Race/ethnicity misclassification of persons reported with AIDS

Abstract

Objectives. To examine differences in race/ethnicity classifications of persons with AIDS among three reporting sources and to estimate the effect of these differences on calculated AIDS rates. Methods. We reviewed case reports from the national AIDS surveillance database, interview (self‐reported) data from 11 state/local health departments, and death certificate information from 16 state/local health departments for agreement in race/ethnicity coding among persons reported with AIDS. Results. Race/ethnicity coding inconsistencies with AIDS case reports were greatest for persons identified as American Indians/Alaska natives on death certificates (46% [47/102] disagreement) and by self‐report (57% 8/14 disagreement). Agreement with AIDS case reports was highest either for persons identified as white from death certificates (4% [1314/31 070] disagreement) and white from self‐reports (2% [26/1068] disagreement) or black from death certificates (3% [440/13 592] disagreement) and black from self‐reports (3% [21/736] disagreement). For other racial/ethnic groups, disagreement with AIDS case reports was intermediate; for Asians/Pacific Islanders, 12% [45/377] disagreement with death certificates and 33% 4/12 disagreement with self‐reports; and for Hispanics, 14% [1151/8527] disagreement with death certificates and 24% [59/249] disagreement with self‐reports: Conclusion. For certain racial/ethnic groups, classification by race/ethnicity can differ substantially by surveillance data source. Because allocation of public health resources may be determined by estimates of disease impact on different population groups, periodic evaluations of the accuracy of race and ethnicity reporting are needed to assure appropriate distribution of these resources.