(365) Tokenization and linkage of state vaccine registry data to improve determination of COVID-19 vaccination status in a large administrative claims database

Background: Administrative claims data have historically under-captured COVID-19 vaccinations. Using tokenization to link claims data to state vaccination registries may provide more complete capture of vaccine status.

Objectives: We compared COVID-19 vaccine uptake in California using a large administrative claims database, with and without linkage to the state vaccine registry.

Methods: We identified all California residents who were enrolled in an insurance plan that contributed closed administrative claims to HealthVerity (e.g. commercial, Medicaid, and Medicare Advantage plans) between December 10, 2020 and September 30, 2023. To be included, participants had to be continuously enrolled for 6 months prior to study start. Using tokenization, we linked HealthVerity claims data to the state vaccine registry. We compared documented COVID-19 vaccine uptake using NDC and CPT codes in the claims alone versus uptake observed after augmenting vaccination status data using CVX codes in the linked vaccine registry.

Results: After claims-based eligibility criteria were applied, claims were available for 7.9 million people (~20% of all Californians). As of September 30, 2023, when using claims data alone, only 25% of the cohort had documented receipt ≥1 COVID-19 vaccine. After adding vaccination history data from the state registry, documented receipt increased to 63%. For the bivalent BA.4/5 vaccine, we found documented uptake of 7% using claims only and 16% when augmented with the registry. Both estimates were lower than what was reported in the full state vaccine registry (79% for any vaccine and 21% for bivalent), likely due to exclusion of a large proportion of adults ≥65 years of age who participate in Medicare Fee-For-Service programs. For example, only 11% of individuals in the claims cohort were ≥65 years versus 16% in all of California. Because COVID-19 vaccine uptake is highest in the elderly, this likely explains the lower uptake seen in our claims-based cohort.

Conclusions: Augmenting claims with state COVID-19 vaccine registry data through tokenization more than doubled vaccine coverage documentation. Although claims-based databases likely under-represent older adults because they do not capture Medicare Fee-For-Service recipients, tokenization of state vaccine registries offers a novel approach to accurately document vaccine receipt in these widely-used claims databases.