Biochemical epidemiology in community-based studies: practical lessons from a study of T-cell subsets.

Elaborate laboratory tests are increasingly being incorporated into traditional epidemiologic research designs, a concept commonly termed biochemical epidemiology. Some of the issues encountered are illustrated by a recent population-based survey of healthy individuals in the Washington, D.C. area designed to examine the effects of demographic characteristics, lifestyle, and medical conditions on peripheral blood T-cell subsets. The study was conducted in three phases: selection of households by random digit dialing (Phase I); telephone interviews (Phase II), and self-administered questionnaires and phlebotomy (Phase III). Although this design facilitated the selection of the final study population, it influenced the participation rates by offering opportunities for nonresponse at each phase. Race was the strongest determinant of response rate despite the use of highly-trained, racially-matched telephone interviewers and repeated attempts at refusal conversion. Also discussed are issues of confidentiality, and logistics of biologic specimen collection and handling. The difficulties encountered in this survey are examined, with suggestions for future population-based investigations involving biochemical epidemiology.