The Need for Accurate CVD Risk Prediction Equations

Editor's Note: Commentary based on DeFilippis AP, Young R, Carrubba CJ, et al. An analysis of calibration and discrimination among multiple cardiovascular risk scores in a modern multiethnic cohort. Ann Intern Med 2015;162:266-75.

Background

In 2013, the American College of Cardiology (ACC)/American Heart Association (AHA) published a cholesterol management guideline that recommends estimating 10-year risk for atherosclerotic cardiovascular disease (ASCVD) to help guide the decision to initiate statins for patients without established ASCVD, diabetes, and with a low-density lipoprotein cholesterol (LDL-C) of 70 to 190 mg/dL.1 With this guideline, the ACC/AHA concurrently published the "Pooled Cohort Equations" for the prediction of 10-year ASCVD risk.2 Individuals with a predicted 10-year risk for ASCVD ≥7.5% using these equations were recommended for consideration of treatment initiation with statins.1 Some, but not all, studies have suggested that the Pooled Cohort Equations over-estimate the risk for CVD.3,4 However, it is unclear whether other published risk prediction equations would be more accurate in estimating ASCVD risk.

Methods

Using data from the Multi-Ethnic Study of Atherosclerosis (MESA), the study investigators followed 4,227 adults between the ages of 50 to 74 years without ASCVD or diabetes for the incidence of ASCVD over a median of 10.2 years.5 The 10-year predicted risk for ASCVD or coronary heart disease was calculated with the following: 1) the Pooled Cohort Equations, 2) the Framingham Coronary Heart Disease Risk Score, 3) the Framingham Risk Score for cardiovascular disease, 4) the Framingham Coronary Heart Disease Risk Score modified for the Adult Treatment Panel III (ATP-III) cholesterol management guidelines, and 5) the Reynolds Risk Score. The accuracy of the five risk scores was assessed by discrimination and calibration statistics. Discrimination assesses whether a risk prediction model produces higher predicted risk for individuals who subsequently have events. Calibration provides assessment of whether a risk prediction model accurately estimates the absolute level of risk.

Results

The mean age of MESA participants in this study was 61.5 years; 54% were women, 42% were white, 26% were African American, 20% were Hispanic, and 12% were Chinese. The ability of the five risk prediction equations to discriminate between individuals who subsequently had and did not have an ASCVD or coronary heart disease event was similar and in the range considered moderate to good. However, the Framingham Risk Scores for Coronary Heart Disease and CVD, the ATP-III CHD risk score and the Pooled Cohort Equations over-estimated the coronary heart disease and ASCVD risk by 37% to 154% in men. In contrast, the Reynolds Risk Score accurately estimated ASCVD plus revascularization events in men. In women, the Framingham CVD risk score and ATP-III coronary heart disease risk score accurately estimated risk while the Framingham Coronary Heart Disease Risk Score and the Pooled Cohort Equations for ASCVD over-estimated risk and the Reynolds Risk Score under-estimated the rate of events.

Conclusion

None of the five most widely cited ASCVD and coronary heart disease risk predictions in the U.S. were accurate for both men and women.

Commentary/Perspective

Risk prediction equations are recommended to assist health care providers in weighing the benefits and risks of treatments for the primary prevention of ASCVD. The 2013 ACC/AHA guideline for cholesterol management recommended using the Pooled Cohort Equations in guiding the decision to consider statin initiation among individuals without ASCVD and diabetes, and with an LDL-C between 70 and 190 mg/dL. It has been estimated that using the Pooled Cohort Equations would more than double the number of U.S. adults recommended consideration of statin therapy for primary prevention due to a predicted risk of ASCVD ≥7.5%.6 The data by DeFilippis and colleagues make an important contribution by suggesting that none of the published ASCVD and coronary heart disease risk prediction equations, including the Pooled Cohort Equations, is accurate among both men and women.

The over-estimation of ASCVD risk with prediction equations may result in individuals with low ASCVD risk receiving statin treatment. Also, over-estimation of risk has implications beyond statin therapy. For example, the U.S. Preventive Services Task Force recommended weighing the benefits of ASCVD risk reduction with the risk for bleeding in the consideration of aspirin prophylaxis.7 Calibrated ASCVD and coronary heart disease risk prediction equations are needed to accurately weigh these benefits and risks.

The availability of very large data sets and increasing computing power is likely to lead to the development and modification of new risk prediction equations. These equations hold great promise for estimating risk for a variety of outcomes, tailored to distinct populations. However, the study by DeFilippis and colleagues demonstrates the need for the ongoing testing and refinement of risk prediction equations as they are incorporated and used in clinical practice.

References

  1. Stone NJ, Robinson J, Lichtenstein AH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation 2014;63:2889-934.
  2. Goff DC, Jr., Lloyd-Jones DM, Bennett G, et al. 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2014;63:2935-59.
  3. Muntner P, Colantonio LD, Cushman M, et al. Validation of the atherosclerotic cardiovascular disease Pooled Cohort risk equations. JAMA 2014;311:1406-15.
  4. Ridker PM, Cook NR. Statins: new American guidelines for prevention of cardiovascular disease. Lancet 2013;382:1762-5.
  5. DeFilippis AP, Young R, Carrubba CJ, et al. An analysis of calibration and discrimination among multiple cardiovascular risk scores in a modern multiethnic cohort. Ann Intern Med 2015;162:266-75.
  6. Pencina MJ, Navar-Boggan AM, D'Agostino RB, Sr., et al. Application of new cholesterol guidelines to a population-based sample. N Engl J Med 2014;370:1422-31.
  7. Wolff T, Miller T, Ko S. Aspirin for the primary prevention of cardiovascular events: an update of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med 2009;150:405-10.

Clinical Topics: Diabetes and Cardiometabolic Disease, Dyslipidemia, Prevention, Lipid Metabolism, Nonstatins, Novel Agents, Statins

Keywords: Adenosine Triphosphate, Adult, African Americans, Aspirin, Atherosclerosis, Calibration, Cholesterol, Cholesterol, LDL, Coronary Disease, Diabetes Mellitus, European Continental Ancestry Group, Hispanic Americans, Health Personnel, Hydroxymethylglutaryl-CoA Reductase Inhibitors, Incidence, Lipoproteins, LDL, Metabolic Syndrome X, Primary Prevention, Research Personnel, Risk Assessment, Risk Reduction Behavior


< Back to Listings