(C-161) Self-controlled Methodology and Test Negative Design to Evaluate Waning Effectiveness of COVID-19 and Influenza Vaccination in the Veterans Health Administration

Background: Self-controlled risk interval (SCRI) analyses implicitly control time invariant confounders; time-varying factors may introduce bias.

Objectives: To evaluate vaccine effectiveness (VE) and VE waning of COVID-19 and influenza vaccines using SCRI, accounting for time-varying viral circulation, and test-negative design (TND) case-control study (comparator analysis).

Methods: We used Veterans Health Administration (VHA) data where event was SARS-CoV-2 (or influenza) medically attended infection. For the COVID-19 VE SCRI analysis, we included individuals who received third dose monovalent vaccination Dec 2021-Aug 2022, and had an event up to Day 238 postvaccination (within the Omicron dominance period). To analyze short-term VE, we compared event rates in the reference period (days 15-56 post-vaccination, when immunity presumably peaks) to days 3-7 (when presumably no immunity was conferred yet by vaccination). We analyzed waning comparing event rates in the reference period with each subsequent week. We used data from individuals without third dose to estimate daily infection rates for each of 80 distinct geographical areas in the country to account for viral circulation changes. We used logistic regression to calculate short-term VE (1-incidence rate ratio [IRR] for reference period vs days 3-7) and estimated the event IRR (95% confidence interval [CI]) for each week versus reference period to estimate waning (defined as IRR(95%CI)>1). We included an offset term to account for difference in length of the periods and for differences in viral circulation. For the TND analysis, we identified individuals tested for SARS-CoV-2 up to 238 days post-vaccination, matching up to 4 controls per case by geographic area within 7 days of test date, conducting conditional logistic regression with confounder adjustment. For waning of influenza VE, we followed similar SCRI methods, using only periods of H3 dominance in the 2022/2023 influenza season, also conducting a TND.

Results: For COVID-19 VE, SCRI and TND analysis yielded, respectively, short-term VE of 47% [95%CI 41%, 50%] and waning by Week 12, and 53% [95%CI 51%, 55%] with waning by Week 13 post-vaccination. For influenza VE, the brief period of H3 dominance limited our ability to assess waning to only 98 days; the SCRI yielded IRRs (95%CI) above 1 for each of Weeks 9-11 post-vaccination vs reference period; for the TND, ORs (95%CI) for Weeks 9-11 post-vaccination vs Weeks 3-8 included the null.

Conclusions: For COVID-19, the SCRI and TND designs yielded similar findings for short-term effectiveness and waning. Analyses for influenza were inconclusive, reflecting the difficulty in estimating waning influenza VE, likely due to the short H3 subtype dominance period.