Abstract
The association between vaccine efficacy (VE) and force of infection (FoI) remains incompletely understood. Previous analyses have been primarily based on trial-level summary data-not accounting for the effect of time and constrained by the number of trials. Here, we leverage individual-level data from three phase 3 randomized, placebo-controlled COVID-19 vaccine trials-the COVE trial (Moderna, CoVPN3001), the AZD1222 trial (AstraZeneca, CoVPN3002), and the ENSEMBLE trial (Janssen/Johnson & Johnson, CoVPN3003)-and contemporaneous geographic-location-specific SARS-CoV-2 surveillance data from the start of the pandemic through November 14, 2021 (including the blinded follow-up periods of the trials) to conduct five cohort- and vaccine-specific analyses: COVE (U.S.), AZD1222 overall (U.S. + non-U.S.), AZD1222 U.S., ENSEMBLE overall (U.S. + non-U.S.), and ENSEMBLE U.S. In AZD1222 U.S., higher VE was associated with higher FoI (p = 0.01). In ENSEMBLE overall, lower VE was marginally associated with higher FoI (p = 0.21), further supported by a region-specific analysis. In COVE, AZD1222 overall, and ENSEMBLE U.S., no VE-FoI association was found. These findings highlighted a new perspective: the VE-FoI association appears complex, potentially influenced by FoI levels, with patterns suggesting an inverted U-shaped relationship, showing a positive association at low FoI levels and a negative association at high levels.