Confidentiality: more than a linkage file and a locked drawer.

S omewhere near the end of a consent form, script, recruitment letter, or brochure, researchers usually state that they will keep information “confidential.” But how is this promise evaluated by the Institutional Review Board (IRB) and kept by the researchers? Given the researchers’ confidentiality plan, the IRB must assess the risks related to participation, including risks created by disclosure of any information shared by subjects, and determine whether the study will be able to provide the level of confidentiality promised to subjects. In clinical research, confidentiality often concerns access to medical records or inadvertent revelation of patient status or of diagnosis. In comparison, survey research may raise fewer flags. When investigators offer confidentiality, it has frequently seemed sufficient to note that the study team will provide a locked cabinet, coded files, or confidentiality statements. In reality, these safeguards may be inadequate or difficult to implement. Indeed, unique and important confidentiality concerns arise during the different stages of a survey research project, but are often under-appreciated in comparison to those of clinical research. According to a recent National Research Council Panel on Institutional Review Boards, Surveys, and Social Science Research, “It is likely that those involved in the human participant protection system... are paying too little attention to the ways in which technological and other changes in the research environment are increasing the risk of disclosure of the identity of participants in research.”1 What attention there is, however, has focused on statistical and administrative techniques to protect confidentiality of largescale public-use micro data in secondary data analyses.2 This paper examines how researchers keep their promise of confidentiality and how that promise is sometimes challenged during primary data collection and analysis for an interview study.