Strengths and Limitations of the ASCVD Risk Score and What Should Go in the Risk Discussion

In 2013, the American College of Cardiology (ACC) and American Heart Association (AHA) published a new guideline on the assessment of cardiovascular (CV) risk.1 The expert panel that created this guideline was charged with examining the scientific evidence on risk assessment in primary prevention and developing a quantitative approach to risk assessment that could be used to guide decisions in primary prevention. The risk assessment work group (RAWG) worked in close conjunction with the cholesterol treatment panel,2 which published its guidelines at the same time. The RAWG endorsed the existing paradigm of using quantitative risk assessment in order to match the intensity of prevention efforts with the absolute risk of the patient.

After extensive review, current risk scores were found to vary widely with regard to the populations from which they were derived, risk marker inputs/covariates, and outcomes of interest. Some of these outcomes included less predictable events including elective revascularizations and heart failure. The RAWG, therefore, judged that new risk equations were needed and that they should focus on the prediction of incident atherosclerotic cardiovascular disease (ASCVD), including coronary death, non-fatal myocardial infarction and fatal or non-fatal stroke.

In deriving the new ASCVD risk equations for primary prevention, the RAWG sought data from cohorts that are community-based, representative of the broad U.S. population, include men and women form diverse race-ethnic groups, have active surveillance components for CV events, and have adequate follow-up time in order to generate robust 10-year risk estimates. Since the purpose is to predict natural history, the RAWG sought to use data that were as contemporary as possible, but without substantial downstream initiation of preventive therapies that might alter that natural history.

A number of cohorts were considered for inclusion and the ones that ultimately contributed to the derivation of the ASCVD risk equations were the Atherosclerosis Risk in Communities (ARIC), Cardiovascular Health (CHS), Coronary Artery Risk Development in Young Adults (CARDIA), and Framingham Original and Offspring Studies. Taken together, nearly 25,000 participants between ages 40 to 79 and free of CVD, contributed to the derivation dataset. Numerous traditional and novel risk factors and risk markers were assessed for inclusion in the equations using state-of-the-art statistical techniques in an unbiased fashion. The final ASCVD Pooled Cohort Equations use data on age, total and HDL-cholesterol levels, systolic blood pressure, antihypertensive therapy status, history of diabetes and current smoking status to estimate 10-year risks for ASCVD using sex- and race-specific equations.1

The advances provided by the new risk assessment approach recommended by the RAWG are important and worthy of mention. First, the ASCVD Pooled Cohort Equations provide risk estimates (for the first time) specific to African American and Caucasian men and women, groups that have different prevalences of risk factors, somewhat different risks associated with those risk factors, and varied underlying event rates of ASCVD. All of these issues reveal the importance of having sex- and race-specific equations.

Second, the RAWG focused on ASCVD rather than continuing the previous focus on only hard coronary heart disease (CHD) events. The risk for preventable strokes increases earlier than the risk for CHD in women and in African Americans; thus, the prior focus solely on CHD was undervaluing the risk, and potential benefits, in those groups, who represent more than half of the population. Third, after the initial quantitative risk assessment, the RAWG recommended a number of other factors that may be considered but are not included in the equations that may help to frame or reclassify the patient’s risk, if there is uncertainty on the part of the patient or clinician about embarking on long-term statin therapy.

These additional factors include a family history of premature CVD, a coronary artery calcium score ≥300 or >75th percentile for age, sex, and ethnicity, an ankle-brachial index <0.9, or a high sensitivity C-reactive protein (hs-CRP) >2.0 mg/L. The RAWG did not recommend routine measurement of these factors, due to lack of data regarding overall efficacy, safety and/or cost-effectiveness, but they can be useful in selected patients. It should also be noted that the interpretation of these markers is substantially enhanced if done in the context of the individual’s quantitative risk, particularly if they are in an “intermediate risk” group (e.g., 5-7.5% ASCVD risk estimate over next decade), rather than in isolation. Based on its review of the evidence, the RAWG specifically recommended against the routine measurement of carotid-intima media thickness by B-mode ultrasound for risk assessment, and called for further evidence of clinical utility on a number of other potential risk markers.

Finally, the RAWG endorsed use of the ASCVD Pooled Cohort Equations as the starting point for a personalized clinician-patient discussion to help determine the expected net clinical benefit of statin use for primary prevention in the context of the patient’s individual situation.2 Since patients and clinicians are often poor at estimating risks for ASCVD,3-5 a tool such as the new ASCVD equations is recommended for starting a clinician-patient discussion aimed at shared decision making around ASCVD prevention with statin medications in particular.

Without quantitative risk estimates, it is very difficult to align expected absolute net benefit in ASCVD risk reduction with potential absolute risk of harms from therapy to determine the expected net clinical benefit. Other factors recommended for inclusion in the clinician-patient discussion include the additional factors (such as family history of premature ASCVD) mentioned above, potential drug-drug interactions that could increase toxicity, and, crucially, patient preferences around use of medications for prevention.2

Despite the rigor with which the new equations were developed, a number of limitations of these equations (and, indeed, all quantitative risk scores) must be acknowledged. First, quantitative risk estimation is a probabilistic exercise that is by nature imprecise. No individual has a 7.5% risk for ASCVD in 10 years; it is either 0% or 100%. The probability assigned by risk equations needs to be understood as a weather forecast, so that patients and clinicians can decide if the risk is high enough to consider carrying an umbrella (i.e., taking a statin). The risk estimate should start the patient-clinician discussion described, not be viewed as a precise estimate that absolutely determines the prescription (or lack of prescription) of a medication. This limitation is of course not limited to risk equations; not everyone with coronary calcification will have a myocardial infarction (in fact the positive predictive values are quite low over five to 10 years of follow-up6), and not all women with a lesion on their mammogram have cancer. There are false positives with any approach, but the absence of any coronary calcification in a person over age 60 is associated with a very low 10-year ASCVD event rate. With a disease like CVD that affects more than half of Americans during their lifetime,7 and with very safe preventive therapies, however, this is less of a concern.

Second, the new Pooled Cohort Equations do not include “novel” risk markers that some consider important for risk assessment. However, the additional information provided by such markers has repeatedly been shown to be small, and their addition is typically only useful in intermediate risk groups rather than as universal screening tests. This is precisely what was observed in the building of the Pooled Cohort Equations.

Third, some concerns have been raised about potential over-estimation of risks for ASCVD because of the cohorts used to derive the Pooled Cohort Equations. This is an important issue for consideration. It should be noted that validation of the equations in appropriate population samples has been difficult given the limited number of contemporary U.S. cohorts that have directly measured risk factors, sufficient follow-up time (of at least 10 years), active surveillance for ASCVD to capture all relevant events, and that have not experienced substantial initiation of aspirin, statins and antihypertensive medication use in their participants that alter the natural history. Recent five-year follow-up data from the most broadly representative contemporary U.S.-based cohort study, the Reasons for Geographic and Racial Disparities in Stroke (REGARDS) study,8 suggest fair discrimination and good calibration of the equations in the relevant primary prevention population.

The truth, as always, likely lies somewhere in the middle. But any potential over-estimation of risk by the Pooled Cohort Equations was built into the recommendations for use by the cholesterol panel in setting the 7.5% threshold for the statin benefit group. Clinical trial data suggested benefit down to the level of 5% 10-year risk, but the cholesterol panel chose the higher threshold in case of over-estimation by the equations, to provide a buffer zone where the margin of net clinical benefit approaches zero, and to allow for consideration of the important other factors to be considered in the clinician-patient discussion.

As a final point, although the Pooled Cohort Equations represent a major advance by providing sex/race-specific equations for Caucasians and African Americans, there is uncertainty regarding risk assessment for other race/ethnic groups. In general, the RAWG recommends using the equations for Caucasians in those other groups, understanding that they likely under-estimate risk in South Asian Americans and over-estimate risk in East-Asian Americans and Hispanics.1

In the larger context of ASCVD prevention, an important debate is occurring as to whether we should continue risk screening to determine potential benefit from drug therapy, or switch to disease screening. Whereas there are compelling observational data to suggest that risk classification is better with screening for coronary artery calcification, for example, it is unclear whether this is a cost-effective strategy, in whom it should be applied (universally vs. risk-enriched samples), at what age, how often, and what is the acceptable level of risk from the small amount of additional radiation exposure. The parallel example here would be mammographic screening for breast cancer. Clinical trials examining the disease screening strategy for ASCVD addressing these issues would be tremendously helpful in focusing and pushing this debate forward. In the end, it will require a sea-change in our thinking about ASCVD screening, but it is likely to be a better strategy in the long run when used in appropriate subgroups. We are just not there yet. In the meantime, risk-based prevention with application of proven therapies in individuals most likely to benefit is a logical and important approach to target effective prevention to those who will benefit the most, while avoiding treatment for large swaths of the population who are less likely to achieve net clinical benefit.


  1. 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.
  2. Stone NJ, Robinson JG, 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. J Am Coll Cardiol 2014;63:2889-934.
  3. Frijling BD, Lobo CM, Keus IM, et al. Perceptions of cardiovascular risk among patients with hypertension or diabetes. Patient Educ Couns 2004;52:47-53.
  4. Petr EJ, Ayers CR, Pandey A, et al. Perceived lifetime risk for cardiovascular disease (from the Dallas Heart Study). Am J Cardiol 2014;114:53-8.
  5. Pignone M, Phillips CJ, Elasy TA, Fernandez A. Physicians' ability to predict the risk of coronary heart disease. BMC Health Serv Res 2003;3:13.
  6. Greenland P, Lloyd-Jones D. Defining a rational approach to screening for cardiovascular risk in asymptomatic patients. J Am Coll Cardiol 2008;52:330-2.
  7. Wilkins JT, Ning H, Berry J, Zhao L, Dyer AR, Lloyd-Jones DM. Lifetime risk and years lived free of total cardiovascular disease. JAMA 2012;308:1795-801.
  8. Muntner P, Colantonio LD, Cushman M, et al. Validation of the atherosclerotic cardiovascular disease Pooled Cohort risk equations. JAMA 2014;311:1406-15.

Clinical Topics: Prevention

Keywords: American Heart Association, Mitogen-Activated Protein Kinases, Primary Prevention, Risk Assessment

< Back to Listings