The CMS Public Use File Reidentification Experience

• Activity Number: 638
• Type: Topic Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Survey Research Methods Section
• Author(s): Susan Hinkins*+ and Patrick Baier and Fritz J. Scheuren
• Companies: NORC and NORC and NORC at the University of Chicago
• Keywords: Statistical Disclosure Limitation ; Public Use File

Abstract

As part of a US Federal Government-wide effort to increase data transparency and access, an effort was undertaken to see if a "disclosure safe" data file could be produced containing beneficiary summary and claims data from the Current Medicare Files. To be "disclosure safe" virtually all the health records had to be certified as not traceable back to an individual. Under the Health Insurance Portability and Accountability Act (HIPAA), such certified data could be made public. To accomplish this goal, separate De-identification and Re-identification teams were assembled. The work of the De-identification team is described in a companion paper in this session. Here we focus on methods employed by the Re-identification team to test whether the data could be traced back to a single entity. The De-identification team had no restrictions on its work, but there was a clear goal to produce something as useful as possible given disclosure constraints. Due to the nature of the data, synthetic methods were of limited use. Other methods, such as data swapping, data coarsening, and suppression were used in combination to produce a file that could be certified as "disclosure safe."