A randomized study of the effect of anonymity, quasi-anonymity, and Certificates of Confidentiality on postpartum women's disclosure of sensitive information.

Background: Under-reporting of substance use and other sensitive information is a substantial threat to internal study validity, particularly during the perinatal period. Anonymous approaches are associated with greater disclosure but are incompatible with longitudinal follow-up. Alternative approaches include use of a U.S. Federal Certificate of Confidentiality (CoC) and quasi-anonymous methods, in which there is no link between name and data. However, the relative effect of these procedures on disclosure is unknown. This randomized study was designed to evaluate the effects of consent condition (anonymous, quasi-anonymous, CoC, and traditional confidentiality) on disclosure of sensitive information among postpartum women.

Methods: Participants were 200 postpartum, primarily African-American women who were randomly assigned to one of the four consent conditions and completed a brief computer-delivered assessment of alcohol and drug use, sexual risk, intimate partner violence, and emotional distress.

Results: Participants in the anonymous and quasi-anonymous conditions disclosed significantly more sensitive information than those in the traditional consent condition. In contrast, no advantage in overall disclosure was observed for the CoC condition. This result was largely consistent across specific content areas with the exception of emotional distress, disclosure of which was unrelated to consent condition.

Conclusions: Although use of a CoC has limited impact on disclosure, the quasi-anonymous method may increase disclosure to a similar extent as full anonymity. Quasi-anonymous approaches should be considered when under-reporting is likely, a context in which the disadvantages of this approach must be balanced against its advantages.