In 2022, one survey each at the Kibera PBIDS and Asembo sites was conducted, and two surveys each at the Kilifi HDSS, Nairobi Urban HDSS ¹⁷ , and Manyatta HDSS sites. One survey at each site was conducted in 2023.

The Kilifi and Nairobi HDSS sites applied a repeated cross-sectional survey design. Using the respective HDSS site’s census as a sampling frame, a randomly selected, age-stratified sample of 850 individuals were selected. The sample included 100 children in each five-year age band <14 years of age, 50 individuals in each five-year age band between 15 and 64 years of age, and 50 adults aged ≥65 years. This sample size was sufficient to measure seroprevalence of 50% with precision (95% confidence interval [CI]) of 47–53%. The Manyatta HDSS, Kibera PBIDS and Asembo PBIDS sites used a longitudinal survey design. A random sample of 225 households, targeting a total of 900 members in the sampled households, was selected from each site’s respective population register. This sample size was sufficient to measure seroprevalence of 50% with an associated 95% CI of ±5% assuming a design effect of 2.

Trained field staff visited selected participants at the households and sought written informed consent for participation. Consent was obtained from the guardians of children aged <18 years; children aged 13–17 years provided written assent as per local ethical guidelines. Approximately 2mL of venous blood were collected from participating children <5 years of age and approximately 5mL from individuals aged ≥5 years. Blood samples were collected by study-employed phlebotomists either at home or at a nearby health facility using either heparinized or serum separator tubes. They were then processed to separate out plasma or serum at a central laboratory. The plasma or serum samples were then biobanked at -80°C for future testing.

Ethical approval to conduct the serosurveys was obtained from the Kenya Medical Research Institute Scientific and Ethics Review Unit (4085, 4168), the Oxford Tropical Research Ethics Committee (44-20), and the London School of Hygiene and Tropical Medicine Research Ethics Committee (26950). Kibera, Asembo, and Manyatta serosurveys were also reviewed by US CDC and were conducted consistent with applicable federal law and CDC policy as provided for in the Code of Federal Regulations (45 C.F.R part 46 and 21 C.F.R. part 56).

Samples were tested for anti-SARS-CoV-2 immunoglobulin G (IgG) antibodies using validated assays. Samples collected at the Kilifi HDSS and Nairobi HDSS sites were tested centrally at the KWTRP laboratories. Those samples were tested for anti-spike IgG using the KWTRP anti-S IgG ELISA with 93% sensitivity (95% CI 88-96%) and 99% specificity (95% CI 98-99%). Development and validation of the assay has been previously described ¹ . Samples were considered positive for anti-S IgG when the ratio of the sample optical density (OD) over the negative control OD was >2. In addition to qualitative (positive/ negative) anti-S IgG results, binding antibody concentrations were also calculated by calibrating the anti-S IgG ELISA optical densities against the WHO International Standard ¹⁸ . Samples collected during the initial round of 2022 surveys at the Kilifi HDSS and Nairobi HDSS sites were also tested for anti-nucleoprotein IgG using the KWTRP anti-N ELISA with 83% (95% CI 76-88%) sensitivity and 91% (95% CI 86-95%) specificity.

Samples collected at the Kibera PBIDS site were tested for anti-S IgG at the CDC-supported KEMRI-CGHR laboratories in Nairobi using the SCoV-2 Detect™ IgG ELISA kit (InBios International, Inc.), whose performance was similar to that of the KWTRP ELISA, with 92% sensitivity and 99% specificity ¹⁹ . Samples collected at the Asembo PBIDS and Manyatta HDSS sites were tested for anti-S IgG using the InBios IgG ELISA kit at the CDC-supported KEMRI-CGHR laboratories in Kisumu.

As of December 2022, SARS-CoV-2 serosurveillance revealed sustained temporal increases in cumulative SARS-CoV-2 infection with rural-urban heterogeneity in seroprevalence ¹⁷ . The latter finding underscores the need for targeted COVID-19 vaccination strategies, given limited public health resources, such as the prioritization of vulnerable groups residing in rural settings. Ongoing SARS-CoV-2 serosurveillance will be important for informing our understanding of age-specific population immunity dynamics, including waning, and the implications for the national COVID-19 vaccination strategy ²⁰ .

To demonstrate the feasibility of integrated serosurveillance and to support the rational use of samples, we proposed to leverage existing assays at KWTRP for optimization or development of multiplex assays. We prioritized the optimization/ development of assays that would support the generation of policy-relevant outputs (Objective #5 of the collaboration). With the support of the Netherlands National Institute for Public Health and the Environment (RIVM), KWTRP investigators had previously implemented a multiplex assay testing for IgG antibodies to seven vaccine preventable diseases (VPDs), i.e., diphtheria, pertussis, tetanus, measles, mumps, rubella and varicella. In addition, the development of an arbovirus multiplex assay began in 2023. Optimization of the seven-plex VPD assay under KEMIS was pursued as VPD serosurveillance can inform priorities for disease-specific immunization programs while arbovirus serosurveillance can inform pandemic preparedness efforts. These applications of serosurveillance can be achieved through identification of population immunity gaps, prediction of outbreak risk, and prediction of the impact of disease control and prevention interventions ^{21– 23} . For example, the generation of age-specific measles serological profiles as part of KEMIS can inform refinement of national vaccine deployment policies including targeted supplementary immunization activities and optimization of the infant immunization schedule.

The RIVM VPD multiplex assays have been validated and described in detail previously ^{24, 25} . These VPD assays were initially optimized for implementation at KWTRP on MAGPIX, a multiplex bead-based fluorescence immuno-assay. We sought to further optimize the seven-plex VPD assay to be run on a different flow cytometry multiplex platform, the Luminex 200. Briefly, commercial antigens (diphtheria toxoid, pertussis toxin, tetanus toxoid, measles virus, mumps virus, rubella virus, and varicella zoster virus) were conjugated to carboxylated polystyrene or magnetic microspheres following standard operating procedures. Successful conjugation was confirmed by assessing the mean fluorescence intensity (MFI) of a reporter conjugated to a secondary antibody bound to the primary antibody of a known sample. Conjugation was validated by assessing correlation after testing a panel of samples with known concentration provided by the RIVM (i.e., validation panel); assessing the linearity and recovery of calibration curves using established standards; and evaluating variation in the concentration of controls as measured in different runs. Samples were run in duplicate dilutions (1:200 and 1:4000) on the Luminex 200 and plates were read using the Luminex xPONENT ^® software. Antibody concentrations were interpolated from MFI adjusted for background fluorescence using five-parameter logistic regression-fit standard curves generated in Belysa software (Sigma). Samples with >20% intra-assay variation were run again. During optimization of the seven-plex assay, we used the same IgG antibody positivity cut-offs used by RIVM for diphtheria, tetanus, measles, mumps, rubella and varicella; these cut-offs were supported by prior evidence ( Table 2) ^{26– 31} . An arbitrary cut-off of 5 IU/mL was used for optimization of the pertussis component of the seven-plex VPD assay; there is no acknowledged correlate of protection of pertussis, though IgG antibody levels ≥20 IU/mL are thought to indicate recent vaccination or infection ^{32, 33} .

The antigens to be included in the multiplex arbovirus assay will be informed by arbovirus recombinant protein ELISAs existing or under development at KWTRP for Rift Valley Fever (glycoprotein [Gn] and C (Gc) antigens) ³⁴ , dengue (non-structural protein 1 [NS1] antigen and envelope [E] protein) and chikungunya (envelope protein 1 [E1] and 2 [E2] antigens). Assay-specific validation will be performed using gold standard negative and positive panels based on PCR or neutralization titers. The assay-specific cut-offs for seropositivity will be designed to maximize specificity. Antigen conjugation will be conducted at KWTRP and conjugation validation performed following a similar approach as for the VPD assay.

Development of multiplex assays for other pathogens will be contingent on funding and the outcome of an exercise on the identification of priority pathogens for serosurveillance, described elsewhere in this article. One aliquot of samples from all other sites is planned to be shipped to Kilifi for testing of antibodies to other pathogens, as agreed upon by the KEMIS collaborators.

Over the first two years of the pandemic in Kenya, a series of mathematical models were developed by a team from KEMRI-Wellcome Trust Research Programme and the University of Warwick Zeeman Institute to support the Kenya Government’s emergency response to COVID-19. This resulted in an enhanced understanding of the underlying epidemiology ³⁵ , predictions of disease incidence from introduced variants of concern ³⁶ and forecasts of the potential impact of non-pharmaceutical interventions and vaccines on cases of disease, hospitalisation, and deaths ³⁷ . This work was underpinned by key data sets, including national PCR and lateral flow SARS-CoV-2 testing data, genomic surveillance of variants, and serosurveys at various time points from locations across the country. The serosurveys, funded through BMGF and FCDO grants as described in the introduction, defined the extent of population transmission of SARS-CoV-2, fundamental to model calibration ³⁵ .

Modelling work for KEMIS continues this evidence synthesis for COVID-19, to capture the dynamics of the virus as it is likely to move towards endemicity. The availability of age and spatial stratified seroprevalence data has enabled the exploration of the potential impact of age and geographical targeting of COVID-19 vaccines and boosters, forecasting the potential impact of future introduction of variants into the population, and the effects of both naturally acquired and vaccine acquired immune decay on disease risk and burden.

The broadened range of infectious disease serosurveys will enable modelling that aims to increase understanding of the epidemiology of vaccine preventable childhood diseases and vector-borne viruses. For example, in the case of mumps, KEMIS will produce the first cross-sectional antibody surveys in Kenya which will inform on the pristine (i.e. in the absence of vaccination) age-stratified rates of transmission of this virus. A compartmental transmission dynamic model incorporating this information will provide predictions of the impact of adding mumps containing vaccine to the current national 2-dose measles-rubella containing vaccine strategy, assessing the potential impact on meningitis and orchitis disease.

An extensive stakeholder engagement exercise was designed to generate a list of priority pathogens for serosurveillance in Kenya (Objective #4 of the collaboration). The study team engaged purposively selected stakeholders from the Kenya Ministry of Health, county departments of health, global and non-governmental organisations involved in disease surveillance, as well as academic and research organisations in developing the list of priority pathogens. A modified Delphi technique was used, involving three stages of stakeholder engagement. At the first engagement, key informants were invited to an in-depth interview to identify a ranked list of priority pathogens for serosurveillance. After the interview, respondents were given an opportunity to refine their recommendations via email – the second point of engagement. The third engagement involved a consensus building workshop where participants representing the stakeholders described above developed a ranked, consensus list of priority pathogens for serosurveillance. The outcome of the consensus building workshop has been described in a Policy Brief ³⁸ and a scientific publication describing the entire qualitative exercise is underway.