Should we consider a subject's knowledge of the etiologic hypothesis in the analysis of case-control studies?

In the context of a case-control study of vitamin supplements and neural tube defects, Werler et al. (JAMA 1993;269:1257-61) have argued that respondents who are aware of the etiologic hypothesis might particularly be prone to misreport their exposure status. They suggest that excluding such participants could offer the best chance for obtaining a valid risk estimate. However, knowledge of an etiologic hypothesis can be the reason for a person's being exposed or nonexposed. In addition, exposed cases selectively may become aware of the etiologic hypothesis after being diagnosed. If one of these, and not recall bias, is the basis for an association between knowledge of the hypothesis and exposure status, then "knowledgeable" subjects should be retained in the analysis so as not to compromise the precision or validity of the results.