Validation of Work Histories for the Purpose of Epidemiological Studies

Abstract

Work histories often serve as the basis for classifying workers by their potential for exposure to occupational hazards for the purposes of epidemiology studies. They can be obtained from a variety of sources and can be of varying degrees of completeness and accuracy. Industry-based studies typically rely on employer records which often contain fairly detailed and complete work histories on employees for the time period they were employed with that employer, but jobs prior to or subsequent to that employment may be missing. Community-based studies generally rely on interviews with subjects or their next of kin to provide a complete listing of jobs over the lifetime of the subject; however, because of the fallibility of human memory, work histories obtained from interviews may not be sufficiently detailed or accurate. These issues are discussed in the context of potential exposure misclassification and its impact on risk estimates. Misclassification that occurs independent of disease classification will usually lead to underestimates of the strength of the exposure-disease relationship, whereas misclassification that is not independent can produce over- or underestimates of risk. In the range of sample sizes that epidemiologists typically encounter, the chance probability is quite low that misclassification is independent of disease classification. Some suggestions for ways to enhance the validity of work history data for future epidemiology studies are offered. These include focusing on longer-term workers within an occupation or industry and soliciting work history data from multiple sources to offset the disadvantages of any single source. Finally, investigators are urged to validate their exposure estimates and adjust risk estimates for misclassification bias.