Data linkage
Linkage process
Cohort data from SDB was first linked to the MCD deterministically. Where cohort records agreed with MCD records exactly on first and last names as well as sex and birthdate, and where each set of values for these fields yielded a single one-to-one match, then these matches were accepted.
Probabilistic linkage was then undertaken on the remaining cohort records to identify probability matches based on similarities in characteristics such as surname, given name(s) and day, month, and year of birth. Sample-based clerical review and post-linkage refinement were also undertaken to further ensure linkage quality. Of the accepted links, 95.5% were a result of unique deterministic links and the remaining resulting from probabilistic linkage.
Linkage results
The final number of accepted links between the cohort and MCD was 3,591,246, an overall linkage rate of 86.6%. Further investigation of unlinked records showed that most unlinked records were records from people on temporary visas on which the visa holder is not immediately eligible for Medicare (see visa information below). These individuals are not in scope of these analyses until they become Medicare eligible. Excluding these records, the linkage rate of in scope records was 97.8%.
There was no notable variation in linkage rates by age or sex.