Abstract
Public Health, Policy and Ethics
.jpg "W. Alton Russell, PhD (he/him/his) photo")
W. Alton Russell, PhD (he/him/his)
Assistant Professor
McGill School of Population and Global Health
Montreal, Quebec, Canada
Disclosure information not submitted.
SARS-CoV-2 serological surveillance in Canada included studies in blood donors and public health surveys. Differences in the sociodemographic representativeness of study populations may bias seroprevalence estimates, necessitating statistical adjustment.
Study
Design/Methods:
We re-analyzed data from the Canadian Blood Services (CBS) blood donor study plasma samples (n=255,881) using the Roche Elecsys® Anti-SARS-CoV-2 (anti-N), meanwhile the Action to Beat Coronavirus (ABC) study that used dried blood spots from participants in a marketing panel (n=13,195), and the prospectively recruited Canadian COVID-19 Antibody and Health Survey (CCAHS; n=13,916) that used dried blood spots both used an in-house IgG‐based enzyme‐linked immunosorbent assays (ELISAs). In each study, we calculated anti-nucleocapsid antibody results for six regions or provinces during periods from 2021 to 2022 when the studies’ sample collection overlapped, calculating crude/unadjusted seropositivity and seropositivity with multilevel regression and poststratification (MRP). MRP adjusted for differences in representation by age, sex, urbanicity, and neighborhood material deprivation.
Results/Findings:
Across regions and time points, unadjusted seroprevalence differed between studies by as much as 20%, and MRP did not consistently improve comparability of seroprevalence across studies (Figure A). In 2022, seroprevalence was consistently highest among blood donors, and MRP reduced seroprevalence across all studies.
Conclusions:
While seroprevalence estimates differed between blood donors and other studies, adjusting for representativeness using common sociodemographic variables did not systematically improve concordance. This suggests that differences between studies are primarily due to other factors, such as differences in representation along other dimensions or differences in blood sample type (plasma or dried blood spots), assay performance (in particular, differences in waning sensitivity), or differences in laboratory procedures such as dilution or assay calibration.