Back

A novel approach for estimating vaccine efficacy for infections with multiple disease outcomes: application to a COVID-19 vaccine trial

Williams, L. R.; Voysey, M.; Pollard, A. J.; Grassly, N. C.

2023-03-02 epidemiology
10.1101/2023.03.02.23286698 medRxiv
Show abstract

Vaccines can provide protection against infection or limit disease progression and severity. Vaccine efficacy (VE) is typically evaluated independently for different outcomes, but this can cause biased estimates of VE. We propose a new analytical framework based on a model of disease progression for VE estimation for infections with multiple possible outcomes of infection: Joint analysis of multiple outcomes in vaccine efficacy trials (JAMOVET). JAMOVET is a Bayesian hierarchical regression model that controls for biases and can evaluate covariates for VE, the risk of infection, and the probability of progression. We applied JAMOVET to simulated data, and data from COV002 (NCT04400838), a phase 2/3 trial of ChAdOx1 nCoV-19 (AZD1222) vaccine. Simulations showed that biases are corrected by explicitly modelling disease progression and imperfect test characteristics. JAMOVET estimated ChAdOx1 nCoV-19 VE against infection (VEin) at 49% (95% CI 37-59) and progression to symptoms (VEpr) at 44% (95% CI 27-58). This implies a VE against symptomatic infection of 72% (95% CI 63-80), consistent with published trial estimates. VEin decreased with age while VEpr increased with age. JAMOVET is a powerful tool for evaluating diseases with multiple dependent outcomes and can be used to adjust for biases and identify predictors of key outcomes.

Published in AJE Advances: Research in Epidemiology · not in our set (fewer than 10 published preprints to learn from) · training set

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.