SARS‐CoV‐2 seroprevalence and implications for population immunity: Evidence from two Health and Demographic Surveillance System sites in Kenya, February–December 2022

We sought to estimate SARS‐CoV‐2 antibody seroprevalence within representative samples of the Kenyan population during the third year of the COVID‐19 pandemic and the second year of COVID‐19 vaccine use.


Conclusion:
More than 70% of Kilifi residents and 90% of Nairobi residents were seropositive for anti-S IgG by the end of 2022.There is a potential immunity gap in rural Kenya; implementation of interventions to improve COVID-19 vaccine uptake among sub-groups at increased risk of severe COVID-19 in rural settings is recommended.

| INTRODUCTION
Serosurveillance for SARS-CoV-2 antibodies emerged as an important tool for estimating the cumulative incidence of SARS-CoV-2 infection in the early phase of the COVID-19 pandemic.It was particularly important in settings with low COVID-19 testing levels, including many countries in Africa. 1 In these settings, cumulative incidence revealed by SARS-CoV-2 serosurveillance far outstripped the rate of infection inferred from case detection. 2In the current context, SARS-CoV-2 serosurveillance remains important for assessing temporal changes in seroprevalence, including waning, and identifying potential gaps in population immunity to inform priorities for COVID-19 prevention measures, including vaccination.
By June 2020, roughly 3 months after the first confirmed case of COVID-19 in Kenya, seroprevalence of SARS-CoV-2 anti-spike (anti-S) immunoglobulin G (IgG) was 4% among blood donors. 3Continued serosurveillance among blood donors in Kenya showed temporal increases in anti-S IgG seroprevalence, reaching 48% by March 2021, 4

| Study design and participants
Two cross-sectional surveys were conducted at each of two sites, the Kilifi HDSS and the Nairobi Urban HDSS.The characteristics of these HDSS sites have been described in detail previously. 6,7The Kilifi HDSS is located in a rural area within Kilifi County in south-eastern coastal Kenya, whereas the Nairobi Urban HDSS is located within Nairobi County, the capital city of Kenya (Figure S1).The population of the Kilifi HDSS was 308,581 (April 2022 census) and about 90,000 at the Nairobi Urban HDSS (October 2021 census).
We used similar methods as previous SARS-CoV-2 serosurveys at the sites. 5In brief, for each serosurvey a random age-stratified sample of 850 participants was drawn from the respective HDSS site population registers.The sample drawn at each site was independent of that drawn for the surveys conducted in 2020-2021.The simple random sample included 100 children in each 5-year age band <15 years, 50 individuals in each 5-year age band between 15 and 64 years, and 50 adults aged ≥65 years.Individuals were eligible to participate in the study if they were resident in the respective HDSS, had no contraindication for blood sample collection, and provided consent.Individuals were considered not contactable after three unsuccessful attempts to visit them.To compensate for participants who were not found at home, a replacement sample was drawn at random from the HDSS population register.
The study research protocol was aligned to the World Health Organization (WHO) UNITY methods for COVID-19 serosurveillance studies. 8

| Participant consent statement
Written parental/guardian consent was obtained for participants aged <18 years, accompanied by written assent for children aged 13-17 years.Written informed consent was obtained from participants aged ≥18 years.

| Data and sample collection
Sociodemographic (e.g., age, sex, and location of residence) information and medical history data (e.g., recent COVID-like symptoms, COVID-19 vaccination history, and previous confirmed SARS-CoV-2 infection) were collected from each participant (Supporting Information).COVID-19 vaccination status was ascertained using either official records (i.e., COVID-19 vaccination certificate or text message confirmation from the national COVID-19 vaccine registry) or verbal report.A single 2 mL (children aged <5 years) or 5 mL (individuals aged ≥5 years) venous blood sample was collected in a heparin-coated tube from each participant and labeled with a unique identifier.

| Laboratory testing
Plasma was extracted from venous blood samples and tested for anti-S IgG to identify either infection-or vaccination-induced antibody response, and for anti-nucleoprotein (anti-N) IgG, to identify infection-induced antibody response. 9Testing was performed using validated KEMRI-Wellcome Trust Research Programme ELISAs.Sensitivity and specificity for the anti-S IgG ELISAs were, respectively, 93% (95% confidence interval 88-96%) and 99% (98-99%). 3Sensitivity and specificity for the anti-N IgG ELISA were 83% (76-88%) and 91% (86-95%).Target-specific IgG positivity was defined as a ratio of the sample optical density (OD) over negative control OD > 2. The WHO International Standard (IS) for anti-SARS-CoV-2 immunoglobulin (NIBSC code 20/136) was included in each anti-S IgG ELISA run and used to calculate sample-specific binding antibody concentrations in binding antibody units per milliliter (BAU/mL).A binding antibody concentration of 1000 BAU/mL was assigned to the WHO IS, as recommended by NIBSC. 10 Sample-specific binding antibody concentrations were calculated by dividing each sample-specific OD ratio by the run-specific IS OD ratio after which the quotient was multiplied by 1000.

| Statistical analysis
The target sample size of 850 participants was sufficient to measure antibody seroprevalence of 50% with an associated 95% confidence interval (CI) of ±3%.Crude seroprevalence was calculated as the proportion of seropositive samples over all samples, along with exact binomial 95% CIs.
Population-weighted anti-S IgG seroprevalence with associated 95% CrIs was also calculated.Anti-S IgG seroprevalence within the entire sample was not adjusted for test performance as validation of the ELISA was performed among unvaccinated individuals; for example, adjustment for test performance using current sensitivity and specificity estimates would overestimate anti-S seroprevalence if in truth assay sensitivity is higher among COVID-vaccinated individuals.To estimate infection-induced seroprevalence, Bayesian populationweighted anti-S IgG seroprevalence was estimated among the subset of participants who were COVID-unvaccinated.Anti-S IgG seroprevalence among COVID-unvaccinated participants was not adjusted for test performance for ease of interpretation given that anti-S IgG seroprevalence within the entire sample was similarly not adjusted for test performance, as described above.To assess the magnitude of inherent assay bias in a sensitivity analysis, Bayesian test-performance adjusted anti-S seroprevalence was also estimated among only COVIDunvaccinated participants.
To assess the contribution of infection to seroprevalence levels, population-weighted and test-adjusted anti-N IgG seroprevalence was calculated separately for COVID-vaccinated and COVIDunvaccinated participants.Although inactivated COVID-19 vaccines can induce anti-N IgG responses, 11 they have had limited use in Kenya; by December 2022, only 31,000 doses of an inactivated virus COVID-19 vaccine (Sinopharm) had been administered out of approximately 23 million doses of vaccine administered nationally. 12Therefore, it was deemed appropriate to perform adjustment for test performance for the anti-N seroprevalence estimates within the entire sample.
For each set of seroprevalence data, seroprevalence was computed overall, by sex and by seven age strata (<16; 16-24; 25-34; 35-44; 45-54; 55-64; ≥65 years).These age strata were selected to allow comparison with seroprevalence data from the same HDSS sites in 2020-2021. 5verse cumulative distribution curves (RCDCs) for anti-S IgG concentrations were plotted.Kolmogorov-Smirnov tests were used to test the equality of the overall distributions across sites, as well as across sex, age group, and COVID-19 vaccination status within each site.The comparison of the distribution of antibodies for COVIDunvaccinated versus COVID-vaccinated individuals was restricted to those aged ≥15 years to minimize confounding by age given that the COVID-19 vaccine rollout has been focused on individuals aged ≥15 years.The proportion of individuals with anti-S IgG ≥ 154 BAU/mL was calculated; ≥154 BAU/mL has been proposed as a threshold of protection against infection with wild-type SARS-CoV-2 following COVID-19 vaccination. 13pulation weighting and adjustments for test-performance were performed in R with RStan, and all other analyses were performed using Stata.

| Role of the funding source
The research was funded by the Bill and Melinda Gates Foundation (INV-039626).The funder had no role in study design, data analysis, data collection, data interpretation, or writing of the report.The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit the paper for publication.

| RESULTS
The surveys at the Kilifi HDSS site were conducted between February The median age of the participants during the surveys ranged between 26 and 27 years.The proportion of females sampled during Survey 2 at the Nairobi Urban HDSS was significantly lower than during Survey 1 (47.5% vs. 37.1%; p < 0.001); this was likely due to differential participant finding, whereby the proportion of females present out of those visited was smaller during Survey 2 than during Survey 1 (Table S1).Among participants in the Kilifi HDSS reporting COVID-19 vaccine receipt, vaccination was ascertained using an offi-  S2).At both sites, a higher proportion of participants reported ≥1 COVID-like symptoms in the  S2).The top three symptoms across the surveys were cough (all surveys), headache (all surveys), runny nose (all surveys except Kilifi HDSS Survey 2), and fever (Kilifi HDSS Survey 2).These events were most commonly reported among children aged <16 years (Figures S3 and S4).

| Anti-S IgG seroprevalence and concentrations among all participants
Population-weighted anti-S IgG seroprevalence was 69.1% (95% CrI 65.8-72.3%)among all participants at the Kilifi HDSS site during Survey 1, increasing to 78.1% (75.1-80.9%)by Survey 2. Between Surveys 1 and 2, seroprevalence increased among children aged <16 years and adults aged 25-34 years (Figure 2, Table S3).Overall seroprevalence at the Nairobi Urban HDSS site during Survey 1 was 88.5% (86.1-90.6%)and was 92.0% (89.9-93.8%)by Survey 2. During both surveys, children <16 years of age within the Nairobi Urban HDSS tended to have lower seroprevalence than some adult age groups (Figure 2, Table S3).Seroprevalence at the Kilifi HDSS site was significantly lower than at the Nairobi Urban HDSS site at both Surveys 1 and 2 as illustrated by non-overlapping 95% CrIs.During Survey 1, seroprevalence was lower across all age and sex strata among Kilifi HDSS residents compared with Nairobi Urban HDSS residents.In Survey 2, seroprevalence was lower among all age and sex strata within the Kilifi HDSS than within the Nairobi Urban HDSS except for individuals aged 16-24 years and adults aged 25-34 years (Figure 2, Table S3).
The geometric mean anti-S IgG concentration was 272 BAU/mL (95% CI 247-299 BAU/mL) and 230 BAU/mL (213-248 BAU/mL) among Kilifi HDSS residents during Surveys 1 and 2, respectively.It was 555 BAU/mL (515-597 BAU/mL) and 439 BAU/mL (411-470 BAU/mL) during Surveys 1 and 2, respectively, within the Nairobi Urban HDSS (Table S4).The proportion of individuals with anti-spike IgG ≥ 154 BAU/mL was 62.7% and 65.5% during Surveys 1 and 2, respectively, within the Kilifi HDSS and 85.9% and 85.5% during Surveys 1 and 2, respectively, at the Nairobi Urban HDSS (Figure 3 and Table S5).Overall anti-S IgG concentrations were significantly lower within the Kilifi HDSS than within the Nairobi Urban HDSS for both surveys ( p < 0.001).Within the Nairobi Urban HDSS, the antibody concentrations were significantly lower at Survey 2 than at Sur-   among UK blood donors was 7%. 16By December 2021, coverage with ≥1 doses of COVID-19 vaccine in the United Kingdom was 69%, and anti-S IgG seroprevalence was 98%, whereas anti-N IgG seroprevalence was 23%. 17 1 year after identification of the first confirmed SARS-CoV-2 infection locally which also coincided with the rollout of the COVID-19 vaccination program by the Government of Kenya.The COVID-19 vaccination program initially targeted frontline workers, older adults aged ≥58 years, and younger adults with comorbidities.It soon opened up to all adults, then to children aged ≥15 years in November 2021, and currently targets individuals aged ≥12 years.To date, the national program has included the following COVID-19 vaccines: Oxford/AstraZeneca (Covishield), Pfizer (BNT162b2), Moderna (mRNA-1273), Johnson & Johnson (Ad26.CoV2.S), and Sinopharm (BBIBP-CorV).While convenience samples, such as blood transfusion donors, can be leveraged to rapidly generate seroprevalence estimates, serosurveillance among randomly selected residents of Health and Demographic Surveillance System (HDSS) sites avoids the selection biases inherent in convenience sampling and provides a more representative sample of the general population.By May 2021, 2 months after initial rollout of COVID-19 vaccine, serosurveillance among an age-stratified random sample of residents of three HDSS sites found anti-S IgG seroprevalence ranging from 25% to 50%. 5 We undertook a repeat serosurvey at two of these three HDSS sites to assess infection-and vaccination-induced seroprevalence within the general population during the third year of the COVID-19 pandemic and the second year of the COVID-19 vaccination program.
Crude COVID-19 vaccine coverage was calculated as the proportion of individuals reporting vaccination divided by the entire sample.Exact binomial 95% CIs were also calculated.Because the population sample did not represent the age structure of the target population, Bayesian population weighting was performed to generate population-weighted coverage and associated 95% credible intervals (CrI).Coverage with ≥1 doses of COVID-19 vaccine and full vaccination coverage were estimated overall and restricted to participants aged ≥15 years.Full vaccination was defined as receipt of at least one dose of Johnson & Johnson vaccine or receipt of ≥2 doses of other COVID-19 vaccines.

15 and May 8 ,
2022 (Survey 1; median March 23, 2022), and between August 24 and November 27, 2022 (Survey 2; median October 3, 2022).The surveys at the Nairobi Urban HDSS site were conducted within a similar timeframe: March 8-June 22, 2022 (Survey 1; median May 1, 2022) and September 27-December 8, 2022 (Survey 2; median November 2, 2022).At the Kilifi HDSS site, 853 participants were sampled during Survey 1 (representing 73.0% of those approached), and 850 were sampled during Survey 2 (i.e., 75.6% of those approached).One sample collected during Survey 1 at the Kilifi HDSS site, representing 0.1% of the 853 samples collected, was not tested due to a sample labeling issue that had not been resolved by the time antibody testing was conducted.The number of participants sampled during Survey 1 at the Nairobi Urban HDSS site was 851 (87.1% of those approached) and 850 (88.0% of those approached) during Survey 2 (Figure 1).Survey 1 at both sites began after the Omicron BA.1 wave, with the sample collection period at the Nairobi Urban HDSS site continuing up to the first half of the Omicron BA.4/BA.5 wave.Survey 2 at the sites occurred after the Omicron BA.4/BA.5 wave, extending into part of the Omicron BQ.1 wave (Figure S2).
cial record in 28.7% (27 of 94) during Survey 1 in the Kilifi HDSS and in 50.9% (59 of 116) during Survey 2. In the Nairobi Urban HDSS, COVID-19 vaccine receipt was confirmed using an official record in 27.8% (79 of 284) of participants during Survey 1 and in 73.8% (225 of 303) during Survey 2. COVID-19 vaccine uptake did not change substantially between the surveys and was significantly lower among residents of the Kilifi HDSS than among Nairobi Urban HDSS residents.At Survey 2, population-weighted coverage with ≥1 doses among all enrolled participants was 11.1% (95% CrI 9.1-13.2%)among the Kilifi HDSS participants against 34.2% (30.7-37.6%)among the Nairobi Urban HDSS participants.When restricted to participants aged ≥15 years of age, coverage for ≥1 doses of COVID-19 vaccine was 17.7% (12.3-23.2%)within the Kilifi HDSS and 49.4% (41.7-56.7%)within the Nairobi Urban HDSS at Survey 2. The distribution of COVID-19 vaccine brand doses received is provided in the Supporting Information; only one participant reported receipt of at least one dose of Sinopharm vaccine (Table

vey 1 .
Within the Kilifi HDSS, antibody concentrations were significantly lower among children <16 years than among all other age groups F I G U R E 1 Study participant flow.T A B L E 1 Population-weighted anti-spike IgG seroprevalence among COVID-unvaccinated individuals by site, sex, and age category.
seroprevalence among COVID-unvaccinated participants-who made up the majority of the study sample-was 76% by November 2022 in Kilifi and 90% by December 2022 in Nairobi, indicative of infectioninduced immune responses.Furthermore, 54% of COVIDunvaccinated participants in Kilifi and 85% in Nairobi were positive for anti-N IgG, supporting the inference of infection-driven anti-S IgG seroprevalence in this group.Anti-S IgG seroprevalence among COVID-vaccinated individuals appeared to be driven by both infection and vaccination as anti-N IgG seroprevalence in that group was 58% in Kilifi and 99% in Nairobi.However, the low number of COVIDvaccinated individuals suggests a marginal role of hybrid immunity 14 at the population level.Our findings also indicate temporal increases in SARS-CoV-2 seroprevalence within the general population in Kenya.Anti-S IgG seroprevalence by the end of 2022 had increased more than threefold within the Kilifi HDSS site and more than twofold within the Nairobi Urban HDSS site since May 2021, when it was 20% and 40%, respectively. 5This temporal increase was driven by COVID-19 vaccination rollout beginning in March 2021, as well as by the four COVID-19 waves occurring since May 2021, that is, the Delta variant, Omicron BA.1, 15 Omicron BA.4/5, and Omicron BQ.1 waves.Between May 2022 and November 2022, increases in seroprevalence in Kilifi were observed among children aged <16 years-likely due to infection given low COVID-19 vaccine uptake in that age group-and among adults aged 25-34 years.Seroprevalence did not change appreciably between June and December 2022 within the Nairobi Urban HDSS; given the high seroprevalence estimates in Nairobi despite nonadjustment for test performance, this suggests near-ubiquitous presence of anti-SARS-CoV-2 antibodies.The findings from Kenya contrast with evidence from settings like the United Kingdom where high anti-S IgG seroprevalence was achieved primarily through COVID-19 vaccination rather than through natural infection.By December 2020, just before phased rollout of COVID-19 vaccine in the United Kingdom, anti-S IgG seroprevalence Thus, natural infection played a small role in the large increase in SARS-CoV-2 antibody prevalence observed in the United Kingdom between December 2020 and December 2021.

F I G U R E 4
Population-weighted and test-adjusted anti-nucleoprotein IgG seroprevalence among study participants overall and by COVID-19 vaccination status, sex, and age category within the (A) Kilifi Health and Demographic Surveillance System (HDSS) and the (B) Nairobi Urban HDSS.
Ethical approval was obtained from the Kenya Medical