PRENTICE ET AL. RESPOND

Abstract

We thank Carter-Pokras and Zambrana for their comments and the opportunity to further discuss how the Los Angeles Family and Neighborhood Survey (LAFANS) protects sensitive information such as respondents’ legal status. Researchers who collect sensitive information have an obligation to protect respondents’ confidentiality, privacy, and safety. Yet there is another significant obligation—to science, the public interest, and the respondents themselves—to have the data used as broadly as possible for research and policymaking. Balancing these 2 obligations is a challenge. To protect respondents, we obtained a confidentiality certificate, which, in principle, protects LAFANS against being forced to divulge data in response to a subpoena. Projects collecting especially sensitive information have often gone further by turning over all private information to an “honest broker,” who may be outside the United States and hence cannot be subpoenaed. Releasing detailed information about the geography of respondents’ lives, such as where they reside, substantially increases the risk of indirect or deductive disclosure. However, this information is essential for developing a better understanding of contextual effects on health and for estimating multilevel models. The LAFANS solution was to provide only the household’s Service Planning Area in the public-use data files that are available through the project’s Web site. (There are 8 Service Planning Areas in Los Angeles County, each with approximately 1.2 million people.) Detailed geographic information is made available in the restricted-use data (see Peterson et al.1), of which there are 3 versions that provide successively greater precision about the respondents’ locations. Researchers requesting the restricted-use data must submit a research plan and a data safeguarding plan approved by their institutional review board. An important principle guiding the design of data release for LAFANS was that all data (except private identifying information such as respondents’ names and addresses) would be available to researchers. This policy actually provides additional protection because it allows researchers to control the release of the data. Researchers are mandated to release their data in response to a demand based on the Freedom of Information Act when data collection is supported by federal funds and the data have been used to support federal actions or regulations. Without data release procedures in place, it is difficult to specify requirements for data release after a Freedom of Information Act request has been made and researchers may lose control of their data.2 Researchers need to continually monitor legislation, such as the Freedom of Information Act and the Patriot Act, that can potentially undermine their best efforts to protect respondents’ confidentiality. Finally, we agree that it is dangerous to collect legal status in a clinical context where requests for such information may deter patients from seeking needed care.