ABSTRACT
BackgroundDifferential SARS-CoV-2 exposure between vaccinated and unvaccinated individuals may confound vaccine effectiveness (VE) estimates.AimWe conducted a test-negative case-control study to determine VE against SARS-CoV-2 infection and the presence of confounding by SARS-CoV-2 exposure.MethodsWe included adults tested for SARS-CoV-2 at community facilities between 4 July and 8 December 2021 (circulation period of the Delta variant). The VE against SARS-CoV-2 infection after primary vaccination with an mRNA (Comirnaty or Spikevax) or vector-based vaccine (Vaxzevria or Janssen) was calculated using logistic regression adjusting for age, sex and calendar week (Model 1). We additionally adjusted for comorbidity and education level (Model 2) and SARS-CoV-2 exposure (number of close contacts, visiting busy locations, household size, face mask wearing, contact with SARS-CoV-2 case; Model 3). We stratified by age, vaccine type and time since vaccination.ResultsVE against infection (Model 3) was 64% (95% CI: 50-73), only slightly lower than in Models 1 (68%; 95% CI: 58-76) and 2 (67%; 95% CI: 56-75). Estimates stratified by age group, vaccine and time since vaccination remained similar: mRNA VE (Model 3) among people ≥â¯50 years decreased significantly (p = 0.01) from 81% (95% CI: 66-91) at < 120 days to 61% (95% CI: 22-80) at ≥â¯120 days after vaccination. It decreased from 83% to 59% in Model 1 and from 81% to 56% in Model 2.ConclusionSARS-CoV-2 exposure did not majorly confound the estimated COVID-19 VE against infection, suggesting that VE can be estimated accurately using routinely collected data without exposure information.