Prediction of coronary heart disease mortality in blacks and whites: pooled data from two national cohorts

Statistical models used to predict personal risk of death from coronary heart disease (CHD) have been based on studies among white populations. We compared the predictive functions derived from black and white men and women, using the pooled data of 2 national cohorts: the First National Health and Nutrition Examination Survey (NHANES I) Epidemiologic Follow-up Study and the Second National Health and Nutrition Examination Survey (NHANES II) Mortality Study. The participants included 6,937 white men, 940 black men, 9,202 white women, and 1,463 black women aged 30 to 74 years who were free of CHD at baseline. The 2 cohorts were followed for 20 and 15 years, respectively. There were no significant differences between blacks and whites in the magnitude of the Cox coefficients for most of the personal risk factors (i.e., age, systolic blood pressure, serum total cholesterol, smoking, and diabetes mellitus status) for men and women. The receiver operating characteristic (ROC) analyses, with all risk factors considered collectively, suggest that the models have similar ability to rank personal relative risk among blacks and whites. The areas under the ROC curve were 0.77 and 0.76 for white and black men, respectively, and 0.84 and 0.82 for white and black women, respectively. However, the equation derived from white men overestimated the 15-year cumulative CHD mortality in black men by about 60%. Thus, predictive functions derived from 1 demographic group (e.g., whites) can be applied to another subgroup (e.g., blacks) to rank personal risk. However, prediction of absolute risk is less accurate.