Linking a Total Ankle Arthroplasty Registry to Medicare Inpatient Claims without Unique Identifiers.

BACKGROUND: Linking clinical registries to administrative claims data enables researchers to capitalize on the specific strengths of each data source with respect to the depth, breadth, and completeness of information. The objectives of this study were to link a health-system-based orthopaedic surgery registry to U.S. Medicare claims data without the use of unique identifiers and to assess the representativeness of the linked records. METHODS: The registry included clinical data for patients ≥65 years of age who underwent elective, inpatient total ankle arthroplasty (TAA) in a single health system during the period of 2007 through 2012. Registry participants were identified within the Medicare data by linking registry procedures to TAA procedures within the claims data using a combination of procedure date, patient date of birth, and patient sex. We assessed the representativeness of the linked records by comparing them to both unlinked registry records and unlinked Medicare records for TAA procedures. Additionally, we described the availability of postsurgical data for linked records. RESULTS: Of 360 TAA registry participants ≥65 years of age, 280 (77.8%) were matched to a Medicare record; 250 (89.3%) of those 280 participants were matched on the basis of a linking rule that required an exact match for procedure date, date of birth, and sex. The 280 linked records comprised 5.5% of all Medicare TAA procedures among beneficiaries ≥65 years of age enrolled in the fee-for-service Medicare program (n = 5,070). Compared with linked records, unlinked records were more likely to be for patients 65 to 69 years old, but the 2 groups were similar in terms of sex, body mass index, and availability of clinical measurements. Of the linked records, 214 (76.4%) had ≥3 years of postoperative follow-up claims data. CONCLUSIONS: Linkage without unique patient identifiers between an orthopaedic registry and Medicare claims data is feasible, allows for assessment of representativeness, and creates a unique resource of longitudinal data for research.

Duke Scholars

Author Sudha Rani Raman Population Health Sciences

Author Bradley Gordon Hammill Population Health Sciences

Author Samuel Bruce Adams Jr. Orthopaedic Surgery

Author Lesley H. Curtis Population Health Sciences