loading  Checking for direct PDF access through Ovid


Disease risk prediction models have been developed to assess the impact of multiple risk factors and to estimate an individual's absolute disease risk. Accurate disease prediction is essential for personalized prevention, because the benefits, risks, and costs of alternative strategies must be weighed to choose the best preventive strategy for individual patients. Cardiovascular disease (CVD) prediction is the earliest example of individual risk predictions. Since the Framingham study reported a CVD risk prediction method in 1976, an increasing number of risk assessment tools have been developed to CVD risk in various settings. The Framingham study results are fundamental evidence for the prediction of CVD risk. However, the clinical utility of a disease prediction model can be population-specific because the baseline disease risk, subtype distribution of the disease, and level of exposure to risk factors differ by region and ethnicity.

It has been proved that CVD prediction models which were developed in high-risk populations, such as the Framingham Risk Score, overestimate an individual's disease risk when applied to a low-risk population without re-calibration. Thus countries of relatively low CVD risk are trying to re-calibrate the existing CVD prediction models or to develop a new prediction model analyzing their own population data. However, even the re-calibrated or newly-developed CVD prediction models are often of little clinical value in a low-risk population. A good example is the CVD prediction in the Korean population. Compared to Western populations, the Korean population has much lower incidence of coronary heart disease. Therefore, the vast majority of individuals fall into the low-risk group when their disease risk is assessed with a prediction model. Even a well-validated prediction model may not identify high-risk individuals who merit aggressive preventive treatment.

A few alternative approaches have been suggested for CVD risk prediction in a low-risk population. First, use of longer-term (or lifetime) CVD risk score can improve risk stratification in a low-risk population. Because the absolute CVD risk largely depends on age, younger people are hardly classified into high-risk group even if they have multiple major risk factors. If we estimate an individual's risk for a longer-term, we can identify more young people who merit earlier preventive intervention. Second, risk prediction for a wide range of CVD may be useful in some low-risk population. Asian populations have relative lower risk for coronary heart disease risk but higher risk for cerebrovascular disease. Risk prediction model for general cardiovascular disease, which includes both coronary heart disease and stroke, can improve risk stratification in Asian population. Third, we may consider lowering cutoff levels for risk stratification in low-risk population. But it requires much more evidence to find the best cutoff for risk stratification.

Related Topics

    loading  Loading Related Articles