Background: Longitudinal studies that collect sensitive data, such as test results for HIV antibodies, present difficult ethical problems for investigators. Personal identifiers are needed for longitudinal follow-up, but current regulations in the United States require that all subjects be informed of their HIV test results when identifiers are retained. Therefore, subjects who do not wish to learn their HIV status must be excluded from prospective research. Because these subjects may be at particularly high risk for HIV infection, this situation may lead to substantial bias.
Methods: We describe a new application of a methodology for blinding such studies, based on the link-file system used for protecting subjects' sensitive research data. This design enables investigators to hold each subject's HIV test result and survey data for multiple ascertainments over time while effectively severing any link between this information and the subject's identity. This objective is achieved by a process of sequential coding, using at least two coders who are not associated with the study.
Conclusions: The method described in this paper permits investigators in the United States to perform important prospective epidemiologic studies while adhering to current regulations for the conduct of such investigations.