(P-437) Old Methods and New Data: A near Real-time Plan for Vaccine Surveillance and Signal Detection in the United States

Background: Real world data (RWD) have great utility in aiding the novel coronavirus disease (COVID-19) pandemic response. As COVID-19 vaccines receive approval, there is an imminent need for real world monitoring adverse events (AE) following vaccine exposure in the general population.

Objectives: To lay out a thoughtful assessment of practical options for vaccine surveillance in the US and suggest an optimal route for efficient and effective surveillance using RWD.

Methods: Regulatory practices require individual companies to conduct formal post-marketing product safety studies as well as safety surveillance for new products, including vaccines. There are a variety of approaches to safety surveillance, all with their benefits and limitations. All factors need to be considered, including: Breadth of coverage, generalizability, Depth/specificity of information being collected/assessed (both for exposure and outcomes), completeness of individual patient data for outcomes and covariates of interest, timeliness of data and analytics, and cost. We assess a number of approaches for real-time signal detection and lay out an innovative approach that applies an old methodology to new data.

Results: While Open Claims databases are not typically the first line resource for epidemiological studies, they are a strong and viable option for signal detection. They are near real-time, include all payers (and hence all age groups), and include prescription and medical claims. While they are not 100% complete longitudinal data for individual patients, the prevalence of hospitalized AEs are comparable to closed claims and hospital chargemaster data. There is a need to shift our thinking from looking at the risk of specific AEs and focus on pharmacovigilance to identify signals for AEs. These data can be analyzed via traditional pharmacovigilance disproportionality analytics which is the standard approach for signal detection of traditional spontaneous report data. Signals can then be evaluated, refined, and determined if/when any specific AEs need a more rigorous epidemiologic study conducted, in which case, the research question can be developed and paired with a supplemental fit-for-purpose dataset.

Conclusions: We need immediate and on-going vaccine surveillance that leverages a large and diverse patient pool. This approach of using existing reliable methods (disproportionality analytics) on new data via real world open claims data, allows for near real-time surveillance with regular updated data.