Composite scoring--methods and predictive validity: insights from the Framingham Study

After three decades of epidemiologic research at Framingham and elsewhere, the risk factor concept is now firmly established. Atherosclerotic cardiovascular disease can now be predicted and highly vulnerable candidates identified from profiles derived from ordinary office procedures and simple laboratory tests [1]. Risk can be estimated over a 20--30-fold range, and close to half of the cardiovascular events are found to occur in a tenth of the population at highest multivariate risk. Categorical risk assessments focusing on the number of "risk factors" present also identify high-risk subjects but tend to overlook high-risk individuals with multiple marginal abnormalities. Multivariate cardiovascular risk profiles made up of the major cardiovascular risk factors can predict all of the major cardiovascular events, even in advanced age, with reasonable efficiency. Such multivariate risk assessments can be made convenient by reproduction of handbooks and use of small programmed calculators, software for personal computers, and slide rules to facilitate office and public health assessments. The sensitivity and specificity of these risk profiles can probably be improved by more detailed lipid information, including HDL-cholesterol [12], vital capacity determination, and other ECG abnormalities. General cardiovascular risk profiles can be devised to predict efficiently all of the major cardiovascular events.