Risk Calculation and Clustering Within Racial/Ethnic Groups

Abstract

Several methods have been suggested to estimate the risk for initial coronary heart disease (CHD) events. Prior to the availability of modern computer methods, there was no easy way to adapt risk estimates to clinical practice. In 1998, the Framingham CHD risk approach was simplified, leading to greater interest in using risk prediction algorithms for clinical care. This formulation used categories of blood pressure, total cholesterol, HDL cholesterol, and simple groupings for smokers and diabetes mellitus to assess CHD risk. In 1999, the National Heart, Lung, and Blood Institute convened a CHD Prediction Workshop to address the use of Framingham CHD risk functions to varied populations. Validation of the Framingham CHD prediction scores was achieved with data from a large number of observational data sources, representing multiple ethnic groups. As part of the proceedings of the CHD workshop, D’Agostino demonstrated the relative effects of most CHD risk factors. The equations derived in Framingham were tested in each site and showed good predictive capabilities in outcomes for the various ethnic groups and ages represented, with some exceptions. In regions where CHD risk was low, such as in the Honolulu Heart Program, the Framingham risk algorithms overestimated the CHD experience of the participants.

By the late 1990s, concern had risen among European scientists as to how well Framingham algorithms predicted CHD risk in their region. A large-scale, multinational European study was undertaken to address this issue – the Systematic Coronary Risk Estimation (SCORE) project – assembling data from 12 primarily Caucasian, European countries. A variety of issues relating to CHD risk estimation are of particular importance to high-risk population groups such as African-Americans. Blood pressure elevation has been consistently shown to be an important CHD risk predictor and levels are typically higher in black American populations. A variety of cardiovascular techniques are now available to assess subclinical arteriosclerosis. Estimation of risk for cardiovascular events is a dynamic field and it is expected that the approaches will undergo modification, as better information is obtained and more contemporary data become available.