greater risk of adverse events.

These risk equations have other limitations, such as includeing race as a predictor, which may potentially worsen existing health disparities.⁸ They also make estimates for a 10-year time frame when decisions, such as the use of a statin, are lifelong. Finally, the equations predict the likelihood of atherosclerotic cardiovascular disease (ASCVD) events but not heart failure (HF). These limitations are addressed by the new PREVENT (Predicting Risk of cardiovascular EVENTs) equations.

The PREVENT equations were developed from a registry of 46 observational cohorts from 1992 to 2017 with individual level data for more than 6 million U.S. adults 30 to 79 years of age without ASCVD or HF at baseline.⁹ Although the registry was originally developed to study chronic kidney disease, it largely included patients without chronic kidney disease, and the mean estimated glomerular filtration rate (eGFR) was 91 mL per minute per 1.73 m². The primary outcomes were ASCVD (a composite of myocardial infarction [MI], stroke, and cardiovascular death) and HF at 10 and 30 years. The data were divided into a derivation group (n = 3,281,919) that was used to develop the risk equation and a validation sample (n = 3,330,085) that was used to test its accuracy. It was validated by the strength of the risk score. Rather than randomly dividing the sample in half, the researchers derived risk scores for 25 data sets and validated them in the remaining 21 data sets.

In addition to the conventional risk factors used in previous models, the eGFR was added (estimated without using race), and the potential value of variables were evaluated, such as A1C levels, a urine albumin-creatinine ratio, and a social deprivation index based on zip code that quantifies levels of social inequity across small areas. They developed separate equations to predict the composite of ASCVD and HF (which is labeled as CVD) and each outcome separately. Overall, the models had good accuracy, with an area under the receiver operating characteristic curve of 0.74 to 0.81, depending on the model and patient sex. Accuracy was similar for different racial and ethnic groups. Calibration, which is the match between observed and expected outcomes, was also good.