Cardiovascular disease risk assessment for primary prevention: Risk calculators

INTRODUCTION — Atherosclerotic cardiovascular disease (ASCVD) is common in the general population, affecting the majority of adults past the age of 60 years. As a diagnostic category, ASCVD includes four major areas:

●Coronary heart disease (CHD) manifested by fatal or nonfatal myocardial infarction, angina pectoris, and/or heart failure

●Cerebrovascular disease manifested by fatal or nonfatal stroke and transient ischemic attack

●Peripheral artery disease manifested by intermittent claudication and critical limb ischemia

●Aortic atherosclerosis and thoracic or abdominal aortic aneurysm

Although CHD is the most common manifestation of ASCVD, CHD represents only approximately 50 percent of the total number of first ASCVD events. The lifetime risk of CHD was illustrated in a study of 7733 participants, age 40 to 94, in the Framingham Heart Study who were initially free of CHD [1]. The lifetime risk for individuals at age 40 was 49 percent in males and 32 percent in females. Even those who were apparently free from disease at age 70 had a lifetime risk of 35 percent and 24 percent in males and females, respectively. The lifetime risk of CHD varies importantly with the aggregate risk factor burden [2]. (See "Atherosclerotic cardiovascular disease risk assessment for primary prevention in adults: Our approach", section on 'Lifetime risk'.)

While a general estimate of the relative risk for ASCVD can be approximated by counting the number of traditional risk factors present in a patient, a more precise estimation of the absolute risk for a first ASCVD event is desirable when making treatment recommendations for a specific individual. This topic discusses the various ASCVD risk models available for use in patients. Our approach to ASCVD risk assessment for primary prevention, including which patients should undergo ASCVD risk assessment, the approach to assessing risk, and the implications of the estimate risk on preventive therapies, is presented separately, as are discussions of the specific approach to the primary prevention treatment of various risk factors. (See "Atherosclerotic cardiovascular disease risk assessment for primary prevention in adults: Our approach" and "Overview of primary prevention of cardiovascular disease" and "Aspirin for the secondary prevention of atherosclerotic cardiovascular disease" and "Low-density lipoprotein cholesterol-lowering therapy in the primary prevention of cardiovascular disease", section on 'CVD risk assessment'.)

MULTIVARIATE RISK MODELS — A number of multivariate risk models have been developed for estimating the risk of initial ASCVD events in apparently healthy, asymptomatic individuals based upon assessment of multiple variables (table 1). While all of the risk models have advantages and disadvantages, no single risk model will be appropriate for all patients [3,4]. The choice of a specific risk model for ASCVD risk assessment should be individualized based on patient-specific characteristics (eg, age, sex, ethnicity). Most experts feel that the use of risk models that predict hard ASCVD events (ie, death, myocardial infarction [MI], stroke) are preferred over those that include other endpoints (ie, revascularization).

Typical variables used in ASCVD estimators — Prediction of ASCVD is based on a set of risk variables that are usually assessed in the outpatient setting. Most estimators use the variables of age, gender, total cholesterol, HDL cholesterol, systolic blood pressure, diabetes mellitus (DM), and current smoking (table 1).

Other variables included in various ASCVD estimators — A variety of factors beyond traditional measures are used in risk estimators (table 1). A brief summary includes:

●Blood pressure treatment (ATP III 2002, Framingham Risk Score 2008, American College of Cardiology/American Heart Association [ACC/AHA] 2013, MESA 2015, China-PAR 2016).

●Family history of CVD (SCORE CVD Death 2003, Reynolds Females 2007, Reynolds Males 2008, JBS3 2014, MESA 2015, Astro-CHARM 2018).

●Rheumatoid arthritis (JBS3 2014).

●Body mass index (QRISK 2007, JBS3 2014).

●Waist circumference (China-PAR 2016).

●Geographic region (SCORE CVD Death 2003, JBS3 2014).

●Urbanization (China-PAR 2016).

●C-reactive protein (Reynolds Females 2007, Reynolds Males 2008, Astro-CHARM 2018).

●Chronic kidney disease (JBS3 2014).

●Atrial fibrillation (JBS3 2014).

●Lipid lowering treatment (MESA 2015).

●Coronary artery calcium score (MESA 2015, Astro-CHARM 2018).

●Statin use at baseline (VA ASCVD 2020).

Key features of the most commonly used risk calculators — A variety of risk models and calculators have been derived over the years, but a select few have been most commonly used in clinical practice. Among the many reasons why specific risk calculators are more popular are ease of use, applicability to a clinician’s patient population, outcomes measured, and professional society recommendations. Some key features of the most commonly utilized risk calculators include:

●Framingham risk score (1998) and ATP III hard CHD risk score (2001) – The 1998 Framingham model used traditional risk factor categories and included data mostly from the 1970s and 1980s with long-term follow-up. Outcomes are limited exclusively to CHD. The 1998 Framingham model was the foundation for the 2002 ATP III hard CHD risk score and ATP III guidelines, which were developed in the United States. This was the primary risk score used in the United States from 2002 to 2013. (See 'Framingham risk score (1998)' below and 'ATP III hard CHD risk score (2002)' below.)

●SCORE CVD death risk score (2003) – The 2003 SCORE CVD risk model included data from a variety of European countries, and included a hard endpoint of CVD death but did not address nonfatal CVD events or morbidity. The 2003 SCORE CVD risk calculator has been recommended for use in Europe beginning with the 2007 European Guidelines [5]. (See 'SCORE CVD death risk score (2003)' below.)

●ACC/AHA pooled cohort hard ASCVD risk score (2013) – The 2013 ACC/AHA pooled cohort hard ASCVD risk score was developed from several United States cohorts, and includes different calculators for White individuals and for Black individuals. Outcomes are limited to both fatal and nonfatal CHD and stroke. Beginning in 2013, the ACC/AHA pooled cohort hard ASCVD risk score was recommended for use in the United States [6,7]. (See 'ACC/AHA pooled cohort hard ASCVD risk calculator (2013), updated algorithm 2018' below.)

●QRISK, QRISK2, and JBS3 (2014) – The QRISK and QRISK2 ASCVD risk calculators used the Framingham (1998) and ATP III (2002) models as a baseline, then added multiple additional medical, ethnic, and socioeconomic variables. JBS3 included the same variables as QRISK/QRISK2 but extended the risk prediction from 10 years to 30 years, and included a variety of fatal and nonfatal ASCVD endpoints. These models were developed in the United Kingdom and, beginning in 2014, are recommended for use by the Joint British Guidelines [8]. (See 'JBS3 risk score (2014)' below.)

Framingham risk score (1998) — The original Framingham risk score (1998), published in 1998, was derived from a largely White population of European descent [9]. Subsequent studies have suggested that the Framingham risk score performs well for prediction of CHD events in Black and White females and males [10].

Endpoints assessed in Framingham CHD risk score (1998)

●CHD death

●Nonfatal MI

●Unstable angina

●Stable angina

ATP III hard CHD risk score (2002) — The Framingham risk score was modified (2002) by the third Adult Treatment Panel (ATP III) for use in their recommendations for screening for and treatment of dyslipidemia [11]. The modifications include elimination of diabetes from the algorithm, since it was considered to be a CHD equivalent; broadening of the age range; and inclusion of hypertension treatment and age-specific points for smoking and total cholesterol.

Endpoints assessed in ATP III hard CHD risk score (revised 2002)

●CHD death

●Nonfatal MI

Framingham general risk score (2008) — The original 1998 and revised 2002 Framingham risk scores do not include all of the potential manifestations and adverse consequences of atherosclerosis, such as stroke, transient ischemic attack, claudication, and heart failure (HF; although manifestations of aortic atherosclerosis were omitted). These patient-important vascular outcomes were included in the development of the 2008 Framingham general cardiovascular risk score, which was shown to have reliable predictive ability (calculator 1 and calculator 2) [12]. The estimated risk of developing a cardiovascular event will be higher with this risk score than with those that predict only CHD events.

Endpoints assessed in Framingham general risk score (2008)

●CHD death

●Nonfatal MI

●Coronary insufficiency or angina

●Fatal or nonfatal ischemic or hemorrhagic stroke

●Transient ischemic attack

●Intermittent claudication

●HF

SCORE CVD death risk score (2003) — SCORE, which is recommended in the 2007 European Society of Cardiology guidelines on cardiovascular disease prevention in clinical practice, included data on more than 200,000 patients pooled from cohort studies in 12 European countries [5,13]. A unique aspect of SCORE is that separate risk scores were calculated for high- and low-risk regions of Europe. The predictive value of SCORE was high in each component study cohort.

SCORE differs from the earlier Framingham risk models (and others) in two important ways: it estimates the 10-year risk of any first fatal atherosclerotic event (eg, stroke or ruptured abdominal aneurysm), not just CHD-related deaths, and it estimates CVD mortality.

Endpoints assessed in SCORE CVD death risk estimator (2003)

●CVD death (including CHD, arrhythmia, HF, stroke, aortic aneurysm, and peripheral vascular disease)

QRISK and QRISK2 — The QRISK and the updated QRISK2 algorithms were developed to predict cardiovascular risk in patients from different ethnic groups living in England and Wales [14,15]. The QRISK2 algorithm included risk predictors used in the modified Framingham/ATP III model, as well as ethnicity, socioeconomic status, family history, and other medical variables such as diabetes, chronic renal disease, atrial fibrillation, and rheumatoid arthritis. QRISK2 more accurately identified those at risk than the modified Framingham/ATP III model in this population.

Endpoints assessed in QRISK ASCVD risk estimator (2007)

●CHD death

●Nonfatal MI

●Coronary insufficiency or angina

●Coronary revascularization

●Fatal or nonfatal stroke

●Transient ischemic attack

●Intermittent claudication

Reynolds risk score for females (2007) — The Reynolds risk score for females was developed from a prospective cohort of nearly 25,000 American females without diabetes [16]. The primary differences between the Reynolds risk score and most other risk estimations are the inclusion of family history of MI and high-sensitivity C-reactive protein (hs-CRP) as variables in the risk calculator. (See "C-reactive protein in cardiovascular disease".)

Endpoints assessed in Reynolds risk score for females (2007)

●Cardiovascular death

●Nonfatal MI

●Nonfatal stroke

●Coronary revascularization

Reynolds risk score for males (2008) — Similar to the Reynolds risk score for females, the Reynolds risk score for males was developed from a prospective cohort of over 10,000 American males without diabetes and also included family history of MI and hs-CRP as variables in the risk calculator [17]. Unlike the Reynolds risk score for females, hemoglobin A1c level was not included as a variable in the Reynolds risk score for males. (See "C-reactive protein in cardiovascular disease".)

Endpoints assessed in Reynolds risk score for males (2008)

●Cardiovascular death

●Nonfatal MI

●Nonfatal stroke

●Coronary revascularization

ACC/AHA pooled cohort hard ASCVD risk calculator (2013), updated algorithm 2018 — Several cohorts of patients were used to develop the 2013 ACC/AHA cardiovascular risk calculator (calculator 3), the first risk model to include data from large populations of both White and Black American patients [6]. The model includes the same parameters as the 2008 Framingham general ASCVD model, but in contrast to the 2008 Framingham model includes only hard endpoints (fatal and nonfatal MI and stroke). However, while the calculator appears to be well-calibrated in some populations similar to those for which the calculator was developed (REGARDS), it has not been as accurate in other populations (Rotterdam) [7,18-20].

A potential limitation of the ACC/AHA calculator is that family history of premature ASCVD is not included in the model. This may underestimate risk in patients with very strong family histories of cardiovascular events. Additionally, the ACC/AHA includes DM only as a yes/no question. Issues that may affect risk with DM include patient age, sex, other cardiovascular risk factors, duration of DM, and whether the patient has type 1 or type 2 DM. A downloadable calculator that incorporates these variables is available for patients with type 2 DM from the UK Prospective Diabetes Study.

Endpoints assessed in ACC/AHA pooled cohort hard ASCVD risk calculator (2013), updated 2018

●CHD death

●Nonfatal MI

●Fatal stroke

●Nonfatal stroke

In 2018, this algorithm was updated [21]. The basic ASCVD risk scoring was maintained, but intermediate risk scores were redefined.

JBS3 risk score (2014) — The Joint British Societies (JBS) released a new risk calculator in 2014 which is based on the QRISK lifetime cardiovascular risk calculator and incorporates many of the same variables from the original QRISK and QRISK2 scores [8]. However, the JBS3 risk calculator extends the assessment of risk beyond the 10-year window of most prior risk estimators and allows for the estimate of "heart age" and the assessment of risk over longer intervals (eg, 50 years for a 45-year-old patient, 30 years for a 65-year-old patient, etc). The JBS calculator also allows for the estimated impact of lifestyle modifications on future risk (ie, the impact of smoking cessation on ASCVD risk), but these data are from statistical models and have not been externally validated. A link to this calculator can be found at the JBS3 risk website.

Endpoints assessed in JBS3 risk estimator (2014)

●CHD death

●Nonfatal MI

●Coronary insufficiency or angina

●Coronary revascularization

●Fatal or nonfatal stroke

●Transient ischemic attack

●Intermittent claudication

MESA risk score (2015) — The Multi-Ethnic Study of Atherosclerosis (MESA) published a risk score in 2015, which was validated in two separate prospective cohort groups [22]. While many of the traditional risk factors are also part of the MESA risk score, this risk estimator also includes multiple ethnic backgrounds (calculator 4). The MESA risk score also incorporates coronary artery calcium (CAC) score to further classify the patients, although an estimate of risk can also be made without entering the CAC score for patients.

Endpoints assessed in MESA risk score (2015)

●CHD death

●Nonfatal MI

●Resuscitated cardiac arrest

●Coronary revascularization in patient with angina

China-PAR risk predictor (2016) — The Predication for ASCVD Risk in China (China-PAR) project published a risk score in 2016, which was validated in two separate prospective Chinese cohort groups [23]. While many of the traditional risk factors are part of the China-PAR risk estimator, this risk estimator also includes waist circumference and geographic region within China for both males and females, as well as urbanization and family history for males. The China-PAR risk estimator was found to be more accurate for persons living in China than the ACC/AHA pooled cohort equations, which tended to overestimate risk in Chinese males while underestimating risk in Chinese females. The China-PAR model has yet to be evaluated for adults living outside China.

Endpoints assessed in China-PAR (2016)

●CHD death

●Nonfatal MI

●Fatal or nonfatal stroke

PREDICT CVD risk predictor (2018) — The PREDICT CVD risk predictor was developed in a multiethnic New Zealand population of over 400,000 patients with no prior history of cardiovascular or kidney disease who were recruited between 2002 and 2015 [24]. The PREDICT CVD risk predictor includes many of the traditional risk factors, along with predictors based on family history and ethnicity. In contrast to nearly all other risk predictors that offer estimates of 10-year or lifetime CVD risk, the PREDICT CVD risk predictor estimates risk at five years. The PREDICT CVD risk predictor was significantly more accurate in this New Zealand population than the ACC/AHA pooled cohort equation, which overestimated the risk by between 40 and 60 percent.

Endpoints assessed in PREDICT (2018)

●Cardiovascular death (including MI, stroke, or atherosclerotic aneurysm)

●Nonfatal MI

●CHD revascularization, both surgical and percutaneous

●Nonfatal stroke

●Transient ischemic attack

●Peripheral vascular disease including revascularization procedures

●HF and cardiomyopathy (unless specified as nonischemic in origin)

Astro-CHARM risk calculator — The Astro-CHARM ASCVD risk calculator was derived from study participants in three cohorts (MESA, Dallas Heart Study, and Prospective Army Coronary Calcium Project; 7382 persons; 55 percent male; 45 percent White individuals) in which family history, coronary artery calcium score, and high-sensitivity CRP were obtained at baseline and incorporated into the risk calculator, along with the traditional risk factors [25]. While showing a small improvement in net reclassification of risk, the Astro-CHARM study involved relatively young persons (mean age 51 years) at relatively low risk of CVD (only 55 events seen among 2057 patients in the validation cohort).

Endpoints assessed in Astro-CHARM

●CHD death

●Nonfatal MI

●Fatal or nonfatal stroke

VA-ASCVD risk calculator — The five-year risk of developing an initial atherosclerotic cardiovascular disease (ASCVD) event or ASCVD mortality was estimated in 1.6 million US veterans. The 2013 Pooled Cohort Equations overestimated ASCVD risk approximately twofold for each group: White females, Black females, White males, and Black males. Baseline statin therapy was associated with a 7 percent lower relative risk of ASCVD and a 25 percent lower relative risk for ASCVD death [26].

Endpoints assessed in VA-ASCVD

●MI

●Fatal or nonfatal stroke

●ASCVD death

Comparison among different risk scores — An important component of multivariate risk models for the estimation of ASCVD risk is that many of the risk factors (eg, age, hypertension, serum LDL-cholesterol) are recognized as producing a graded increase in risk (figure 1) [27]. In addition, these models estimate risk of an individual patient over the next 10 years, even though the risk models have been derived from large population-based studies.

Several studies have suggested that the Framingham criteria either overestimate or underestimate the risk of initial CHD events in other populations such as Japanese American and Hispanic males, Chinese patients, Native American females, individuals in various European countries (eg, the United Kingdom and France), and African American males and females, as well as individuals older than age 85 years [3-5,10,13,23,24,28-35]. These differences are partly explained by the research methods used, adjudication procedures, time intervals studied, and calendar year of the baseline evaluations. Multiple models, including SCORE and QRISK2, have been developed in an attempt to provide better predictive accuracy for individuals in European countries [5,13-15,31-33].

How a risk score performs is largely dependent on population characteristics along with the presence or absence of primary preventive therapies to address relevant risk factors.

●In an attempt to assess the performance of risk scores in a diverse population, five risk scores (1998 Framingham, 2002 ATP III, 2008 Framingham, Reynolds risk score, and 2013 ACC/AHA score) were calculated for 4227 racially diverse participants of the MESA cohort (42 percent White, 26 percent Black, 20 percent Hispanic, 12 percent Chinese) ages 50 to 74 years and without ASCVD or diabetes at baseline (follow-up 10.2 years) [3]. When comparing the predicted and observed rates of ASCVD, four of the five risk scores significantly overestimated the 10-year ASCVD risk (between 25 and 115 percent), with only the Reynolds risk score slightly underestimating risk (3 percent underestimation).

●In a separate study comparing the 2013 ACC/AHA calculator with the ATP III risk estimator among 2435 participants in the Framingham offspring and third-generation cohorts who were followed for over nine years, the 2013 ACC/AHA calculator identified significantly more patients as statin eligible and more accurately identified patients who were at risk for incident ASCVD [36].

●A separate study assessing the accuracy of the 2013 ACC/AHA calculator in six cohorts (the original four used to develop the pool cohort equations plus MESA and the Jackson Heart study) found that the 2013 ACC/AHA calculator overestimated risk by an average of 20 percent [37].

A variety of issues may underlie these differences, including secular changes in risk factors, greater use of cardiovascular prevention medicines and strategies over the past two decades, and a general decline in ASCVD morbidity and mortality in more recent times. The exact reasons for overestimation of risk are not known. In addition to these findings reinforcing the importance of being familiar with multiple risk scores and choosing the most applicable risk score for each individual patient (based upon the patient's unique characteristics), the potential for overestimation of ASCVD risk should be recognized in the discussion of risk and the decision-making process regarding therapies aimed at primary prevention.

Limitations of current vascular disease prediction models — The following represent important limitations to current risk models:

●Risk assessments that stratify patients according to the number of defined risk factors can identify high-risk persons, but they tend to falsely reassure persons deemed to be at low risk who may have multiple marginal abnormalities [2,38]. (See "Overview of established risk factors for cardiovascular disease", section on 'Prevalence and impact of cardiovascular risk factors'.)

●Risk models identify patients who are more or less likely to develop ASCVD within a defined period (eg, 10 years for CHD in the Framingham model). This approach does not consider lifetime risk, which might be substantially higher and amenable to aggressive risk factor reduction [2,39-41]. However, the 2013 ACC/AHA guideline on the assessment of cardiovascular risk calculators does offer an estimate of lifetime risk [6]. (See "Atherosclerotic cardiovascular disease risk assessment for primary prevention in adults: Our approach", section on 'Lifetime risk'.)

●Some risk scores have been found to overestimate 10-year risk of ASCVD. (See 'Comparison among different risk scores' above.)

●Most ASCVD risk scores do not account for an individual’s socioeconomic status or geographic/neighborhood location, which can significantly impact the accuracy of one’s estimated ASCVD risk [42].

●There is significant variability in predicted risk when multiple risk scores are applied to the same population, particularly when used in a population that is different from the validation cohort of the risk score [3,4].

●The severity and frequency of the first vascular disease event has decreased over the last 40 years [43].

●The relative effects of traditional risk factors differ according to the vascular disease outcome being evaluated [43].

●Some risk models do not include patient-important ASCVD outcomes such as stroke, HF, or development of symptomatic peripheral artery disease.

ADDITIONAL RISK ASSESSMENT — For most patients who require ASCVD risk assessment, we do not routinely perform additional testing above and beyond that which is part of the standard risk calculator used (eg, lipid values, blood pressure reading, etc). However, in select patients, such as those whose estimated risk is near a threshold to treat (or not treat) or patients whose risk factor(s) may not be adequately included in the standard risk calculators (eg, strong family history of premature ASCVD in multiple relatives), additional risk assessment may be helpful to guide treatment decisions.

There is likely a small added benefit to assessing genetic information for the purpose of ASCVD risk stratification, and cost benefit of these approaches are not well established [44].

The use of evaluations beyond traditional risk factor assessment has been considered by multiple major professional societies:

●In 2018, the United States Preventive Service Task Report issued a recommendation statement (and associated evidence report) in which it concluded that there was insufficient evidence to assess the balance of benefits and harms of adding high-sensitivity C-reactive protein (hs-CRP), coronary artery calcium (CAC) score, or ankle-brachial index (ABI) to standard risk assessment [45,46].

●In 2013, the American College of Cardiology/American Heart Association issued a Risk Assessment report in conjunction with new cholesterol guidelines [6]. The opinion of this expert committee was that quantitative risk assessment should occur first, and if a risk-based treatment decision is uncertain, assessment of family history of ASCVD, hs-CRP, CAC score, or ABI may be considered to inform treatment decision making. The committee did not recommend routine measurement of carotid intima-media thickness for risk assessment for a first ASCVD event. The committee also did not recommend use of apolipoprotein B (apoB), chronic kidney disease, albuminuria, or cardiorespiratory fitness evaluation for risk assessment for a first ASCVD event.

In a 2016 study that assessed the impact of 13 negative risk markers (including CAC score of 0, normal carotid intima-media thickness, CRP <2 mg/L, B-type natriuretic peptide <100 pg/mL, and negative family history, among others) on the future risk of all ASCVD and CHD in 6814 participants in the Multi-Ethnic Study of Atherosclerosis (MESA) without known ASCVD at baseline, a CAC score of 0 was the strongest modulator of future risk and had the greatest impact on net reclassification of risk [47]. In a 2020 study which assessed the impact of baseline CAC score and high-sensitivity cardiac troponin (hs-cTnT) levels on ASCVD risk among the MESA cohort, patients with undetectable hs-cTnT levels had similar rates of incident ASCVD as did patients with CAC = 0 (5.2 versus 5.0 per 1000 person-years, respectively), while patients with both undetectable hs-cTnT and CAC = 0 had the lowest event rates (2.8 per 1000 person-years) [48]. Future studies that incorporate additional risk factors into the traditional risk models may further guide clinicians in the assessment of individual ASCVD risk. (See "Coronary artery calcium scoring (CAC): Overview and clinical utilization", section on 'Patients we screen'.)

Specific lipid subfractions contribute differently to predictions of overall CHD risk (see "Lipoprotein classification, metabolism, and role in atherosclerosis"). Lipid subfractions are generally classified by relative density, with low-density lipoprotein (LDL) and high-density lipoprotein (HDL) most commonly measured. Many additional measurements are available, including further subfractions of LDL based on density [49] and measurement of various apolipoproteins (apo) A-1 and B.

There are numerous studies examining the power of each of these various lipoproteins in predicting CHD risk. There is generally a consensus that non-HDL cholesterol (total cholesterol minus HDL-C) or the total cholesterol to HDL-C ratio is a better predictor of CHD risk than LDL cholesterol (LDL-C) alone [50,51]. The literature is largely consistent on this matter:

●In an analysis of data from the Framingham study, the ratio of total cholesterol to HDL-C significantly affected the risk for CHD events (hazard ratio [HR] 1.39, 95% CI 1.22-1.58), while LDL-C was not a significant predictor of such events (HR 1.11, CI 0.97-1.27) [52].

●In an individual patient meta-analysis of 302,430 people, the ratio of non-HDL-C to HDL-C was a better predictor of CHD risk than LDL-C (HR 1.49 versus 1.38) [53]. The authors concluded that measurement of total cholesterol and HDL-C was simpler and provided more accurate risk estimation than measuring LDL-C.

●In a study examining the predictive power of lipid levels on CHD events in patients on statins, the predictive power of non-HDL-C was superior to that of LDL-C (p = 0.002) [54].

While LDL subfractions may perform better than LDL-C alone, they have not been adequately assessed relative to existing risk scores [55]. The lack of additional predictive power of LDL-C is reflected in most predictive tools for CHD risk, which incorporate HDL-C and total cholesterol, but not LDL-C. (See "Atherosclerotic cardiovascular disease risk assessment for primary prevention in adults: Our approach", section on 'Choosing a risk calculator'.)

There is ongoing debate about the utility of apolipoprotein markers in the prediction of CHD risk. There is particular interest in the use of apoB and apolipoprotein A-1 (apoA-1). Overall, apoB and non-HDL cholesterol perform relatively similarly in terms of predicting CHD risk [56]. Use of the apoA-1/apoB ratio has also been examined. In an analysis of Framingham data, for example, multivariate analyses showed that ratios of total cholesterol/HDL, LDL/HDL, and apoA-1/apoB all performed similarly, and all were superior to any single measure alone [52]. Similarly, in an individual patient meta-analysis, the incorporation of apoB and apoA-1, lipoprotein(a), and lipoprotein-associated phospholipase A₂ mass added only slightly to CHD prediction [57].

The discussion of these additional tools for risk assessment is provided separately. (See "Overview of established risk factors for cardiovascular disease", section on 'Family history' and "Coronary artery calcium scoring (CAC): Overview and clinical utilization" and "Noninvasive diagnosis of upper and lower extremity arterial disease", section on 'Ankle-brachial index' and "Carotid intima-media thickness" and "C-reactive protein in cardiovascular disease".)

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Assessment of cardiovascular risk".)

SUMMARY AND RECOMMENDATIONS

●Introduction – Atherosclerotic cardiovascular disease (ASCVD) is common in the general population, affecting the majority of adults past the age of 60 years. While a general estimate of the relative risk for ASCVD can be approximated by counting the number of traditional risk factors present in a patient, a more precise estimation of the absolute risk for a first ASCVD event is desirable when making treatment recommendations for a specific individual. (See 'Introduction' above.)

●Multivariate risk models – A number of multivariate risk models have been developed for estimating the risk of cardiovascular events in apparently healthy, asymptomatic individuals based upon assessment of multiple variables (table 1). While all of the risk models have advantages and disadvantages, no single risk model will be appropriate for all patients. The choice of a specific risk model for ASCVD risk assessment should be individualized based on patient-specific characteristics (eg, age, sex, ethnicity). However, our experts feel that the use of risk models that predict hard events (ie, death, myocardial infarction, stroke) are preferred over those that include other endpoints (ie, revascularization). (See 'Multivariate risk models' above.)

●Limitations – All of the risk models are subject to limitations, the most important of which is the ASCVD risk prediction over a defined period of time (usually 10 years) which may poorly characterize a person's long-term ASCVD risk, particularly in younger patients. Many individuals with a low 10-year risk, as calculated using the above models, will still have a high lifetime risk, as the largest driver of risk in most risk calculators is age. (See 'Limitations of current vascular disease prediction models' above and "Atherosclerotic cardiovascular disease risk assessment for primary prevention in adults: Our approach", section on '10-year risk versus 30-year (lifetime) risk'.)

●Our approach – Our comprehensive approach to ASCVD risk assessment (algorithm 1) is discussed separately. The use of the risk models in decision making for the management of hypercholesterolemia or initiation of aspirin therapy for primary prevention is discussed elsewhere. (See "Atherosclerotic cardiovascular disease risk assessment for primary prevention in adults: Our approach" and "Aspirin for the secondary prevention of atherosclerotic cardiovascular disease" and "Low-density lipoprotein cholesterol-lowering therapy in the primary prevention of cardiovascular disease".)

●Certain individuals with established ASCVD or ASCVD risk equivalents are known to be at high risk of recurrent cardiovascular events and should be treated with appropriate secondary prevention measures. (See "Prevention of cardiovascular disease events in those with established disease (secondary prevention) or at very high risk".)