Assigning race to occupational cohorts using census block statistics.

Race is an important determinant of disease frequency, yet the race of subjects in retrospective epidemiologic studies is frequently unknown. If addresses are available, the race of study subjects may be estimated from the racial composition of the blocks on which they have resided. Such information can be obtained from census block statistics for Standard Metropolitan Statistical Areas and, with the 1990 Census, probably for the entire United States. The authors assigned black race to persons on blocks with greater than 60% black residents and white race to those residing on blocks containing less than 40% blacks. The validity of the procedure was tested on 341 decedents of known race who resided at the time of death within the Detroit, Michigan, Standard Metropolitan Statistical Area. Thirteen individuals who lived on blocks with 40-60% blacks were excluded from analysis, as well as 18 others for whom racial composition of blocks could not be ascertained. In 293 (94%) of the remaining 310 persons, race assigned on the basis of census information agreed with race obtained from death certificates. This method should prove useful for assigning race to study subjects lacking racial identification.