for Linking Data Sets Without Unique Identifiers Leveraging Administrative Data for Program Evaluations: A Method

AbstractIn community-based wellness programs, Social Security Numbers (SSNs)are rarely collected to encourage participation and protect participant pri-vacy.Onemeasure ofprogram effectiveness includes changesinhealth careutilization. For the 65 and over population, health care utilization is cap-tured in Medicare administrative claims data. Therefore, methods asdescribed in this article for linking participant information to administrativedata are useful for program evaluations where unique identifiers such asSSN are not available. Following fuzzy matching methodologies, participantinformation from the National Study of the Chronic Disease Self-Management Program was linked to Medicare administrative data. Linkingvariables included participant name, date of birth, gender, address, and ZIPcode. Seventy-eight percent of participants were linked to their Medicareclaims data. Linking program participant information to Medicare adminis-trative data where unique identifiers are not available provides researcherswith the ability to leverage claims data to better understand programeffects.Keywordsadministrative data, fuzzy matching, linkage, Medicare, program evaluation