Solving Problems of Disclosure Risk in an Academic Setting: Using a Combination of Restricted Data and Restricted Access Methods

The health and retirement study collects a vast amount of information about a sample of the U.S. population over age 50 from biennial interviews, supplemental questionnaires, and through linkages with administrative data including Social Security earnings and benefits records and Medicare claims records. To honor its pledge to the respondents that their data will be kept confidential, but at the same time meet its objective of providing useful data to researchers, it has developed procedures for stripping sensitive information (i.e., information that could facilitate re-identification of sample members) from data sets that are publicly released, and also for providing mechanisms for qualified researchers to gain access to a variety of restricted-access data files. These mechanisms include a procedure whereby highly qualified researchers—in particular, only those who have a current grant from a federal agency—can apply to obtain restricted-access data sets for a limited amount of time, with the understanding that they will make no attempt to re-identify sample members and that they will be audited to ensure that they have adhered to the agreed-upon safeguards. For those who meet some but not all of the requirements for receiving these data, the files can be analyzed in a data enclave (a controlled, secure environment in which eligible researchers can perform analyses). This paper focuses on approaches to restricting data access that may need to be considered by investigators who plan to share their data, and by their institutional officials who will need to support that effort with appropriate infrastructure and policies. It also provides guidance to investigators and institutional review boards (IRBs) who seek access to restricted data generated and archived elsewhere.