Short-Term Effectiveness of Bivalent COVID-19 mRNA Booster Vaccinations in the Veterans Health Administration Using Self-Controlled Risk Interval and Test-Negative Case-Control Designs

Background: On 31 August 2022, FDA authorized COVID-19 bivalent vaccines. At the time, data on the benefits of bivalent vaccines were limited.

Objectives: To evaluate the short-term relative vaccine effectiveness (rVE) of bivalent COVID-19 mRNA vaccine boosters compared to prior vaccination using two methods: a self-controlled risk interval (SCRI) design, a method in development, as primary analysis; and a test-negative design (TND) case-control study, as secondary analysis.

Methods: Using Veterans Health Administration data we identified individuals who received a bivalent booster on or after 01 September 2022, had no vaccination 60 days prior, and no history of tixagevimab/cilgavimab treatment. For the SCRI analysis, we identified SARS-CoV-2 infections (positive SARS-CoV-2 PCR, antigen test, or COVID-19 treatment) days 14-17 post-booster (booster period), when bivalent vaccination is presumed to have induced immunity, and on days 4-7 post-booster (control period), at a time before immunity presumably has been induced. For TND, we identified patients with a SARS-CoV-2 PCR or antigen test. For both analyses, we included infections identified through the end of Omicron BA.4/BA.5 predominance, excluding infections detected during hospitalization or with a COVID-19 history during the prior 90 days. For SCRI analysis, we used logistic regression to calculate relative risk (RR) (95% confidence interval [CI]) of infection in the booster vs control period. For TND, after matching, we used conditional logistic regression to calculate the odds ratio (OR) (95%CI) to estimate the association between a positive test and receipt of a bivalent booster.

Results: For the SCRI analysis, we identified 76 and 95 SARS-CoV-2 infections in the booster and control intervals, respectively, yielding a rVE of 20% (95%CI [-8, 41]). For TND, we identified 4,467 cases and 17,868 controls, finding a rVE from the adjusted matched analysis of 56% (95%CI [46, 64]); rVE was significantly higher with >8 months vs < 3 months since prior vaccination.

Conclusions: SCRI analysis showed a low, non-significant rVE with wide CIs compared to TND. SCRI analysis had limited power. The potential bias due to assumptions regarding relationship between positivity date and infection date can be explored further in medical record review. Methodological refinement of the SCRI study design is warranted to further develop this as a method for evaluation of VE.