Introduction
To enable the rapid
implementation of early COVID-19 sero-epidemiological investigations to
guide policy decision-making 1, WHO developed new, or
adapted existing,2 standardized generic protocols
branded as the ‘Unity Studies’, a key focus was supporting low- and
middle-income countries to implement these studies 1.
Unity-aligned studies meet criteria including aspects of study design,
study population, sampling, recruitment and using well performing
serological tests 1. The most frequently implemented
studies in the Africa region were seroprevalence, transmission and
vaccine effectiveness 3. As of the
24th February 2023, 179 Unity-aligned SARS-CoV-2
seroprevalence studies were identified in the WHO African region4.
The main methodological challenges reported in a 2022 external
evaluation of the Unity Studies were delays caused by protocol
finalisation, ethical approval and gaining access to funding and test
kits 5. Gaps were also reported in human resources in
laboratory science, data analytics and communications5.
To gain a deeper understanding of the methodological challenges faced
when conducting seroprevalence studies in Africa, we conducted
unstructured interviews with purposively sampled study teams that had
implemented at least one seroprevalence Unity-aligned Study (Burkina
Faso (NFK), Ghana (IOD), Mali (SB), South Africa (JK) and South Sudan
(JFW)). We present four key reoccurring methodological challenges
identified during these interviews and potential solutions to aid future
implementation.
Recruitment and retention
Seroprevalence studies require the inclusion of a predetermined minimum
number of participants to estimate the population prevalence with good
precision 6. Teams faced challenges with recruitment
and retention of participants and therefore, their ability to meet their
minimum sample size.
“People were reluctant to participate because of fear and stigma.”
(Ghana)
“When initially enrolling participants into the study, we informed them
that we would provide them with their test results as soon as they were
available. However, there were delays with the delivery of the test kits
provided by WHO, as such, we were unable to get the results to
participants before the next round of follow-up visits. People didn’t
want to continue with the study and give a new sample before they
received their results.” (Burkina Faso)
These examples add weight to the well documented importance of risk
communication and community engagement in surveillance and operational
research 7–9. Community engagement should be
implemented as a routine activity of enhanced surveillance. Engagement
opportunities include identifying people that the community trusts and
building relationships with them and involving them in decision-making,
utilising human resources from the community, and importantly,
disseminating study findings with all stakeholders7,9,10.
The Malian team reported that one of the main successes of their study
was that:
“We were able to build a young, dynamic, multidisciplinary team capable
of implementing future studies, with involvement of stakeholders at all
levels including policy makers, community leaders, epidemiologists,
immunologists, biologists, data mangers, sociologists and
anthropologists.” (Mali)
Engagement with and continual support of teams such as these is an
important way to ensure timely community-based investigations during
future pandemics.
Sampling frame
In an idealized setting, a perfect sampling frame (study population from
which the sample is selected in order to adequately address study
objectives and extrapolate conclusions appropriately to the broader
target population) would be a list where each person with known
characteristics (e.g. age) is listed once 2,11. Often
this does not exist in real-world settings. At the outset of the
investigation, the Burkina Faso team did not have enough information
about the population they were going to sample as there was no recent
census data. As such:
“We were obliged to conduct a census before we could sample. The last
population census was conducted in 2006 and the data needed to be
updated. After randomly sampling the enumeration areas (the smallest
geographic statistical unit), we visited the households in these areas
and gathered information on the sociodemographic characteristics and
sizes of the households. In this way we did a small census to find out
about our sample and better establish our sampling frame.” (Burkina
Faso)
This process was time and resource intensive and delayed the start of
the study.
Sampling issues also affected the South Sudan team:
“We used satellite imagery to randomly select shelters for inclusion in
the study, some of the sampled households could not be located in the
community due to poor internet access in some remote locations which
affected the use of GPS to locate the shelters, some households were
also empty when study teams arrived, hence the need for supplementary
sampling.” (South Sudan)
Before implementation, partners need to check if an ad-hoc census is
required and feasible based on available resources.
Sample and data
management
Sample and data management was noted as a key area that could be
improved for future studies.
Using easy-to-collect samples like dried blood spots (DBS) significantly
decreases the complexity around sample management. However, prior
validation of serological tests using an appropriate panel of paired
samples, and a sufficient number of samples, to study test accuracy when
using DBS samples is required. Validation of both serological tests and
using DBS samples was challenging at the start of the SARS-CoV-2
pandemic which meant that the robustness of non-validated study results
was unknown. As such, it is recommended to use well validated
serological tests and sample types where possible and that results are
adjusted for test performance (see below).
Manual capturing of participant identifiers on collection forms and
samples lead to errors, which further resulted in delays while these
discrepancies were resolved. A possible solution suggested by several
teams was the development of sample collection SOPs including how
samples will be linked through unique identifiers to epidemiological
data collected such as through the use of barcodes.
Respondents noted that having access to and training on offline
electronic data collection tools is necessary.
“We used web-based data collection forms on a platform the team was
familiar with as we didn’t have time to learn a new software. We had
several challenges with connectivity due to limited cell phone network
connections in our sampling areas. A better tool would have been a
reliable, user-friendly application that you can capture data with
offline.” (South Africa)
To overcome connectivity issues, some teams used offline mobile data
collection tools. They suggested that trainings on how to set-up and use
these tools should be done in the preparation phase, so that when
studies need to be implemented, they are familiar to study teams and
ready for deployment.
Data analysis and
presentation
The WHO Unity team collaborated with SeroTracker 4 to
provided online workshops and tailored support by producing code and
analysis instruction across several analysis tools. Post-hoc adjustment
support was concentrated on population and clustering adjustment to
control for selection bias in participant sampling, and test adjustment
to account for potential biases in estimates introduced by serological
assay performance (sensitivity and specificity values).
One of the main aims of the Unity Studies is to provide robust evidence
for rapid policy decision making 1. To be effective in
this aim, findings need to be presented in a clear and concise manner.
Teams requested assistance with data visualization and advice on how to
best present findings to policy makers 12.
Tools to address
challenges
Tools (some similar to those identified during the Unity Studies
evaluation 5) to aid future study implementation were
identified during the interviews (Figure 1), along with more detail on
how such tools can be tailored for the African context. These included
risk communication and community engagement (RCCE) guidance and tools,
sample collection standard operating procedures (SOPs) and access and
training in a customizable electronic offline data collection tool. For
data management and analysis, having access to software in which the
team is already trained, a draft plan of analysis and data visualization
tools are required. In addition to updating study protocols for
respiratory pathogens of pandemic potential, the Unity Studies
initiative intends to generate and distribute such tools with input from
partners. Countries in the African region have unique public health
challenges, as such the tools created for global application, will need
to be adapted to fit the context of each study population. Trainings and
site visits to implement these tools should be encouraged.