Kenya’s experience implementing event-based surveillance during the COVID-19 pandemic

Abstract

Event-based surveillance (EBS) can be implemented in most settings for the detection of potential health threats by recognition and immediate reporting of predefined signals. Such a system complements existing case-based and sentinel surveillance systems. With the emergence of the COVID-19 pandemic in early 2020, the Kenya Ministry of Health (MOH) modified and expanded an EBS system in both community and health facility settings for the reporting of COVID-19-related signals. Using an electronic reporting tool, m-Dharura, MOH recorded 8790 signals reported, with 3002 (34.2%) verified as events, across both community and health facility sites from March 2020 to June 2021. A subsequent evaluation found that the EBS system was flexible enough to incorporate the addition of COVID-19-related signals during a pandemic and maintain high rates of reporting from participants. Inadequate resources for follow-up investigations to reported events, lack of supportive supervision for some community health volunteers and lack of data system interoperability were identified as challenges to be addressed as the EBS system in Kenya continues to expand to additional jurisdictions.

Summary box

• While it was known that event-based surveillance (EBS) can be useful as a complementary early warning system to traditional surveillance systems, it was not clear whether EBS would work during the COVID-19 pandemic and how it could contribute to epidemic intelligence.

• This manuscript shows how an existing EBS structure in two counties in Kenya allowed for rapid adaptation to include detection of COVID-19 cases during the pandemic.

• This approach can serve as a model for other countries, where EBS platform can serve as an early warning system prepandemic and readily adapted to detect clusters during an ongoing pandemic.

Introduction

Event-based surveillance (EBS) is defined as ‘the organised collection, monitoring, assessment and interpretation of mainly unstructured ad hoc information regarding public health events or risks’.^{1 2} This is accomplished through recognition and immediate reporting of signals of events that may represent acute threats to public health and can include reporting from multiple sources and levels, such as the mass media (media scanning EBS (MEBS)), health facilities (health facility EBS (HEBS)), hotlines (phone EBS) and community health volunteers (CHVs) and community members (community EBS (CEBS)). Leveraging these structures, EBS can be used in most settings to detect all health hazards, including respiratory illnesses, which can be challenging to investigate in many settings.³ An EBS system compliments the disease-specific indicator-based surveillance (IBS), and sentinel surveillance systems (SSS) by enabling early detection and immediate reporting of outbreaks and other public health threats. Thus, EBS can be useful as an early warning system to detect a public health threat early and can also be rapidly leveraged for detection of outbreaks and changes in virus behaviours in an ongoing pandemic. It also improves the detection rates of public health threats due to the multiplicity of its sources of information.

In September 2019, the Kenya Ministry of Health’s (MOH) Division of Disease Surveillance and Response began implementation of a CEBS programme in select jurisdictions within two counties (Nakuru and Siaya) in Kenya. The first case of COVID-19 in the country was confirmed by the MOH on 13 March 2020, in a traveller returning to Nairobi.⁴ As the pandemic progressed in Kenya, the MOH made the decision to initiate HEBS alongside the CEBS system and added COVID-19-related signals to the EBS programme.

In this manuscript, we describe Kenya’s CEBS infrastructure prior to the COVID-19 pandemic, modifications made to include COVID-19 reporting, and highlight the experience of expanding CEBS and piloting HEBS during a pandemic.

Community event-based surveillance structures

In September 2019, with support from the US Centers for Disease Control and Prevention (CDC), Henry Jackson Foundation (HJF), Kenya Medical Research Institute (KEMRI), Medic Mobile and Washington State University, the Kenya MOH implemented a pilot of CEBS in 37 community units in Siaya and Nakuru counties, 2 out of Kenya’s 47 total counties.⁵ These counties represented approximately 1.6% of the total country population.⁶

The CEBS pilot was based on existing community health services, such as the Ministry of Health’s community health strategy and the Kenya Ministry of Agriculture, Livestock, Fisheries and Cooperatives’ (MOALFC) community animal health systems, as well as Integrated Disease Surveillance and Response (IDSR) implementation strategies. The IDSR reporting structure in Kenya includes CHVs, who provide health promotion and engage in disease surveillance through existing community networks.⁷ As a general rule, 20 households are assigned to 1 CHV and 50 CHVs constitute a community health unit (CHU), which is functionally linked to a health facility. The CHVs in each CHU are supervised by a community health assistant (CHA) who are healthcare workers within the MOH.

The CEBS pilot also embraced the One Health concept with the involvement of human and animal (including wildlife where present) health sectors. At the CHU level, community animal disease reporters (CDRs) were included alongside the CHVs with the animal health assistants (AHAs) cosupervising this level one work force. Similarly, the subcounty veterinary officer (SCVO), county director of veterinary services and the national director of veterinary services from the veterinary department shared CEBS roles with the subcounty disease surveillance coordinator (SCDSC), county director of public health and national director medical services from the MOH.

Community networks are social, cultural and economic groups of people that exist and operate within a community. The networks include, but are not limited to, family, friends, gender-based and age-based groups, farmers, traditional healers, local administrators, religious leaders, learning institutions and community-based organisations, among others, who reside in the community and regularly interact with residents and farmers hence form networks of key informants. The community networks are the ‘eyes’ and ‘ears’ on the ground greatly increasing chances of signal detection in the community.

These networks are sensitised on the community signals whereupon observing, witnessing or being informed, they report to their respective CHV/CDR in person or through any other available means. The CHVs and CDRs are responsible for sensitising the community networks on signals to facilitate detection. The CHVs/CDRs can use household visits, barazas, dialogue days, etc to sensitise the networks. Traditional communication channels, such as banners, fliers, posters, etc, can also be used to aid detection of signals.

CHVs and CDRs are expected to detect and report on priority signals/syndromes to the CHAs and AHAs, respectively. CHAs and AHAs in turn report to the subcounty health and veterinary departments, respectively. A subcounty is overseen by the SCDSC, and veterinary officer (oversees all animal-related occurrences) who review the reports and provide appropriate response to occurrences that may represent public health threats.

The existing IDSR reporting structure was adapted for CEBS (online supplemental figure 1): in addition to disease-specific case definitions, CHVs and CDRs were trained to detect patterns that may predict emerging and re-emerging events (ie, signals) and report to the CHAs or AHAs. The CHAs and AHAs are trained to receive, triage, and verify the signals and report to the SCDSC/SCVO. The SCDSC then conducts a risk assessment (RA) to inform appropriate response, such as further investigation, deploy a rapid response team or escalate to the county level (table 1).

Table 1.

Predefined signals used for community event-based surveillance (CEBS) and hospital event-based surveillance (HEBS) in Kenya during pilot implementation and COVID-19 expansion

Signals for CEBS pilot (Implemented September 2019)	Additional signals for COVID-19 expansion (Implemented June 2020)
1. Two or more people presenting with similar signs and symptoms in a community (village, estate, school, other institution, community gathering, eg, funeral, wedding, market) within a week 2. Any death in the community due to unknown cause 3. Any child less than 15 years with a sudden onset of weakness of the limb/s not due to injury 4. Any person 5 years of age or more with lots of watery diarrhoea on the same day 5. Increased sickness including abortions and/or deaths of animals (wild or domestic and poultry/birds or fish) 6. Any person who has been bitten by a sick looking animal including dogs 7. Any event that causes public health anxiety/concern	8. Any recent local (cross-county border) or international (cross-national border) traveller with or without symptoms of respiratory illness (hotness of the body, cough or difficulty in breathing) 9. Any person who has had recent close contact (work, transport, residence, visit, etc) with a person confirmed or suspected to be having COVID-19 with or without symptoms of respiratory illness (hotness of the body, cough or difficulty in breathing) 10. Two or more cases with a respiratory illness (hotness of the body, cough or difficulty in breathing) from the same locality/village or social group (family, school, workplace, function, estate, etc) within 1 week.
Signals for HEBS pilot∗ (implemented August 2020)
H1. Two or more cases with similar symptoms within a week from the same location that required hospitalisation H2. Any healthcare worker with severe illness during/after caring for a patient with similar symptoms. H3. Unexpectedly large increase of cases with the same symptoms H4. Any hospitalised case with unexplained/unusual clinical manifestation of a known disease H5. Any infectious case that fails known therapy	H6. Any recent local (cross-county border) or international (cross-national border) traveller with or without symptoms of respiratory illness (hotness of the body, cough or difficulty in breathing) H7. Any person who has had recent close contact (work, transport, residence, visit, etc) with a person confirmed or suspected to be having COVID-19 with or without symptoms of respiratory illness (hotness of the body, cough or difficulty in breathing) H8. Two or more cases with acute respiratory illness (difficulty in breathing, cough or fever (≥38.0°C)) from the same community or social grouping (family, institutions (workplace, schools or faith-based organisation), functions (funeral, wedding or market), village/estate, etc) within a week

*Signals H6, H7 and H8 for HEBS were designed to detect events related to COVID-19.

Electronic reporting system: m-Dharura

At the time of CEBS implementation, CHVs were using paper-based forms to report cases. The MOH implemented an electronic reporting system called m-Dharura to streamline the reporting process throughout the public health structures (figure 1). m-Dharura is a mobile phone-based short message services (SMS) application with a web-based interface. The m-Dharura system is part of the community health toolkit, which provides open-source resources to design, build and deploy digital tools for community health targeting people delivering care in hard-to-reach areas and works with or without Internet connectivity.^{8 9}

Figure 1.

The CEBS pilot reporting system and workflow in m-Dharura. AHA, animal health assistant; CEBS, community event-based surveillance; CDR, community animal disease reporter; CHV, community health volunteer; SCDSC, subcounty disease surveillance coordinator; SCMOH, subcounty Ministry of Health; SCVO, subcounty veterinary officer; SMS, short message service.

On detection of signals, the community networks report them to the CHVs who then log them into the m-Dharura system by submitting an SMS with the signal code corresponding to the signal being reported (table 1, figure 1). After successful submission of a signal, the CHAs are notified of a logged signal through SMS and can conduct signal triage by gathering more information from the reporting CHV to evaluate if the report met the threshold of any of the predefined signals; and has not been previously reported. Once triaged, the CHA verifies the public health-focused signals and AHA the veterinary focused signals by contacting the signal reporter to determine authenticity. Any signal triaged and verified as true becomes an event and is reported to the SCDSC by the CHA/AHA through the m-Dharura platform. Once the SCDSC receives a notification, they initiate the process of RA using a risk categorisation matrix in the national EBS guidelines to determine the level of risk presented to public health by the event and submit a report to the subcounty health authorities for response. Depending on the resource capacity, expertise and the event’s level of risk, the response could be at the subcounty level or escalated to the county or national levels.

The m-Dharura reporting system also populates a web-based dashboard through which the subcounty, county and national level EBS track performance using the EBS indicators, view basic analytics on EBS accuracy and timeliness of reports, and a signal log with downloadable data to facilitate additional analysis.

CEBS enhancement to include COVID-19 reporting

On 8 June 2020, in response to the COVID-19 pandemic, the existing CEBS system was leveraged on to detect signals of potential COVID-19 cases or clusters in communities (online supplemental figure 2). During this period, the country was experiencing importation of COVID cases and sporadic and isolated outbreaks with hotspots in the two largest cities in Kenya, Nairobi and Mombasa. The CEBS system with the new COVID-19-related signals was expanded from pilot sites in Siaya and Nakuru Counties to include all subcounties in Siaya (6) and Nakuru (11), and 2 subcounties in Marsabit county increasing community units covered under CEBS from 37 to 617 (online supplemental figure 3).

At the time, community transmission had not been established outside of Nairobi and Mombasa counties, and the signals were designed to also detect travel-related cases using criteria such as recent local or international travel, in addition to close contact with persons suspected or confirmed to have COVID-19, and clustering of two or more cases of respiratory illness within 1 week (table 1).

After the inclusion of COVID-19-related signals, training of staff on the revised signals and adding new subcounties in June 2020, a total of 2343 personnel in Siaya, 3100 in Nakuru and 305 in Marsabit counties were trained on the CEBS system (online supplemental table 1).

HEBS strengthening in response to COVID-19

Following widespread transmission of SARS-CoV-2 by August 2020 and the need to enhance COVID-19 surveillance in the healthcare facilities, the MOH piloted HEBS in a total of 10 health facilities (3 each in Nakuru and Meru counties and 4 health facilities in Mombasa County) (online supplemental figures 2 and 3). All the health facilities were selected from the same subcounty in the respective counties, taking into consideration those with high workload, levels of service delivery, ownership and active performance in weekly routine surveillance reporting.

Within the existing IDSR reporting structure, healthcare workers (HCWs) in these facilities report notifiable diseases to their health facility surveillance focal persons, who then report to SCDSC. This existing structure was enhanced for HEBS: the HCW detected signals through interaction with patients, caretakers, or family members, and gathered relevant demographic and clinical information of the presenting condition through a chart review (online supplemental figure 1). If the information extracted matched any of the predefined signals, the HCW reported through m-Dharura to the health facility surveillance focal person, who then undertook triage and verification and submitted an event report to the SCDSC for RA. HEBS reporting was conducted through the phone-based m-Dharura application.

Eight HEBS signals were developed by the MOH to capture: (1) clusters of diseases, (2) hospital acquired infections among healthcare workers, (3) changes in trends for endemic conditions, (4) changes in disease agents and (5) antimicrobial resistance, as well as three signals to detect COVID-19-related events (table 1). The public health programme officers and surveillance supervisors at the county and subcounty levels, along with healthcare workers, were trained on these signals through in-person trainings. Posters were also developed and put up as job aids in various departments of the participating health facilities.

A total of 288 HCWs were trained on HEBS in Nakuru, Mombasa and Meru Counties across the reporting cascade (online supplemental table 2).

Findings from Kenya’s community and HEBS systems

In May 2021, an evaluation of Kenya’s EBS system was initiated and included qualitative analyses through field visits conducted in Nakuru county, a pilot site for both CEBS and HEBS implementation. The field visits included key informant interviews (KIIs) with CEBS and HEBS focal points and focus group discussions (FGDs) with CHVs and CDRs to assess the performance and acceptability of the EBS system. In addition, data collected from CEBS and HEBS sites between June 2020 (when COVID-19-related signals were implemented) and June 2021 (when field visits were conducted) were extracted from m-Dharura and analysed.

From June 2020 to June 2021, 8734 signals were reported through CEBS, with 2961 (33.9%) verified as events. The most frequently reported signals included signal 8 (recent traveller with or without respiratory illness, 28.1%), signal 7 (any event that causes public health anxiety, 23.7%), signal 5 (increased sickness and/or deaths of animals, 15.7%) and signal 2 (any death due to unknown causes, 13.3%) accounting for 80.7% of reported signals and 77.5% of events (table 2). On the other hand, of the events, 1026 (34.7%) underwent RA, of which 866 (84.4%) were responded to. Of the events, 863 (29.1%) were confirmed to be COVID-19-related events.

Table 2.

Number of reported event-based surveillance (EBS) signals and events in Kenya, June 2020–June 2021

EBS signals	Total no (%) of reported signals	Total no (%) of events	Event to signal ratio
CEBS signals (June 2020–June 2021)	n=8734	n=2961	0.34
1. Two or more people sick with similar signs and symptoms in a community within a week	806 (9.2)	341 (11.5)	0.42
2. Any death in the community due to unknown causes	1162 (13.3)	358 (12.1)	0.31
3. Any child less than 15 years with a sudden onset of weakness of the limb/s not due to injury	128 (1.5)	38 (1.3)	0.30
4. Any person 5 years of age or more with lots of watery diarrhoea and dehydration in the same day	163 (1.9)	73 (2.5)	0.45
5. Increased sickness including abortions and/or deaths of animals	1370 (15.7)	566 (19.1)	0.41
6. Any person who has been bitten by a sick-looking animal	371 (4.2)	132 (4.5)	0.36
7. Any event that causes public health anxiety/concern	2066 (23.7)	590 (19.9)	0.29
8. Any recent local or international traveller with or without symptoms of respiratory illness	2451 (28.1)	782 (26.4)	0.32
9. Any person who has had recent close contact with a person confirmed or suspected to have COVID-19	141 (1.6)	48 (1.6)	0.34
10. Two or more cases with a respiratory illness from the same locality/village or social group within 1 week	76 (0.9)	33 (1.1)	0.43
HEBS signals (August 2020–June 2021)	n=56	n=41	0.73
1. Two or more cases with similar symptoms within a week from the same location that required hospitalisation	4 (7.1)	2 (4.9)	0.50
2. Any healthcare worker with severe illness during/after caring for a patient with similar symptoms	3 (5.4)	2 (4.9)	0.67
3. Unexpectedly large increase of cases with the same symptoms	5 (8.9)	3 (7.3)	0.60
4. Any hospitalised case with unexplained/unusual clinical manifestation of a known disease	4 (7.1)	3 (7.3)	0.75
5. Any infectious case that fails known therapy	6 (10.7)	4 (9.8)	0.67
6. Any recent local or international traveller with or without symptoms of respiratory illness	22 (39.3)	20 (48.8)	0.91
7. Any person who has had recent close contact with a person confirmed or suspected to have COVID-19	11 (19.6)	6 (14.6)	0.55
8. Two or more cases with a respiratory illness from the same locality/village or social group within 1 week	1 (1.8)	1 (2.4)	1.00

CEBS, community event-based surveillance.

The HEBS system generated 56 signal reports including 32 (57.1%) from Nakuru county, 19 (33.9%) from Meru County and 5 (8.9%) from Mombasa County. A total of 41 (73.2%) signals were verified of which 30 (73.2%) underwent RA. Of those events that underwent RA, 23 (76.6%) were responded to. Thirty-four (60.7%) of the HEBS reports were for COVID-19-related signals, of which 22 (64.7%) were travel related (H6) and 11 (32.4%) COVID-19-related signals were associated with close contact. In addition to the COVID-19-related signals, 6 (10.7%) reports were for signal 5 (infectious case that fails known therapy) and 5 (8.9%) reports of signal 3 (large increase in cases with same symptoms) (table 2).

Results from both CEBS and HEBS show that most COVID-19-related signals were detected from local or international travellers early in the pandemic when there was low community transmission. As the pandemic progressed, reports from this signal decreased over time (figure 2). The waning utility of this signal following widespread community transmission of SARS-CoV-2 in Kenya in later months emphasises the need for continuous refinement of CEBS signals.

Figure 2.

Number of CEBS signal reported in Kenya, September 2019–June 2021. CEBS, community event-based surveillance.

The signal ‘any death in the community due to unknown cause’ did not yield significant reports; however, this system can be a platform to strengthen mortality surveillance (crude, specific, all-cause, etc). Although the MOH, Kenya included community reporting of ‘any child less than 15 years with a sudden onset of weakness of the limb/s not due to injury’, not many reports were received. It remains to be seen in the upcoming months if community surveillance for Acute Flaccid Paralysis (AFP) may add value to this programme.

Both CEBS and HEBS continued to be used to report non-COVID-19-related events (2098 CEBS events and 14 HEBS events) throughout the pandemic, including animal health events, unexpected deaths of unknown etiology and other infectious and non-infectious causes (figure 2). The CEBS system resulted in consistent notification of COVID-19-related signals even during ongoing healthcare worker strikes, highlighting the crucial contribution of community structures in early detection. Operating an EBS system during a pandemic may provide a means for public health agencies to continue to monitor non-pandemic disease threats.

Qualitative data from KIIs and FGDs found that the CHVs and CHAs understood the expanded signals and had few challenges navigating m-Dharura application. However, the KIIs and FGDs also highlighted several challenges, including lack of resources for follow-up investigations and insufficient supervision of CHVs. Lack of funds for CHAs to travel to the field to investigate reported signals led to delays in signal triage and verification and/or lack of investigation. In addition to missing potential events of concern, the lack of follow-up on potential events negatively impacted CHV and HCW engagement and motivation for reporting. Similarly, lack of capacity at the SCDSC level also impeded timely action and response for some verified events and led to inadequate monitoring and supervision of CHVs in some instances. This lack of supervision was an additional contributor to the decrease in motivation at the CHV and CHA level.

Data extracted from m-Dharura corroborated the findings of the evaluation and showed an uptake in reporting during the initial months following training and implementation (July–September 2020), and a steep decline in subsequent months (figure 2). The MOH, in collaboration with CDC, is exploring ways to mobilise resources to facilitate the SCDSCs to support prompt verification, investigation and response activities. Processes to support SCDSCs for ongoing monitoring of the system and providing routine supervision of the CHAs and CHVs are being revised and implemented.

During pilot implementation of HEBS, only a small number of HCWs were sensitised and trained on HEBS, 288 (13.1%) of 2193 HCWs across the participating health facilities. To accommodate staff schedules, training for HEBS was conducted using an existing continuing medical education (CME) model, where a subset of HCWs were trained on a weekly basis and ultimately the entire workforce would be trained over time. However, the CME sessions also included other competing training topics, which limited the number of HEBS sessions and total number of staff engaged. During the earlier months of the pandemic, trainings were also halted to limit large close-quarter gatherings. Competing priorities for HCWs coupled with fragmented trainings led to a low number of HCWs trained and limited reporting from the health facilities. To address this challenge, the MOH is developing new strategies that include working with health facilities to prioritise staff roles critical to HEBS signal reporting to ensure delivery of training first to those roles and additional HEBS training sessions added to the CME schedule to ensure regular training opportunities. Implementation of training outside the CME model is also being considered, structured in a tiered schedule to allow for different roles and different shifts to take part in HEBS training with minimal disruption to health facility operations. Finally, the MOH is also exploring routine support supervision and mentorship as additional ways of ensuring HEBS implementation is strengthened.

KIIs and FGDs reported that frequent updates to the m-Dharura application resulted in system downtimes and led to delays in receiving signals throughout the reporting cascade. To address challenges with the m-Dharura application, the MOH has established regular meetings to revise signals and materials and coordinate these changes with application upgrades at a regular interval to limit system downtime. Currently, the data being reported through m-Dharura populates an interactive dashboard which allows users to view signals stratified over time, however, it does not link to other surveillance or data systems. The MOH is developing systems to link this dashboard to the national and county emergency operations centres (EOCs), which would allow reported events to inform ongoing response work. Additionally, while the EBS dashboard does not currently connect with Kenya’s Health Information System (KHIS Aggregate)—a web-based health data management platform—the technical requirements for developing links between the two systems are being defined to enable better tracking of individual cases that are part of reported events.¹⁰

Lack of EBS data system linkage with KHIS also posed challenges to connecting laboratory results back to reported events, making it difficult to fully assess the impact of EBS reporting and subsequent investigations. This is largely because the case investigation forms used for sample collection during an event response were not designed to incorporate unique identifiers that would enable linkage between the systems. The MOH, with support from CDC, is planning to incorporate a unique identifier on forms and samples for all cases that are being tested related to a singular reported event, which would link the laboratory results with the EBS system.

Using the lessons learnt and recommendations from the pilot and evaluation, EBS processes and systems are being strengthened, and the system will be expanded to up to three additional counties throughout Kenya in 2022. In preparation for this expansion, the m-Dharura application underwent upgrades through a weeklong hackathon and technical peer-review process to address usability and downtime issues. The EBS signals have undergone several cycles of revisions from internal and external partners to ensure clarity and utility of each individual signal. These signals will be incorporated into revised training and resource materials, as well as the m-Dharura application. Resensitisations and trainings, for existing counties with EBS will be conducted, as well as trainings for the new expansion counties. All community units who participated in the pilot, including those that joined during the expansion, are expected to continue to engage in the development and consolidation of EBS in Kenya.

It is crucial to highlight that designing a system with an appropriate combination of sensitivity and specificity is one of the key challenges of EBS.³ This requires the inclusion of a sufficiently diverse and sensitive set of signals to meet national priorities without relying on disease-specific strategies. For that reason, continuous evaluation and refinement of signals becomes essential.

Finally, although the animal health sector was not the primary focus of the evaluation, it is important to note that the MOALFC also has a CDR system monitored by the AHA/SCVO; however, due to human resource limitations, this system may not be as extensive as community health. The MOALFC may have no animal healthcare centres but provide health extension services largely through licensed private practitioners. Despite these challenges, approximately one-fifth of all notified events were related to animal health.

Conclusion

An existing EBS structure allowed for rapid adaptation to include detection of COVID-19 cases during the pandemic. This approach can serve as a model for other countries, where an EBS platform can serve as an early warning system prepandemic or for routine detection of local outbreaks¹¹ and the same platform can be readily adapted to detect clusters and hotspots during an ongoing pandemic. This model could contribute to the development of a more effective system for global health security.¹²

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.

Ethics statements

Patient consent for publication

Not applicable.