Cardiovascular risk factors vary somewhat for each specific atherosclerotic cardiovascular disease event; however, there is sufficient commonality to enable crafting of a global CVD risk-assessment instrument that accurately predicts total CVD risk and the risk of its individual components.
Framingham Heart Study statistical investigators have crafted new, sex-specific, general atherosclerotic cardiovascular disease (CVD) prediction instruments to encourage primary care physicians to make greater use of multivariable risk-assessment tools.1 Despite the availability of several validated, CVD-specific, risk-prediction algorithms, their use has lagged in the preventive primary care setting. Framingham investigators reasoned that prevention-minded physicians would prefer a user-friendly risk-assessment tool that had the propensity to include all major atherosclerotic cardiovascular events. Because these events share major predisposing risk factors, global risk assessment is a feasible and effective means for devising preventive management. While physicians might sometimes prefer to target risk assessment and preventive measures to a specific cardiovascular end point, depending, for example, on a patient’s family history, age, diabetes status, or predisposition to a particular CVD outcome due to fibrillation or heart valve disease, most physicians engaged in preventive health maintenance want to assess the risk of any major atherosclerotic event.
Framingham investigators devised 2 simple risk scoring instruments: 1 that is based on all traditional risk factors, including lipid and blood glucose laboratory tests, and another based on nonlaboratory predictors, substituting body mass index (BMI) for the laboratory tests. Simple scoring tables that make the multivariable CVD risk formulation more practical were devised (Tables 1 and 2).
Subjects and methods
To craft the new risk-assessment model, the risk of a first CVD event was evaluated in 8491 Framingham Heart Study participants, aged 30 to 74 years, who were free of CVD and attended routine examinations.1 Sex-specific multivariable risk functions were derived that incorporated age, total and high-density lipoprotein (HDL) cholesterol levels, systolic blood pressure, hypertension treatment, smoking history, and diabetes status. The performance of the general CVD algorithm for assessing global CVD risk and the individual CVD events that comprise it (coronary artery disease [CAD], stroke, peripheral artery disease, or heart failure) was assessed over a 12-year follow-up period.
A total of 1174 participants (456 women) developed a first CVD event. All traditional risk factors evaluated predicted CVD risk (multivariable adjusted, P <.001). Simple adjustments to the general CVD risk algorithms allowed estimation of the risks of each cardiovascular component. The risk profile crafted discriminated participants destined for CVD events from those who would do well. The top sex-specific quintiles of predicted risk identified 49% of men and 60% of women who experienced an initial CVD event during the follow-up period (sensitivity). Over this same period, 85% of men and 84% of women who were not in the top quintile of cardiovascular risk did not experience an event (specificity). The previous Framingham CVD risk functions2 performed less well for predicting global CVD risk than the new CVD risk prediction models. The sensitivity of the top quintile of predicted risk using the CVD risk functions were slightly lower (47% in men and 56% in women) although specificity was similar (85% in men and 83% in women).
Use of the nonlaboratory-based CVD predictor that substitutes BMI for the laboratory-based predictors enables physicians to make a risk assessment at the office, allowing prompt interaction with patients without requiring them to return for laboratory test results (Table 1). Physicians who find patients at reasonably high risk of CVD events after using nonlaboratory risk assessment can then order the necessary laboratory tests to determine whether dyslipidemia, glucose intolerance, or insulin resistance need correcting. This model also allows physicians to determine baseline risk, decide which risk factors to focus on, and to monitor improvement in patients undergoing treatment. Other risk factors not included in the general risk profile also should be taken into account when evaluating CVD risk and selecting the most efficacious treatment. These include abdominal obesity, evidence of left ventricular hypertrophy, high triglyceride levels, evidence of insulin resistance, and a strong family history of premature CVD.
Although the effects of risk factors vary from 1 specific CVD type to another, there is sufficient commonality to warrant developing a single general CVD risk reduction instrument that can accurately predict global CVD risk as well as determine the risk of individual component diseases. Evidence shows that measures taken to prevent 1 CVD outcome can also prevent the risk of the other outcomes. Therefore, use of a general CVD risk score is an attractive option for office-based primary care practices, which is the setting where most primary prevention of CVD takes place.
Formulation of a general CVD risk-assessment instrument is not a new innovation in Framingham research. Framingham investigators first formulated a general CVD risk function score in 1976.3 This risk-assessment tool was offered to the American Heart Association Council on Epidemiology and Prevention for promotion, but was deemed too complicated for physicians. Evidently, this appraisal was correct, because this tool languished in the literature for 30 years. The new global CVD risk prediction model extends and expands on the previous general CVD risk formulation and incorporates newer risk factors, such as HDL cholesterol level, which was not available in 1976. The new version is also more robust because it is based on a much larger data set. The addition of low-density lipoprotein cholesterol and triglyceride levels, diastolic blood pressure, and BMI to this assessment did not improve performance of the model.
Examining other CVD risk-assessment instruments
The key modifiable risk factors that account for a majority of CVD burden are well known. Numerous reports have shown the tendency of risk factors to cluster and have demonstrated the conjoint influence of multiple risk factors in mediating vascular disease risk.4 Based on this information, a number of prediction instruments have been devised that synthesize weighted effects of vascular risk factors to yield estimates of absolute global CVD risk in individual patients.5-9 Ridker and colleagues proposed a Reynolds risk score (RRS) for predicting CVD in women.6 The RRS incorporates family history of CVD and high-sensitivity C-reactive protein (CRP) into the profile. The Framingham study has noted a limited reliability of family history in determining CVD risk.10 Although opinion is divided on the usefulness of CRP as a necessary component of prediction instruments, CRP deserves consideration among several potential biomarkers warranting evaluation for inclusion in risk prediction formulations. Studies have shown the utility of some markers (eg, B-type natriuretic peptide and urine albumin excretion) in predicting CVD risk,11 but without data from clinical trials, it is unclear if a treatment strategy linked to CRP is superior to that using standard risk factors.11,12 Furthermore, the RRS applies to women alone and does not include CVD end points, such as peripheral arterial disease. Furthermore, its usefulness for disease-specific prediction is unknown.
The Systematic Coronary Risk Evaluation (SCORE) project formulated a CVD risk-assessment instrument (HEARTSCORE) that has been adopted by the Joint European Society and incorporated into their guidelines on CVD prevention.13 While this model has the advantage of being based on European epidemiologic data, the HEARTSCORE predicts only fatal CVD events, resulting in an underestimation of the total CVD burden. The metabolic syndrome, another multivariable risk profile, also can be used to predict CVD risk. Compared with the Framingham CAD risk score, the metabolic syndrome was found to be a significant predictor of CAD, but it was not quite as sensitive a predictor as the Framingham risk score.14 The QRISK and ASSIGN risk formulations from the United Kingdom also incorporate a family history of CVD, a person’s BMI, and social deprivation measures.15,16
CVD multivariable risk-assessment instruments derived from the Framingham Study are the most commonly recommended instruments in the United States.17 The Framingham CVD risk equations have been adopted because they were developed in a large, long-term prospective cohort of US men and women aged 30 to 74 years, have been validated in diverse population samples, and discriminate well between those destined to have a CVD event from those who are apt to escape it.
A sex-specific multivariable risk factor algorithm can be conveniently employed to assess general global CVD risk and the risk of individual atherosclerotic CVD events, including coronary and cerebrovascular events, peripheral arterial disease, and heart failure. Estimated absolute CVD event rates can be used to quantify risk and guide preventive management. The validity and transportability of the new Framingham CVD risk profile needs to be evaluated and formally compared with the performance of other CVD risk scores. After this risk-assessment instrument has been evaluated and validated, the Framingham Study anticipates setting up a Web-based risk-assessment instrument for CAD similar to the current Framingham CAD instrument on the Internet.
From the National Heart, Lung, and Blood Institute’s (NHLBI) Framingham Heart Study, National Institutes of Health (NIH). Framingham Heart Study research is supported by NIH/NHLBI Contract No. N01-HC-25195.