METHODS AND DATA
This analysis uses the presence of antibodies to the nucleocapsid protein as an indication of past infection and the presence of anti-spike antibodies to represent the overall seroprevalence representing both vaccine-induced and infection-acquired antibodies. Prior to vaccine roll-out in Jan 2021, the presence of anti-spike antibodies also indicates past infection. The seroprevalence estimates presented in this report are from three different sources: 1) Blood donors from Canadian Blood Services and Héma-Québec; 2) anonymized discarded or residual blood samples from provincial laboratories; and 3) participants in CITF-funded research cohorts. The data are assumed to provide an assessment of seroprevalence reflecting infection approximately 14 days or more earlier than their collection date, given the time it takes on average for infected individuals to develop measurable IgG antibodies in response to infection.
SOURCES OF DATA
Data were drawn from projects funded by the Government of Canada through its COVID-19 Immunity Task Force (CITF) and projects that have made their seroprevalence estimates publicly available. The funded projects and engaged partners reflect efforts by the CITF to assess seroprevalence of SARS-CoV-2 antibodies across Canada, as per its mandate beginning in April 2020. The types of projects collecting the data include provincial serosurveys and pan-Canadian studies of the general population, studies focusing on specific age-groups, special, and/or vulnerable subpopulations, and studies in COVID-19 “hotspots”, such as in occupational cohorts.
ASSAYS USED TO DETECT SARS-COV-2 ANTIBODIES
The measurement of antibodies against the spike (S), receptor binding domain (RBD), and nucleocapsid (N) proteins was performed by provincial laboratories, blood operators, and academic research laboratories. The assays used by provincial labs were Health Canada approved commercial ELISAs (Enzyme linked Immunosorbent Assay), including:
- Anti-nucleocapsid IgG assay – Abbott Laboratories
- Anti-nucleocapsid Total Ig antibodies – Roche Laboratories
- Anti-spike Total Ig antibodies – Roche Laboratories
- Anti-spike IgG antibodies – Diasorin
- Anti-SARS-CoV-2 antibodies – MesoScale Discovery multiplex assay
Academic laboratories used both commercial assays and assays that they developed. These laboratory-developed assays used constructs of the SARS-CoV-2 proteins produced at the National Research Council Canada in Montreal or by other reputable, commercially available sources. Antibody detection was done via ELISA or high-throughput chemiluminescence assays. The development of global standards to help calibrate SARS-CoV-2 IgG 8 assays, Biological Arbitrary Units, has helped to decrease heterogeneity and facilitated comparison of and combination of results. For the nucleocapsid assays, the threshold for prior infection was at the manufacturer’s predetermined cut-off for the assay.
This report describes population estimates of SARS-CoV-2 seroprevalence measured across the course of the pandemic, within three distinct phases based on the predominant variant of concern at the time:
- Pre-Delta: before August 1, 2021;
- Delta: August 1 to December 14, 2021; and
- Omicron: December 15, 2021, ongoing.
Anti-spike protein seropositivity can result from vaccination or infection. However, anti-nucleocapsid protein seropositivity only occurs following infection and does not occur following administration of any of the vaccines approved for use in Canada. Therefore, pre-vaccination (prior to Dec 15, 2020) evidence of infection-acquired antibodies includes anti-spike or anti-nucleocapsid seropositivity. After Dec 15, 2020, only anti-nucleocapsid seropositivity is taken as evidence of infection-acquired antibodies.
Seroprevalence results were summarized or presented by:
- Geography: Canada and by province or region (no data are available for Canada’s three territories).
- Average Age: Estimated from the age range of the participants
STATISTICAL MODEL FOR AVERAGE SEROPREVALENCE
Using a time-series approach with the aggregate-level seroprevalence estimates reported by each project, average seroprevalence over time was estimated using a Bayesian multivariate generalized linear model. The model, appropriate for count data where the proportion positive cannot exceed 100%, used a logit link with beta-binomial distributed errors. The predictors for anti-N seroprevalence included natural splines for the time predictor, and province-specific intercepts and coefficients with partial pooling. Hence, the time trend could vary by province while sharing information across provinces. The predictors for anti-S seroprevalence included natural splines for the time predictor and logit of anti-N seroprevalence. The 95% credible intervals were between the 2.5% and 97.5% quantiles of the Markov chain Monte Carlo (MCMC) samples.
The serological assays used are heterogeneous, with different assays using different measurement units. By reporting the proportion of positive samples (seropositivity) we avoid the need to standardize different measurement units. However, it should be noted that different assays have different inherent characteristics (sensitivity, specificity, thresholds), which affect how positive samples are determined. Further, using the nucleocapsid antibody as the indication of prior infection has limitations. These limitations are due to factors such as antibody levels decay over time, low immunogenicity, the inability to know exactly when infection occurred, a lag of one- to two-weeks for maximal antibody generation, and the fact that it is not possible to differentiate between the first and recurrent infections. Consequently, the heterogeneity in estimates has increased during the Omicron era. In BC, for example, the increasing variation in seroprevalence estimates is due to including a broader age range in the source population. Given these limitations, estimates of seroprevalence should not generally be interpreted as direct measures of cumulative infections over the course of the pandemic. However, absolute changes in anti-N seroprevalence over short intervals (e.g., a few months) are likely to accurately represent sudden increases in infections over the interval.