Modeling heart disease mortality with census tract rates and social class mixtures.

The relationship between social class and 1980 heart disease (HD) mortality in eight urban U.S. counties was examined by regressing age and sex adjusted census tract specific HD rates (N = 1211) on tract social class characteristics. The regression model indicated that lower middle class residents experienced a HD mortality rate 1.9 (95% CI = 1.3, 2.8) times the rate in the upper middle/middle class, while the working poor experienced a HD rate 4.4 (95% CI = 3.5, 5.7) times the rate in the referent class. Similar class effects were seen for both black and nonblack residents. The crude race effect (1.3 with 95% CI = 1.2, 1.4) was explainable by the concentration of blacks in the lower classes. The methods illustrate the ecologic regression of mixtures of mortality rates on mixtures of exposure in the presence of random tract effects which eliminates some of the problems associated with small denominators or zero rates in some tracts.