Detailed methods have been described previously (7). Briefly, ten facilities in each of Chadiza, Chipata, and Chongwe districts, and nine in Lusaka district enrolled up to 20 pregnant women aged 15–49 years attending their first ANC visit (ANC1) each month from September 2021 to September 2022. Consenting participants completed an electronic questionnaire and provided a blood sample for the SARS-CoV-2 antibody test. Facility-level challenges and best practices were identified by district research assistants (RAs) during weekly quality assurance and control visits to the study sites. The RAs reviewed study documentation to ensure the accuracy and completeness of data and improve efficiency in the implementation processes. During the health facility visits, staff sought clarity and reported issues identified during participant recruitment and data collection.
The study facilitated WhatsApp groups for clinic teams through which ad hoc problems experienced during implementation were shared and possible solutions provided rapidly. The study and data coordinator reviewed records submitted to the server daily and performed weekly data extractions for presentation on a dashboard during check-in meetings. All identified gaps were communicated to the district RAs for follow-up and resolution during their weekly visits to health facilities. Findings and actions taken were relayed to the study coordinator immediately or during weekly check-in meetings.
Study Implementation Process and Progress
Stakeholder Engagement
In the initial stages of implementation, study staff were primarily involved in preparatory activities including engaging key stakeholders and facilitating community sensitization meetings. Following approval of the CIDRZ study protocol by the National Health Research Authority (NHRA) –Zambia’s health research regulatory body – all relevant study documentation was shared with the provincial and district health directors to obtain authorization for implementation in selected districts and health facilities. PATH submitted an amendment to its existing malaria surveillance study protocol to the University of Zambia Research Ethics Committee (UNZABREC) before submitting it to the NHRA.
Study Trainings and Community Sensitization Meetings
Health care workers (HCWs) (midwives and laboratory technicians) and community-based volunteers (CBVs) participated in the trainings which included practical sessions tailored to staff-specific roles during implementation. Facility staff were trained in the protocol, research ethics, and data and sample collection and completed an assessment to determine levels of competence. After the trainings, staff led two-hour community sensitization meetings with members of neighborhood health committees (NHCs) within their health facility’s catchment areas. The study coordinator, data coordinator, and RAs were on hand to reinforce any missing information and to help answer questions.
COVID-19 Mitigation Measures
Before registering for attendance at training sessions, all potential study staff were screened for COVID-19 symptoms using an MOH screening checklist and tested for SARS-CoV-2 using a rapid antigen test irrespective of symptoms to reduce the likelihood that trainings would be foci for SARS-CoV-2 transmission. Staff who tested positive were not allowed to participate and were referred to a health facility to have a confirmatory nucleic acid amplification test (NAAT). Trainings were limited to 30 participants to accommodate social distancing. Wearing of masks, hand hygiene, Infection Prevention and Control (IPC), and COVID-19 vaccination were encouraged for both staff and study participants during training sessions and at the facility level.
Sampling and Participant Recruitments
CBVs provided details of the ANC COVID-19 Surveillance Study to pregnant women during routine health education sessions. A health facility-specific sampling interval was defined based on historic monthly mean ANC attendance during fiscal year 2020 and was initially used by CBVs to systematically select women for participation. However, due to staff unavailability and fluctuating ANC attendance, some facilities transitioned to a convenience sampling approach to recruit the expected number of participants. Pregnant women were approached for eligibility screening and those eligible were provided with an information sheet and the consenting process was initiated. Thereafter, the participant proceeded with registering for ANC, and routine services were provided by a midwife.
Data Collection and Storage
Study questionnaires were programmed using Open Data Kit (ODK) and administered via a tablet, but paper-based questionnaires were available as backups. Regardless of the process used to collect data from the participant, data were reviewed for accuracy and completeness while the participant was still at the health facility and before forms were uploaded to a central server. A paper log was completed by the midwife to create a paper trail in case of mislabeling. The paper log was stored at the health facility and reviewed by the district RA every week. To maintain confidentiality during recruitment, unique 16-digit study identification numbers were printed on adhesive labels in triplicate for use on each participant’s DBS card, hard copy questionnaire (if used), and consent form. Initially, the identification numbers were manually keyed into the electronic questionnaire by trained midwives. Logic checks for the identification number had not been programmed into the electronic questionnaire and some facility staff entered incorrect numbers which could not be matched to the identification numbers on the DBS card or laboratory form. Consequently, some staff inadvertently duplicated study identification numbers, leading to a total of 358 (12%) incorrect identification numbers. These problems were identified at the time of scrutinizing and comparing the LIMS and ODK data extracts four months after the first participant was recruited into the study.
Sample Collection, Transfer, and SARS-CoV-2 Serologic Testing
At the time of drawing blood from the participant for routine testing of HIV and syphilis using a finger prick or venipuncture, a dry blood spot (DBS) sample was collected for SARS-CoV-2 antibody testing. Specimens and questionnaires were immediately labeled with an identification sticker that had a pre-printed unique study ID. At the beginning of implementation, samples from Chipata, Chongwe, and Lusaka districts were transported via the existing HIV viral load sample courier system directly to the testing laboratory at the University Teaching Hospital (UTH). However, when problems were identified at the time of reviewing laboratory data extracts, the transportation of samples was assigned to respective district RAs who were required to check the DBS cards for accuracy and completeness at every quality assurance visit to the health facilities before transporting them to the CIDRZ laboratory for cataloging in the laboratory information management system (LIMS) and then to UTH for testing. Chadiza samples were transported by PATH study personnel to the National Malaria Elimination Centre (NMEC) laboratory.
Testing for SARS-CoV-2 was done using the Tetracore FlexImmArray SARS-CoV-2 Human IgG Antibody Test (Tetracore Inc., Rockville, MD) and analyzed using a MAGPIX machine (Luminex Corporation). Samples were tested according to manufacturer instructions as described by Tartof et al (7). Sample analysis was batched, and up to 20% of samples were re-run for quality assurance. Four controls were incorporated into every test plate and three different SARS-CoV-2 antigens were used in the multiplex assay (receptor binding domain (RBD) of the spike-1 protein, nucleocapsid protein (NP), and an RBD-NP hybrid antigen).
Study Outputs and Outcomes
Planned study activities, targets, and actual achievements during implementation are shown in Table 1. Thirty-nine out of 40 planned health facilities participated in the study. The only private hospital that was selected to participate in the study withdrew from the study at the beginning of implementation due to the perceived additional time and costs associated with implementing study tasks. Due to staff shortages and COVID-19-related restrictions on the number of in-person attendees per training, we trained 113 facility staff and CBVs instead of the expected 120 staff. We sensitized 870 Neighborhood Health Committee (NHC) members in three districts immediately after the trainings. In Chipata District, sensitization meetings were suspended due to the escalating COVID-19 cases at the beginning of the Delta wave and the enforcement of COVID-19 restrictions but resumed almost three months after the HCW and CBV trainings. Meetings were held with 300 NHC members in Chipata district.
Table 1
Study Outputs and Outcomes
Activity
|
District
|
Output
|
Planned
|
Achieved
|
Study site selection
|
Chadiza
|
10 public health care facilities
|
10 public health care facilities
|
Chipata
|
10 public health care facilities
|
10 public health care facilities
|
Chongwe
|
10 public health care facilities
|
10 public health care facilities
|
Lusaka
|
9 public health care facilities
1 private health care facility
|
9 public health care facilities
|
Stakeholder engagement
|
Chadiza, Chipata, Chongwe, and Lusaka
|
2 Provincial Health Directors
4 District Health Directors
4 MNCH Coordinators
4 Laboratory Scientists
|
2 Provincial Health Directors
4 District Health Directors
4 MNCH Coordinators
4 Laboratory Scientists
|
Study trainings
|
Chadiza,
|
156 Midwives, Phlebotomy staff and CBVs
|
28 staff†
|
Chipata
|
32 staff
|
Chongwe
|
27 staff
|
Lusaka
|
26 staff
|
Community sensitization meetings
|
Chadiza, Chipata, Chongwe and Lusaka
|
40 Community sensitization meetings
30 NHC members
|
39 Community sensitization meetings
30 NHC members
|
Participant recruitments
|
Chadiza
|
2,600 participants
|
1,755 participants§
|
Chipata
|
2,600 participants
|
2,396 participants
|
Chongwe
|
2,600 participants
|
2,527 participants
|
Lusaka
|
2,600 participants
|
2,332 participants
|
Samples tested
|
Chadiza
|
2,600 samples
|
§ 1,755 samples
|
Chipata
|
2,600 samples
|
2,161 samples
|
Chongwe
|
2,600 samples
|
2,455 samples
|
Lusaka
|
2,600 samples
|
2,187 samples
|
†Another 30 staff were retrained after incorporation of the ANC COVID-surveillance protocol into the existing malaria surveillance pilot. §Participant recruitment ended in July 2022, MNCH: maternal, newborn, and child health; CBV: community-based volunteer; NHC: neighborhood health committee
|
Between September 2021 and September 2022, the study collected samples from 9,010 recruited participants across 39 health facilities. Chipata, Chongwe, and Lusaka districts recruited 7,255 participants in the 13 months of study implementation; Chadiza District recruited 1,755 participants from October 2021 to July 2022. Five percent of the participants did not have test results as the samples were misplaced. While only 179 (2%) participants reported that they had confirmed COVID-19 before enrollment, seroprevalence during the last month of data collection was 57% in Chadiza District (July 2022), 95% in Chipata and Chongwe districts, and 100% in Lusaka District (8).
Study Experiences
Delayed Implementation because of Delta Wave
Field activities were initiated in June 2021 when Zambia started experiencing a third wave of COVID-19 caused by the Delta variant, the most severe variant to affect Zambia (Table 2) (9). On 16th June, to contain the Delta wave, MOH announced strict COVID-19 mitigation measures which included prohibiting in-person workshops (10). The NHRA followed with a notice temporarily suspending COVID-19-related research activities which were deemed high risk for researchers and participants during this period of intense transmission (11). Trainings and community sensitization meetings in Chipata District were most impacted by these restrictions. Because of these delays, participant enrollment in all four districts did not begin until September 2021 (CIDRZ) and October 2021 (PATH).
Even after the Delta wave, COVID-19 continued to impact implementation, especially interactions between health facility staff and participants. MOH guidelines reinforced the use of face masks in public places and scaled up the availability of rapid SARS-CoV-2 test kits and COVID-19 vaccines. Additional personal protective equipment was procured for staff and participants and distributed immediately after each district-level training. In August 2021, the NHRA lifted the national research suspension when COVID-19 cases were consistently decreasing and authorized the study to recommence field activities after demonstrating adequate COVID-19 risk mitigation procedures.
Division of Study Districts between two Implementing Partners
This study’s protocol was modeled after an existing pilot of malaria surveillance in Chadiza District by the National Malaria Elimination Program and PATH’s Program for the Advancement of Malaria Outcomes with support from the U.S. President’s Malaria Initiative (12). During the preparatory stages, CIDRZ planned to implement study activities in all four selected districts. However, due to PATH’s existing presence in the Chadiza District, the study districts were divided between the two partners, with PATH expanding its existing malaria surveillance study to include SARS-CoV-2. This approach facilitated implementation in Chadiza, though it required revision of the electronic study questionnaire and retraining staff to undertake malaria and ANC COVID-19 surveillance simultaneously. This approach additionally required close coordination and collaboration between the two partners to ensure that processes were the same in all sites.
Use of Ministry of Health Staff for Study Implementation
The study leveraged existing resources in the MNCH departments at implementing facilities. This not only reduced implementation costs but eliminated the need to hire additional and enhanced acceptability of the study as nurses and CBVs were already known and trusted by the community, which was particularly important to counteract hesitancy around COVID-19 activities. To reinforce participation by MOH staff and CBVs, facility staff were given 30 Zambian Kwacha (approximately 1.85 USD) per participant enrolled (which is common for studies in Zambia).
Although facility staff were trained in research ethics, some had difficulties adhering to research requirements during implementation. Consequently, the study reported several protocol deviations such as the recruitment of a minor or the consenting of an illiterate participant, both in the absence of a witness. Repeated onsite orientations and supportive supervision visits were conducted by RAs to enhance research knowledge and fulfill the requirements of the protocol.
Professional staff who were trained to help with study implementation were selected from a limited pool of MOH employees and were still expected to perform their routine tasks and were entitled to permissible vacation time and study leave. To mitigate the impact of staff absences, CIDRZ conducted onsite orientations for additional staff who had not been trained at the beginning of the study. Additionally, the RAs provided continued support for participant recruitment when health facility staff were unavailable at short notice.
To resolve the identified issues and harmonize the capturing of identification numbers with PATH, CIDRZ reprogrammed the electronic questionnaire with logic checks and redesigned its labels to enable barcode scanning using the tablet. The barcodes were printed in triplicate and were to be placed on the DBS card, laboratory log form, and informed consent form (for CIDRZ) and malaria RDT form (for PATH). All previously generated study documentation (i.e., screening and recruitment logs, DBS cards, and laboratory forms) were relabeled with the newly designed barcodes. Further, participant records from the CIDRZ Laboratory Information Management System (LIMS) and ODK databases were matched. Identification numbers of any unmatched records were shared with the RAs to verify that a sample had been collected from each participant. After follow-up, the RA returned to the CIDRZ laboratory to correct the details on the DBS cards.
Sample Collection, Transportation, and Storage
At the beginning of the study, blood samples collected from the CIDRZ health facilities were transported using the national HIV viral load sample courier system from the health facilities to central hubs for storage before being transferred to the University Teaching Hospital (UTH) laboratory in Lusaka for testing. Despite being distinctively packaged and labeled at implementing facilities, study DBS cards were batched together with those used for early infant diagnosis of HIV at delivery to the UTH laboratory. This resulted in 693 missing samples and required samples to be sorted manually to be recovered (35%). The process of verifying the number of samples received from implementing facilities was slow and hampered by the unavailability of permanent study staff at the UTH laboratory. Samples collected from the Chadiza health facilities for the previous month were collected by the second month and transported to the NMEC laboratory by the PATH study team.
To prevent further loss of samples, in March 2022, RAs were tasked with documenting all collected samples in a study laboratory logbook, checking all DBS cards for accuracy and completeness of participant information before submitting them to the CIDRZ central server, and transporting samples to the CIDRZ laboratory for entry into LIMS before sending them to the UTH laboratory for testing. This lessened the possibility of samples missing and helped to ensure that data was accurate and complete but also required more staff to perform data entry into two separate systems and the purchase of additional barcode scanners.
Since implementation was done by different partners, study data were collected and stored on different servers. The Chadiza dataset was to be shared by PATH monthly with the CIDRZ study team for review. However, due to the remote location of some health facilities in Chadiza District and the resulting poor cellular network connection, there were often delays in uploading completed questionnaires to the PATH server and consequently, delayed submission to CIDRZ. The delays in dataset sharing impacted the cleaning, merging, query resolution, and reporting of the data.
Sample Testing
DBS samples were tested using the Tetracore FlexImmArray SARS-CoV-2 Human IgG Antibody Test and processed using Luminex MAGPIX machines. While multiple MAGPIX machines were available in Zambian laboratories, they were infrequently serviced and required maintenance before sample processing could begin. This issue was not identified until April 2022. In July 2022, laboratory engineers from South Africa traveled to Zambia to perform maintenance of the machines. Prior assessment of equipment and service records and earlier engagement of maintenance staff may have circumnavigated this complication and could potentially have avoided delays and issues with sample processing quality. Additional calibration and validation kits for retesting a proportion of the samples were purchased. These costs were not initially budgeted for and therefore had an additional cost implication on the budget.
Due to international travel restrictions, UTH and PATH laboratory technicians were initially oriented on the assay protocol virtually, but a lab scientist later traveled from CDC in Atlanta to Zambia to provide hands-on technical assistance. Because of these delays, although study implementation began in September 2021, laboratory results were only available starting in August 2022. This delay impeded the study objective of being able to monitor survey results in near real-time for new COVID-19 waves.
Delivery of Clinical and Laboratory Supplies Challenges
Because this study was integrated into routine health services in the health facilities, 20 DBS cards for study sample collection were availed through MOH district laboratories as each implementing facility ordered DBS cards for infant HIV testing. In Chadiza, DBS cards were procured by PATH and distributed directly to health facilities. In February 2022, facility staff in Lusaka District began reporting phased roll-out point of care (POC) testing for the early infant diagnosis (EID) of HIV which no longer relied on the use of DBS cards. As MOH scaled up POC testing, routine procurement of DBS cards was reduced so additional stock needed to be procured by the study to supplement the district supplies, which was an unanticipated cost to the study.
Myths and Misconceptions
This study was conducted during the 3rd and 4th waves (Delta and Omicron, respectively) of COVID-19 in Zambia. Consequently, there was still much stigma and mistrust around SARS-CoV-2 infection and COVID-19 vaccination, which had only become available in Zambia in April 2021. Vaccine awareness and education efforts were in their infancy and uptake was poor. Rumors around the vaccine causing infertility, vaccine effects on pregnancy, and even death were commonplace.
Added to this was wariness around COVID-19 testing as nasal swabbing was new to most and had not previously been used widely in Zambia for disease diagnosis. Although our study did not utilize nasal swabs, the use of DBS came with its challenges as some people were hesitant to allow their blood to be drawn. These myths and misconceptions were addressed during group talks and individually when consenting participants. Additionally, over time, the MOH embarked upon widespread vaccine campaigns and worked actively to correct the misinformation.
Lessons Learned
Integration into Pre-existing Surveillance Platforms
This protocol was adapted from, and integrated into, the ANC-based malaria surveillance study protocol that was already being implemented by PATH in Chadiza district. Integration required close communication with the district, clear trainings and job aids, careful monitoring of work burdens and leave schedules, and careful data collection. Engagement of the community also required a deliberate approach to avoid confusion and to ensure that the pre-existing ANC-based malaria surveillance study was not compromised in any way. Our study demonstrated that it is possible to integrate COVID-19 surveillance into an already established service delivery and study, in this case, the ANC-based routine malaria surveillance study. This approach could be more likely to be sustainable since it utilizes existing structures. A similar surveillance approach could be used for other diseases of public health importance.
Adequate Implementation Preparations
Implementers endeavored to identify and engage local partners and stakeholders both before and during implementation to encourage community ownership and acceptability of the study. Since surveillance systems are only as good as the data they produce, incorporating lessons from prior studies and ensuring stakeholder buy-in, collaboration, and support was critical for ensuring activities were acceptable, sustainable, and successful. At the initial proposal writing stage, the identification of stakeholders was meticulous to ensure that a comprehensive list was developed. Lastly, assuring current laboratory testing equipment maintenance as part of the pre-study processes helped to avoid unexpected delays in producing results.
Ensure Data Completeness and Integrity from the Beginning
Before study implementation, all available records and data collection tools were reviewed and piloted to assess usability and allow for adaptation and identification of potential challenges. Simple and clear record-keeping and probing study data early and frequently helped to ensure data queries were addressed immediately. Performing continuous quality assurance checks for limited issues after study completion.
Using tablets for data collection with strict data checks improved the completeness of data but added to costs and required additional training of staff, maintenance of tablets, and adequate security. Hard copies of questionnaires were needed as a backup in the event of malfunctioning tablets, highlighting that for future studies, preparing for device failures and other technological failures is important to consider at the planning stage.
Buffer Budgeting and Absorbing Unanticipated Expenditures
Despite careful planning, the study still ran into unanticipated challenges, and additional funds were required to compensate for delays in enrolling (e.g., extending staff contracts and incentives), additional buffer stock of DBS cards, lancets, and unanticipated laboratory supplies to process the samples. Providing nominal rewards or incentives added to costs but also encouraged engagement and participation and fostered accountability for deliverables. Maintaining a budget buffer that was rapidly accessible was key to maintaining study momentum and avoiding unnecessary delays and interruptions.
Using CBVs to Ease the Burden on Health Facility Staff
MOH staff are burdened with a high patient-nurse ratio, particularly in rural areas. Recruiting CBVs to assist with some of the study procedures such as conducting health education, screening and selecting potential study participants, obtaining consent, and administering the questionnaire reduced the additional workload of the study on the professional staff.
Strengths and Limitations of this Study
A major strength of our study is that we successfully integrated SARS-CoV-2 surveillance into routine antenatal care visits making the system more sustainable and reducing the costs and resources required. Monitoring implementation through multiple in-person and virtual platforms facilitated early detection and prompt resolution of challenges. An additional strength was the fact that study implementation took place during the 3rd and 4th waves of COVID-19 in Zambia, and mitigation strategies put in place can serve as a model should additional COVID-19 waves hit Zambia or different disease outbreaks occur. We were also unable to assess the potential for early outbreak detection due to delays in laboratory testing. While retrospective analysis still adds value to understanding COVID-19 epidemic dynamics, our results could not be used for real-time decision-making. Lastly, even though pregnant women have been used as a healthy general population proxy for other diseases (e.g., HIV), it is not known if they are representative of the general population for SARS-CoV-2.