Curriculum in cardiologyConcept and usefulness of cardiovascular risk profiles☆
Section snippets
Evolution of multivariable risk formulation
Extensive epidemiological research in the past half century at the Framingham Study and elsewhere was devoted to the creation of mathematical models for predicting CVD.3, 4, 5, 6, 7, 8, 9, 10 The study was a leader in defining and quantifying the impact of risk factors.11 This ongoing 50-year study has been engaged in making periodic measurements of standard and novel risk factors in the original cohort and their offspring, and observing them for initial and recurrent CVD events for extended
Framingham perspective on risk stratification
When the Framingham Study was initiated in 1949, the prevailing concept was that we should seek out a single essential cause sufficient to produce CVD. It soon became apparent that it was only going to be possible to suggest guilt by association, which gave rise to the concept of predisposing “risk factors,” a term coined in the Framingham Study.17 In every instance, the CVD hazard imposed by any particular major risk factor varied substantially in relation to the burden of risk factors that
Mathematical predictive models
Because of the multifactorial predisposition to CVD, and the need to determine and quantify the net and joint contribution of predisposing risk factors, multivariable risk formulations were needed. The first of these was devised in the 1960s and subsequently followed by risk formulations devised on the basis of longer periods of follow-up, better predictive variables, and increasingly sophisticated statistical methods, including logistic regression, Cox proportional hazards regression, and
Framingham study disease-specific CVD risk profiles
Epidemiological research in the Framingham Study identified, for either sex and specified ages, a set of major correctable risk factors that impacted strongly and independently on the rate of development of each of the major clinical manifestations of CVD.24 These included, in addition to age and sex, blood lipid levels, blood pressure, glucose tolerance, smoking, and left ventricular hypertrophy. Although all the standard risk factors contribute powerfully to coronary disease, for
Coronary risk assessment
Statistical prediction models for coronary disease were largely made on the basis of the logistic regression model in the past; now, Cox regression and Weibull models have replaced this.7, 9, 16 Framingham Study coronary risk formulations included age, sex, blood pressure, total and HDL cholesterol levels, smoking, diabetes mellitus, and left ventricular hypertrophy.7 Age, total cholesterol level, and HDL cholesterol level were used in the equations as continuous variables, whereas smoking,
Stroke risk profile
Stroke, the most feared of the atherosclerotic sequelae of CVD risk factors, becomes a serious hazard in patients aged ≥65 years. A number of CVD risk factors and cardiac conditions have been identified that independently contribute to stroke rates. With the Cox proportional hazards regression model, these were formulated into a stroke-risk profile, including age, sex, systolic blood pressure, diabetes mellitus, cigarette smoking, echocardiography-left ventricular hypertrophy, coronary and
Peripheral artery disease
Peripheral artery disease is a hazard in patients aged ≥65 years and is also an ominous harbinger of other CVD events. Persons with intermittent claudication have a 2-to 4-fold increased risk for coronary disease, stroke, or heart failure.25 The significant independent risk factors found to be useful in devising a predictive risk profile for peripheral artery disease include age, sex, blood pressure, diabetes mellitus, cigarettes, cholesterol level, and coronary disease status.21 By using a
Heart failure
Heart failure is a terminal stage of cardiac disease, with a survival experience little different from cancer. When overtly manifested, the median survival period in patients with this condition in the Framingham Study was only 1.7 years for men and 3.2 years for women, and sudden death was a prominent feature of the mortality rate.26 Recent declines in CVD mortality have not been accompanied by a reduction in the prevalence of heart failure.27 Patients who are at high risk for heart failure
Risk assessment in patients undergoing treatment
Statistical models for estimating CVD risk from population data initially made the unwarranted assumption that risk factors in patients undergoing treatment carry a risk of CVD identical to that in patients who are not undergoing treatment. At that time, this was of no concern because available treatment was not very effective or widely used. As the Framingham cohort was observed for 5 decades, more and more effective therapy for controlling risk factors was introduced. This was recognized as
Risk appraisal for subsequent CVD
In the Framingham Study, calculation of risk appraisal functions for men and women with prior CHD or stroke at the time of examination indicates that for men, only age, log ratio of total/HDL cholesterol level, and diabetes mellitus remain in the model. In women, log systolic blood pressure and smoking remain in the model with the other variables. Risk factor scoring sheets for men did not include systolic blood pressure or cigarette smoking. For women, these variables are included (Table V).
General CVD risk profile
Although the impacts of risk factors vary from one atherosclerotic CVD entity to another, there appears to be a sufficient commonality to warrant considering all clinical varieties as a unit. Also, 1 CVD event tends to presage another, so a patient with coronary disease, for example, is substantially more likely to have a stroke, peripheral artery disease, or heart failure than patients without coronary disease.6, 25 The prospect that CVD as a group may be effectively predicted from a
Transportability of risk formulations
Caution is necessary in generalizing multivariable risk from the Framingham Study to dissimilar population samples, particularly patients with a low CHD incidence. The Framingham Study multivariable coronary disease risk factor model was tested in a variety of other population samples and found to be reasonably accurate, except when applied in areas in which the incidence of coronary disease is quite low.29, 30, 31, 32, 33 However, even in these low-incidence populations, it was possible to
Preventive implications
It is now recognized that atherosclerotic CVD is attributable to a variety of factors and has several clinical manifestations. In every instance, the hazard of a particular risk factor varies widely depending on the burden of associated accompanying risk factors. Almost half of CVD events occur in the tenth of the population at highest multivariate risk.
Single risk factor detection and correction may be worthwhile for prevention of CVD on a population basis, but is inefficient on an individual
References (45)
- et al.
A multivariate analysis of the risk of coronary heart disease in Framingham
J Chronic Dis
(1967) - et al.
A general cardiovascular risk profilethe Framingham Study
Am J Cardiol
(1976) - et al.
Multiple risk functions for predicting coronary heart diseasethe concept, accuracy and application
Am Heart J
(1982) The use of classification and regression trees in clinical epidemiology
J Clin Epidemiol
(2001)- et al.
Internal validation of predictive modelsefficiency of some procedures for logistic regression analysis
J Clin Epidemiol
(2001) - et al.
Primary and subsequent coronary risk appraisalnew results from the Framingham Study
Am Heart J
(2000) - et al.
Atherosclerosis risk factors
Pharmacol Ther
(1987) - et al.
Representativeness of the Framingham risk model for coronary heart disease mortalitya comparison with a national cohort study
J Chronic Dis
(1987) Lp(a) lipoprotein in cardiovascular disease
Atherosclerosis
(1994)- et al.
Homocysteine and coronary atherosclerosis
J Am Coll Cardiol
(1996)
Prevention of coronary heart disease in clinical practicesummary of recommendations of the second joint task force of European and other societies on coronary prevention
J Hypertens
Estimation of the probability of an event as a function of several independent variables
Biometrika
An updated coronary risk profilea statement for health professionals
Circulation
Probability of strokea risk profile from the Framingham Study
Stroke
Prediction of coronary heart disease using risk factor categories
Circulation
The Munster Heart Study (PROCAM)results of follow-up at 8 years
Eur Heart J
Artificial neural networksopening the black box
Cancer
Validation of the Framingham coronary heart disease prediction scoresresults of a multiple ethnic groups investigation
JAMA
Factors of risk in the development of coronary heart disease—six-year follow-up experience; the Framingham Study
Ann Intern Med
Cited by (118)
Identification by Ultrasound Evaluation of the Carotid and Femoral Arteries of High-Risk Subjects Missed by Three Validated Cardiovascular Disease Risk Algorithms
2015, American Journal of CardiologyCitation Excerpt :The incidence of events has decreased substantially since their recognition and the implementation of successful risk reduction strategies.1 Nevertheless, it has also been established that while successful in a population analysis, these predictors have been less successful in identifying subjects at personal risk for a cardiovascular event.15–17 A recent analysis of the Atherosclerosis Risk in Communities (ARIC) population has confirmed the decrease in CVD risk attributable to traditional risk factors particularly in white men.3
Electrocardiographic changes improve risk prediction in asymptomatic persons age 65 years or above without cardiovascular disease
2014, Journal of the American College of CardiologyThe impact of financial crisis on coronary artery disease burden in Greece
2019, Hellenic Journal of CardiologyEndovascular treatment for thromboangiitis obliterans (Buerger’s disease)
2023, Cochrane Database of Systematic ReviewsPOST EVENT-CARDIOVASCULAR RISK PERCEPTION SURVEY: VALIDITY AND RELIABILITY IN CARDIAC PATIENTS POST HEART EVENT
2023, Journal of Nursing Measurement
- ☆
Framingham Study research is supported by the National Institutes of Health/National Heart, Lung, and Blood Institute (NIH/NHLBI Contract N01-HC-38038) and the Visiting Scientist Program, which is supported by Servier Amérique.