Using sero-epidemiology to monitor disparities in vaccination and infection with SARS-CoV-2

Abstract

Background: As COVID-19 vaccines continue to be rolled-out, the "double burden" of health disparities in both exposure to infection and vaccination coverage intersect to determine the current and future patterns of infection, immunity, and mortality. Serology provides a unique opportunity to measure biomarkers of infection and vaccination simultaneously, and to relate these metrics to demographic and geographic factors.

Methods: Leveraging algorithmically selected residual serum samples from two hospital networks in San Francisco, we sampled 1014 individuals during February 2021, capturing transmission during the first 11 months of the epidemic and the early roll out of vaccination. These samples were tested using two serologic assays: one detecting antibodies elicited by infection, and not by vaccines, and one detecting antibodies elicited by both infection and vaccination. We used Bayesian statistical models to estimate the proportion of the population that was naturally infected and the proportion protected due to vaccination.

Findings: We estimated that the risk of prior infection of Latinx residents was 5.3 (95% CI: 3.2 - 10.3) times greater than the risk of white residents aged 18-64 and that white San Francisco residents over the age of 65 were twice as likely (2.0, 95% CI: 1.1 - 4.6) to be vaccinated as Black residents. We also found socioeconomically deprived zipcodes in the city had high probabilities of natural infections and lower vaccination coverage than wealthier zipcodes.

Interpretation: Using a platform we created for SARS-CoV-2 serologic data collection in San Francisco, we characterized and quantified the stark disparities in infection rates and vaccine coverage by demographic groups over the first year of the pandemic. While the arrival of the SARS-CoV-2 vaccine has created a 'light at the end of the tunnel' for this pandemic, ongoing challenges in achieving and maintaining equity must also be considered.

Funding: NIH, NIGMS, Schmidt Science Fellows in partnership with the Rhodes Trust and the Chan Zuckerberg Biohub.

Figure 1:. Schematic of parameters to be estimated using serosurveillance platform (shown in red, blue and gold).

Red represents the probability of vaccination given prior infection, Pr(vacc|inf), blue represents the probability of prior infection, Pr(inf), and gold represents the probability of vaccination given no prior history of infection, Pr(vacc|uninf).

Figure 2:. Sample characteristics.

(A) Age distribution by hospital week of sample collection within the University of California, San Francisco (UCSF) and San Francisco Department of Public Health (ZSFG) hospital networks. Each point represents a sample and colors correspond to age bins used for analysis. (B) Proportion of samples from a given San Francisco zipcode plotted against the proportion of the San Francisco population within that zipcode. Colors show the percentage of residents below the poverty line within that zipcode, as determined by the American Community Survey 2019.

Figure 3:. Maps showing geographic disparities in SARS-CoV-2 within San Francisco.

Maps show (a) estimated probability of prior infection and (b) probability of vaccination by ZIP code in San Francisco, as of February 2021.

Figure 4:. Stratified seroprevalence by assay and by demographic group.

(A) Univariate Roche seropositivity estimates by age (elicited by prior infection). (B) Univariate Vitros estimates by age (elicited by either prior infection or vaccination). (C) Univariate Roche estimates by race/ethnicity. (D) Univariate Vitros estimates by race/ethnicity.

Figure 5:. Relationship between probability of vaccination and probability of prior infection by race/ethnicity.

a) Probability of infection vs. probability of vaccination by age and race/ethnicity. b) Infographic showing the number of estimated people vaccinated for every one person previously naturally infected in San Francisco within each racial/demographic group.