Skip to main content Skip to left navigation Skip to footer
Skip Left Navigation

Browse by Topic

Cardiovascular Disease, Risk Assessment With Nontraditional Risk Factors, 2018

* Indicates an old grade definition

Recommendations: Cardiovascular Disease: Risk Assessment With Nontraditional Risk Factors


Burden of Disease

Cardiovascular disease is the most common cause of death among adults in the United States, accounting for 1 in 3 deaths each year. Although it remains a major cause of morbidity and mortality, CVD mortality has been decreasing over time in the United States. Currently, the annual incidence of new cases of myocardial infarction and cerebrovascular accident in the United States is 580,000 and 610,000, respectively.1

Scope of Review

The USPSTF commissioned a systematic evidence review719 to update its 2009 recommendation on using nontraditional risk factors in assessment of coronary heart disease risk.20 Unlike the 2009 recommendation, the current recommendation focuses on 3 nontraditional risk factors—the ABI, hsCRP level, and CAC score. The USPSTF chose these risk factors because they have the most promising evidence base, are reliably measured, are independently associated with CVD risk or CVD events, and the prevalence and distribution of abnormal and normal values have been described in the target population. The review focused on the health benefits (reduction in CVD events, CVD mortality, and overall mortality) and harms of CVD risk assessment and treatment guided by nontraditional risk factors combined with the Pooled Cohort Equations or Framingham Risk Score compared with using either risk assessment model alone. The review also evaluated whether the use of nontraditional risk factors, when added to existing CVD risk assessment models, improves measures of calibration, discrimination, and risk reclassification.

At the same time, the USPSTF also commissioned a separate systematic evidence review to update its 2013 recommendation on screening for peripheral artery disease and CVD risk assessment with the ABI.21

Accuracy of Screening Tests

The USPSTF reviewed evidence of whether the ABI, hsCRP level, or CAC score improves calibration, discrimination, or risk reclassification when added to CVD risk assessment models using traditional risk factors. Calibration measures the agreement between observed and predicted outcomes, discrimination measures the ability to distinguish between persons who will and will not have an event, and reclassification measures the ability to (correctly) reassign persons into clinically meaningful risk strata. The USPSTF found 10 articles representing 22 cohorts for the ABI, 25 articles representing 49 cohorts for hsCRP level, and 19 articles representing 10 cohorts for CAC score, although few studies reported all 3 measures; most did not use the published versions of the Framingham Risk Score or the Pooled Cohort Equations as the base model.

In general, all cohort studies examining calibration (5 for the ABI, 9 for hsCRP level, and 8 for CAC score) found that adding 1 of these 3 nontraditional risk factors improved calibration, although preferred measures of calibration were rarely reported and only 1 study (of CAC score) used the Pooled Cohort Equations as a base model. The calibration plots available for hsCRP demonstrate that, although adding hsCRP level improved calibration for some groups, it worsened calibration for others.2223 An individual patient data meta-analysis of 18 cohorts found that the ABI improved discrimination when added to the Framingham Risk Score, but only for women.24 A separate analysis of the Multi-Ethnic Study of Atherosclerosis (MESA) cohort (which was not included in the individual patient data meta-analysis) found no improvement from adding the ABI to the Pooled Cohort Equations.25 Evidence (25 studies) for adding hsCRP was inconsistent, showing at most a small improvement in discrimination.7 In the only study that added hsCRP to the Pooled Cohort Equations (MESA), hsCRP level did not improve discrimination.25 Adding CAC score (18 studies) to various risk assessment models resulted in at least a small, and often larger, improvement in discrimination.7 However, the magnitude of improvement decreased as the discrimination of the base model improved. Four studies25-28 that added CAC score to the Pooled Cohort Equations found a very small to small improvement (0.02-0.04) to the area under the curve.

The evidence for risk reclassification was largely similar to the evidence for discrimination. Different studies used different risk strata, but those that used the Framingham Risk Score as the base model generally used less than 10% for low risk, 10% to 20% for intermediate risk, and greater than 20% for high risk, while studies using the Pooled Cohort Equations as the base model used greater than 7.5% for increased risk. In general, studies found that the ABI, hsCRP level, and CAC score tended to have positive event net reclassification (ie, more persons who had a CVD event were correctly reclassified to a higher-risk category than were incorrectly reclassified to a lower-risk category). The ABI (in women) and CAC score tended to have negative nonevent net reclassification (ie, more persons who did not have a CVD event were incorrectly reclassified to a higher-risk category than were correctly reclassified to a lower-risk category). Because only a few persons in the general population have CVD events (myocardial infarction, stroke, or CVD death) in a given period, this suggests that on balance, more persons would be inappropriately than appropriately reclassified.7

Effectiveness of Risk Assessment and Treatment

The USPSTF found only 1 study that directly assessed the potential benefit on clinical outcomes of adding 1 of these 3 nontraditional risk factors to traditional risk assessment models.29 This fair-quality randomized clinical trial (RCT) assigned asymptomatic volunteers (N = 2137) with no history of CVD to CAC scoring plus risk factor assessment counseling vs risk factor assessment counseling alone. At 4 years, there was no difference in CVD outcomes between the 2 groups; however, the study was not adequately powered to detect a difference in patient health outcomes.29 The USPSTF found no studies that assessed the incremental benefit on health outcomes of adding the ABI or hsCRP level to traditional risk factor assessment. The Viborg Vascular (VIVA) screening trial30 recently reported interim results; this trial randomized men aged 65 to 74 years to invitation for a triple screening (screening for high blood pressure, abdominal aortic aneurysm, and peripheral artery disease using the ABI) or no screening and found a decrease in mortality with screening; however, it was not possible to determine how much of the decrease was attributable to screening for peripheral artery disease and how much was attributable to screening for abdominal aortic aneurysm and high blood pressure, both of which are already recommended screenings.

The USPSTF found no trials evaluating the additional benefit of adding the ABI, hsCRP level, or CAC score to traditional risk assessment models for guiding decisions about specific interventions to prevent CVD. The USPSTF found a few studies evaluating the use of a nontraditional risk factor as a single intervention to guide decisions about specific preventive medications compared with usual care. Two RCTs (total N = 4626) compared using the ABI to guide decisions to start aspirin therapy vs usual care and found no benefit in CVD outcomes at 7 years of follow-up.3132 However, both studies used atypical cutoff points for diagnosing peripheral artery disease, and the results may not be applicable to current practice. One RCT (Justification for the Use of Statins in Prevention: An Intervention Trial Evaluating Rosuvastatin [JUPITER]; N = 17,802) compared hsCRP screening vs usual care to guide high-intensity statin therapy and found benefit at 1.9 years of follow-up in the reduction of CVD events in the hsCRP group.9 However, because the study only randomized persons with elevated hsCRP levels, it is not known whether patients with lower hsCRP levels would also have benefited from high-intensity statin therapy. Further, many of these patients met criteria for statin therapy based on traditional CVD risk assessment and would already have been candidates for treatment. One study (n = 1005) of using CAC score to guide statin therapy found no benefit at 4 years in the reduction of CVD events.33

A systematic review that addressed the effect of screening with CAC score on risk perception, adherence to medication, and behavioral therapies found only 2 studies comparing traditional CVD risk assessment vs CAC score. Neither of these studies found that screening with CAC score was superior to traditional CVD risk assessment for preventive medication use or risk factor management.5

Potential Harms of Screening and Treatment

The main potential harm of adding nontraditional risk factors to CVD risk assessment is radiation exposure from CAC score testing, although the dosage (0.4 to 2.1 mSv) is relatively low.7 More general potential harms are false-positive test results and subsequent invasive diagnostic procedures (such as coronary angiography). Three studies assessing the effect of CAC score on health care utilization found conflicting results. The Early Identification of Subclinical Atherosclerosis by Noninvasive Imaging Research (EISNER) study, an RCT of CAC score use in an academic setting, found no statistically significant increase in downstream cardiac testing and procedures.29 In contrast, a retrospective study of Medicare data found that use of CAC score increased downstream cardiac testing and procedures compared with use of hsCRP and lipid screening,34 while a second smaller observational study found no difference.35 A systematic review of 7 studies found that the prevalence of incidental findings on computed tomography for CAC score ranged from 8% to 58%. The ultimate outcomes of subsequent diagnostic procedures for these incidental findings, whether positive or negative, are not known.36 Two studies found no short-term psychological harms from use of CAC score in CVD risk assessment.3738

Treatment with aspirin and statins to prevent CVD events has some potential harms (specifically bleeding and increased incidence of diabetes, respectively), but these harms are generally accepted to be a reasonable trade-off among persons at higher risk of CVD events.39

Estimate of Magnitude of Net Benefit

The USPSTF found adequate evidence that adding the ABI, hsCRP level, and CAC score to existing CVD risk assessment models results in small improvements in discrimination and reclassification. However, the clinical meaning of these changes is largely unknown. Evidence on adding the ABI, hsCRP level, and CAC score to the Pooled Cohort Equations is sparse, which makes it difficult to infer the clinical significance of these findings. The USPSTF found inadequate evidence to assess whether treatment decisions guided by the ABI, hsCRP level, or CAC score, in addition to risk factors in existing CVD risk assessment models, leads to reduced incidence of CVD events or mortality. Few studies were available and were either underpowered or used atypical test thresholds for intervention. The USPSTF found adequate evidence to bound the harms of early detection and interventions as small. The USPSTF concludes that the current evidence is insufficient to assess the balance of benefits and harms of using the ABI, hsCRP level, or CAC score in risk assessment for CVD in asymptomatic adults to prevent CVD events.

Response to Public Comment

A draft version of this recommendation statement was posted for public comment on the USPSTF website from January 16, 2018, to February 12, 2018. Many comments expressed belief that the evidence for risk assessment with CAC score was strong enough to warrant a separate positive recommendation. Although adding CAC score to traditional risk assessment models improved discrimination and reclassification, the USPSTF found inadequate evidence that this change would translate into improved health outcomes among asymptomatic patients.

Several comments expressed concern that the USPSTF overestimated the harms associated with CAC score testing (radiation exposure, downstream testing). The USPSTF added language to clarify that it determined that the harms associated with the addition of nontraditional risk factors, including CAC score, are small in magnitude.

Several comments noted that the addition of nontraditional risk factors, especially CAC score, is useful for patients whose risk stratification is unclear or for those who fall into intermediate-risk groups. The USPSTF did not find convincing evidence that adding nontraditional risk factors to traditional risk factors improves reclassification in intermediate-risk groups. As clinical practice moves toward a single threshold for treatment, this concern may no longer be relevant in clinical decision making. Some comments also expressed belief that CAC score testing leads to better adherence to preventive therapies (ie, medications and lifestyle changes). The USPSTF carefully reviewed the available evidence and concluded that CAC score testing showed no benefit over traditional CVD risk assessment in preventive medication use or risk factor control. The USPSTF added language to address this point.

Several comments recommended including more information on the differences between the Framingham Risk Score and the Pooled Cohort Equations as well as population distribution of risk. The USPSTF included information in the Clinical Considerations section to clarify these differences and provide more information on risk in the general US population. Last, comments noted that the USPSTF assessment may not be applicable across sex, race/ethnicity, family history, and socioeconomic status. The USPSTF included language indicating the need for more studies in these subpopulations.

Other Sections:

Other Considerations:

Research Needs and Gaps

A substantial number of studies demonstrate an association between the ABI, hsCRP level, and CAC score and cardiovascular outcomes, so additional association studies are unlikely to add more information. Similarly, studies assessing nontraditional risk factors in isolation are of limited value, given that current treatment recommendations are based on risk assessment with the Framingham Risk Score or Pooled Cohort Equations. Good-quality studies comparing traditional risk assessment with traditional risk assessment plus the ABI, hsCRP level, or CAC scores are needed to measure the effect of adding nontraditional risk factors on clinical decision thresholds and patient outcomes (CVD events and mortality). Studies are especially needed in more diverse populations (women, racial/ethnic minorities, persons of lower socioeconomic status), in whom assessment of nontraditional risk factors may help address the shortcomings of traditional risk models. In addition, well-designed prospective studies reflective of real-world practice are needed to identify the downstream effects of CAC score on additional testing and procedures.


Update of a Previous Recommendation

This recommendation replaces the 2009 USPSTF recommendation.20 The previous recommendation considered the evidence on the addition of several risk factors to the Framingham Risk Score. The major change in the current recommendation is that the USPSTF evaluated the Pooled Cohort Equations in addition to the Framingham Risk Score and focused on only 3 nontraditional risk factors—the ABI, hsCRP level, and CAC score.


Recommendations of Others

The American Association of Clinical Endocrinologists’ 2017 guidelines include hsCRP level, as part of the Reynolds Risk Score, as a possible CVD risk assessment tool and to stratify borderline cases, and also states that CAC score can be useful in refining risk stratification.40 The American College of Cardiology and American Heart Association encourage using the Pooled Cohort Equations to assess 10-year risk of an initial hard CVD event (defined as stroke, nonfatal myocardial infarction, or CVD death). If risk-based treatment is still uncertain, they recommend using 1 or more of the nontraditional risk factors (including the ABI, hsCRP level, or CAC score) or family history to help clarify treatment decisions.4 The Canadian Cardiovascular Society encourages use of a modified Framingham Risk Score risk assessment tool in asymptomatic persons to assess 10-year risk of any CVD event. It recommends judicious use of secondary testing among patients for whom the need for statin therapy is unclear.41 The European Society of Cardiology uses the Systemic Coronary Risk Evaluation (SCORE) risk charts, which do not include the ABI, hsCRP level, or CAC score, to determine 10-year risk of a fatal CVD event.42 The UK National Institute for Health and Care Excellence uses the QRISK3 risk tool, which does not include the ABI, hsCRP level, or CAC score, to estimate 10-year risk of a CVD event.43 The Scottish Intercollegiate Guidelines Network (SIGN) uses the ASSIGN risk score to determine the 10-year risk of a CVD event, which does not include the ABI, hsCRP level, or CAC score.44

Prevention TaskForce

You are now leaving Prevention TaskForce website and going to
This website contains links to other federal and state agencies and private organizations. The United States Prevention Services Taskforce (USPSTF) cannot attest to the accuracy of information provided by these website links. Linking to other sites does not constitute an endorsement by USPSTF of the sponsors or the information and products presented on the sites. You will be subject to the destination site's privacy policy when you leave the Prevention TaskForce website.

Continue Cancel