2C-01 - Is the Climb Worth the View? Estimating COVID-19 Vaccine Effectiveness (VE) using Nested Trials vs. Traditional Matched Cohort Design

Background: Effectiveness studies of COVID-19 vaccination often rely on traditional matched cohort studies. However, target trial emulation with a nested trials (NT) approach allows for evaluation of patients on each day of the intake period, crossover of unexposed patients to exposed, and enhanced statistical efficiency.

Objectives: We performed the same analyses using both designs to assess the VE of the 2023-2024 monovalent mRNA-1273.815 COVID-19 vaccine against COVID-19 hospitalization.

Methods: Using HealthVerity administrative claims, US adults ≥18 years, vaccinated with mRNA-1273.815 (12 Sep 2023-31 Dec 2023) were matched 1:1, to individuals who did not receive mRNA-1273.815, on calendar date, demographics, 2022-23 bivalent vaccine receipt, and frequency of prior medical claims. For NT, risk-sets for each day a vaccination occurred were created and the same 1:1 matching process was repeated each day to generate a sequence of nested ‘trials’. Data were pooled across NT. Patients were followed through 26 Jan 2024 for the outcome of COVID-19 hospitalization. If a patient was sampled into the unexposed arm, they could later be included as exposed, if eligibility criteria were met. Adjustment with inverse probability of treatment weighting (IPTW) was used in both designs, and inverse probability censoring weighting (IPCW) was also used in NT, estimated by multivariable logistic regression. Weighted Cox models were used to estimate hazard ratios (HR) and VE (1-HR*100).

Results: We matched 1,272,161 exposed individuals to unexposed individuals in the cohort study, and 1,186,748 in NT. Demographics were similar, and after weighting, adjusted standardized differences showed no meaningful differences between exposed and unexposed individuals using either approach. In the cohort study, 9.4% of the unexposed were censored due to receipt of a mRNA-1273.815 vaccine; similarly in NT, 9.4% of the unexposed received mRNA-1273.815; of these, 64% were eligible for entry into the exposed cohort. A higher proportion of patients were vaccinated later in the NT. In the cohort study, VE (95% CI) was 51% (48%, 54%), compared with 44% (40%, 48%) in NT.

Conclusions: The VE from the nested trials approach was lower than the cohort study. Timing of vaccination and evolving variant predominancy may have driven the difference in VE estimates. Both methodologies confirmed the effectiveness of vaccination with mRNA-1273.815 against COVID-19 hospitalization. Implementing NT is computationally intensive but may be “worth the view” in certain situations, such as when crossover is expected to significantly impact estimates. In this instance, the results of both methods resulted in a similar interpretation of results.