Overview
Several studies measured antibodies to the SARS-CoV-2 virus to estimate the proportion of the people in Canada who were infected at different points during the COVID-19 pandemic. Each study had different characteristics. In this project, we will re-analyze data from multiple studies to understand how study characteristics impacted their findings, and whether we can adjust for these differences to make study results more comparable. We will focus on two elements of study design: which sociodemographic groups were over- or under-represented in a study and which antibody assays were used. Our first analysis will compare population antibody estimates across studies that used different recruitment methods, leading to different sociodemographic characteristics. We will apply different statistical methods to remove “representation bias” from estimates of population immunity levels. We will analyze whether adjusting for representation bias makes estimates between studies more comparable and compare different adjustment methods to one another. Our second analysis will compare population antibody estimates from studies in the same province that used different assays. Within Alberta, Saskatchewan, and Manitoba, we will compare a study in blood donors that used a Roche assay to studies that tested leftover blood sent to provincial laboratories for routine testing that used an Abbott assay. Both assays are imperfect, with different probabilities of falsely returning a negative result in someone who was infected or falsely returning a positive result in someone who was not infected. We will assess whether findings are more comparable after applying an adjustment for each assay’s performance, which will be derived using Bayesian latent class analysis. This is a statistical model that can estimate the true prevalence in situations where multiple imperfect assays are used. Together, these analyses will help inform best practice methods for analyzing data on population immunity levels for SARS-CoV-2 and other pathogens.
Calendar
- Start Date
- 2023-05-16
Contact Details
- Name
- Dr. Alton Russel
- Institution
-
McGill University