Seroprevalence of Antibodies to SARS-CoV-2 in Rural Households in Eastern Uganda, 2020-2022

Key Points Question What proportion of a rural population in eastern Uganda has been infected and reinfected with SARS-CoV-2? Findings In this cohort study of 441 participants, by the end of the SARS-CoV-2 Delta wave and before the widespread availability of vaccination, 67.7% of the cohort study population had been infected with SARS-CoV-2. During the subsequent Omicron wave in early 2022, 84.8% of unvaccinated, previously seronegative individuals were infected for the first time, and 50.8% of unvaccinated, already seropositive individuals were likely reinfected. Meaning These results contribute to a growing body of seroprevalence data from Africa suggesting that there are very high SARS-CoV-2 infection rates despite low case ascertainment.


eAppendix. Supplemental Methods
PRISM Border Cohort study enrollment: All households in the parishes of Osukuru, Kayoro, and Buteba parishes in eastern Uganda were enumerated to generate a sampling frame to recruit households into the PRISM Border Cohort study. In August 2020, enumerated households were randomly selected and screened for eligibility to participate. Inclusion criteria for a household to participate included having at least two members aged 5 years or younger. All permanent members of an enrolled household who met eligibility criteria were screened for study enrollment. The cohort was dynamic, so any permanent members that joined an enrolled household were also screened for enrollment.
Covalent antigen coupling to magnetic microsphere beads: Antigens were covalently coupled to MagPlex carboxylated magnetic microspheres (Luminex Corp, Austin, USA). 12.5×10 6 beads for each bead region were suspended by vortexing at medium speed for 2 minutes. The beads were transferred from the stock bottle into a 1.5ml micro-centrifuge tube with low protein binding surface (Eppendorf, UK) to minimize bead loss and centrifuged at 16,000g for 5 minutes. The bead pellet was pulled on the tube side by placing into a magnetic rack for 2 minutes before pipetting off the supernatant. The beads were washed twice by suspension in 1000µl distilled water, vortexed for 30 seconds and centrifuged at 16,000g for 5 minutes. Supernatant was removed after settling the beads on a magnetic rack for 2 minutes. Beads were suspended in 800µl monobasic Sodium Phosphate (NaH2PO4, pH 6.2 activation buffer). To activate the carboxyl surface of the beads, 100µl of 50mg/ml Hydroxysulfosuccinimide (Sulfo-NHS) (Thermo Fisher scientific, UK) was added, vortexed briefly and immediately followed by addition of 100µl of 50mg/ml 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) (Thermo Fisher scientific, UK). The beads were incubated for 20 minutes at room temperature in the dark, wrapped in aluminum foil on a plate shaker. Beads were pelleted by centrifuging at 16,000g for 5 minutes, placed tubes in a magnetic holder for 2 minutes, the supernatant carefully removed, and washed 3 times with 1000µl PBS. Previously determined amounts of antigen (38.5µl/ml spike, 42ug/ml RBD, and 29.8ug/ml nucleocapsid proteins) were added to each bead region in PBS to make a final volume of 1000µl. The beads/antigen suspension was incubated for 2 hours on a rotating shaker in the dark, covered in aluminum foil. Coupled beads were pelleted by centrifugation at 16,000g for 5 minutes and washed of excess antigen 3 times with 1000µl of phosphate buffered saline 0.05% tween20 (PBS-TBN). The antigen coupled beads were suspended in 1000µl storage buffer (PBS-TBN-Sodium Azide) and stored at 4°C.
Assay to measure total IgG: Total IgG responses to spike, RBD, and nucleocapsid proteins were assayed in plasma at 1:400 dilution using the multiplex bead array using the Luminex Magpix machine (Luminex Corp, Austin, Texas). 10ul of each of the three coupled bead regions was added to 5ml PBS-TBN. 50µl of the pooled bead suspension was added to each well, washed and incubated with 50µl of 1:400 test plasma or control sample for 1.5 hours at room temperature. The plates were washed and 50ul of 1/250 goat anti-human IgG rPE labeled antibody (Jackson Immuno Research Laboratories) was added and incubated for 1.5 hours on a shaking platform. The plates were washed, 1xPBS added and read on a MagPix machine.The blank well MFI was deducted from each well to determine the net MFI. We required a minimum of 50 beads per analyte for a measurement to be included in this analysis. Cutoff for seropositivity was established as the highest value among the negative controls; this was determined to be 516 for background-subtracted spike median MFI, and 0.024 for relative spike antibody concentration.
Estimating seroprevalence and attack rate using a Bayesian measurement model: The test performance characteristics of a single assay (i.e., sensitivity (Se) and specificity (Sp)) can be determined from a 2 x 2 table of positive/negative control samples and their binary classification on that assay using a binomial model as follows: In Equation 1, N_pos represents the total number of positive control samples tested and y_pos represents the number of positive control samples that tested positive. In Equation 2, N_neg represents the total number of negative control samples tested and y_neg represents the number of negative control samples that tested negative. For a given serosurvey where N total samples were tested and y samples tested positive, the adjusted seroprevalence p_adj can be estimated using the binomial model in Equation 3: The same approach can be used to estimate the attack rate.
Testing for clustering of seroconversions within households using pairwise odds ratios: The pairwise odds ratio is the odds of seroconversion for an individual in a household if another individual from the same household seroconverted, relative to their odds of seroconversion if the individual did not seroconvert. Pairwise odds ratio values greater than 1 indicate clustering within households. We estimated within-age category and between-age category pairwise odds ratios by time interval, and assumed that the pairwise odds ratios were constant across all households. The probabilities and odds were estimated in a multinomial Bayesian framework with a 4-simplex distribution parameter representing the possible serostatus permutations in a pair of individuals.  Figure 3, boosting was defined as a ≥ 4 fold increase (orange line). The Round 3 antibody response is shown on the x-axis, and the fold change between the Round 4 and Round 3 antibody response is shown on the y-axis. This figure is limited to participants who were vaccinated at Round 4, and were separated by tertiles of Round 3 response (the second tertile is shown in the gray shaded rectangle). The proportion of individuals within each tertile that demonstrated antibody boosting is shown in orange text at the top. The colors of the points represent the manufacturer of the vaccine received.