ArticlePDF Available

Electronic Data Management for Vaccine Trials in Low Resource Settings: Upgrades, Scalability, and Impact of ODK


Abstract and Figures

Background: ODK provides software and standards that are popular solutions for off-grid electronic data collection and has substantial code overlap and interoperability with a number of related software products including CommCare, Enketo, Ona, SurveyCTO, and KoBoToolbox. These tools provide open-source options for off-grid use in public health data collection, management, analysis, and reporting. During the 2018–2020 Ebola epidemic in the North Kivu and Ituri regions of Democratic Republic of Congo, we used these tools to support the DRC Ministère de la Santé RDC and World Health Organization in their efforts to administer an experimental vaccine (VSV-Zebov-GP) as part of their strategy to control the transmission of infection. Method: New functions were developed to facilitate the use of ODK, Enketo and R in large scale data collection, aggregation, monitoring, and near-real-time analysis during clinical research in health emergencies. We present enhancements to ODK that include a built-in audit-trail, a framework and companion app for biometric registration of ISO/IEC 19794-2 fingerprint templates, enhanced performance features, better scalability for studies featuring millions of data form submissions, increased options for parallelization of research projects, and pipelines for automated management and analysis of data. We also developed novel encryption protocols for enhanced web-form security in Enketo. Results: Against the backdrop of a complex and challenging epidemic response, our enhanced platform of open tools was used to collect and manage data from more than 280,000 eligible study participants who received VSV-Zebov-GP under informed consent. These data were used to determine whether the VSV-Zebov-GP was safe and effective and to guide daily field operations. Conclusions: We present open-source developments that make electronic data management during clinical research and health emergencies more viable and robust. These developments will also enhance and expand the functionality of a diverse range of data collection platforms that are based on the ODK software and standards.
Content may be subject to copyright.
published: 04 November 2021
doi: 10.3389/fpubh.2021.665584
Frontiers in Public Health | 1November 2021 | Volume 9 | Article 665584
Edited by:
Carolina Varela Rodríguez,
Hospital Universitario 12 de
Octubre, Spain
Reviewed by:
Helene Langet,
Swiss Tropical and Public Health
Institute (Swiss TPH), Switzerland
Cesar Llorente,
Vall d’Hebron University
Hospital, Spain
Chrissy H. Roberts
Specialty section:
This article was submitted to
Digital Public Health,
a section of the journal
Frontiers in Public Health
Received: 11 May 2021
Accepted: 30 September 2021
Published: 04 November 2021
Marks M, Lal S, Brindle H, Gsell P-S,
MacGregor M, Stott C, van de
Rijdt M, Almazor GG, Golia S,
Watson C, Diallo A, Toure A,
Houlihan C, Keating P, Martin H,
Restrepo A-MH, Anokwa Y and
Roberts CH (2021) Electronic Data
Management for Vaccine Trials in Low
Resource Settings: Upgrades,
Scalability, and Impact of ODK.
Front. Public Health 9:665584.
doi: 10.3389/fpubh.2021.665584
Electronic Data Management for
Vaccine Trials in Low Resource
Settings: Upgrades, Scalability, and
Impact of ODK
Michael Marks 1, Sham Lal 1, Hannah Brindle 1, Pierre-Stéphane Gsell 2,
Matthew MacGregor 1, Callum Stott 3, Martijn van de Rijdt 4, Guillermo Gutiérrez Almazor 3,
Suman Golia 1, Conall Watson 1, Abdourahamane Diallo 2, Alhassane Toure 2,
Catherine Houlihan 1, Patrick Keating 5,6 , Hélène Martin 3, Ana-Maria Henao Restrepo 2,
Yaw Anokwa 4and Chrissy H. Roberts 1
1London School of Hygiene & Tropical Medicine, London, United Kingdom, 2World Health Organization, Geneva,
Switzerland, 3Nafundi LLC, San Diego, CA, United States, 4Enketo LLC, Denver, CO, United States, 5Médecins Sans
Frontières UK, London, United Kingdom, 6UK Public Health Rapid Support Team, London School of Hygiene & Tropical
Medicine, London, United Kingdom
Background: ODK provides software and standards that are popular solutions for
off-grid electronic data collection and has substantial code overlap and interoperability
with a number of related software products including CommCare, Enketo, Ona,
SurveyCTO, and KoBoToolbox. These tools provide open-source options for off-grid
use in public health data collection, management, analysis, and reporting. During the
2018–2020 Ebola epidemic in the North Kivu and Ituri regions of Democratic Republic of
Congo, we used these tools to support the DRC Ministère de la Santé RDC and World
Health Organization in their efforts to administer an experimental vaccine (VSV-Zebov-GP)
as part of their strategy to control the transmission of infection.
Method: New functions were developed to facilitate the use of ODK, Enketo and Rin
large scale data collection, aggregation, monitoring, and near-real-time analysis during
clinical research in health emergencies. We present enhancements to ODK that include a
built-in audit-trail, a framework and companion app for biometric registration of ISO/IEC
19794-2 fingerprint templates, enhanced performance features, better scalability for
studies featuring millions of data form submissions, increased options for parallelization
of research projects, and pipelines for automated management and analysis of data. We
also developed novel encryption protocols for enhanced web-form security in Enketo.
Results: Against the backdrop of a complex and challenging epidemic response,
our enhanced platform of open tools was used to collect and manage data
from more than 280,000 eligible study participants who received VSV-Zebov-GP
under informed consent. These data were used to determine whether the
VSV-Zebov-GP was safe and effective and to guide daily field operations.
Marks et al. ODK Upgrades for Vaccine Trials
Conclusions: We present open-source developments that make electronic data
management during clinical research and health emergencies more viable and robust.
These developments will also enhance and expand the functionality of a diverse range
of data collection platforms that are based on the ODK software and standards.
Keywords: Ebola, open-data-kit, open source software, electronic data collection, low and middle income
ODK (1) provides open-source, community developed software
and standards that have found broad utility in public health
research (2,3), epidemiology (4), disease mapping (5), and
anthropological (6) studies in low-and-middle-income countries
(LMICs). It has also been used to positive effect in clinical
trials (7) as well as in disease surveillance (8) during outbreaks.
The core ODK tools cover both front-end and back-end user
tools. Front end tools are primarily facilitated by electronic
data collection (EDC) on Android devices through the ODK
Collect App and the open-source Enketo system which is a
major component of the ODK ecosystem and provides webform
based data entry tools. Back-end tools consist primarily of a
web accessible ODK Aggregate server for data aggregation and
ODK Briefcase which provides downstream data management
(3). ODK Briefcase has both a graphical user interface (GUI)
and command line interface (CLI), the latter of which makes
the off-server management of data automatable via command
line tools. The core ODK tools are some of the most widely
used EDC tools in the world, with over 400,000 active users
each month and more than one million app instals to date. They
also form the basis of the ODK ecosystem (9), which underpins
the function of many other electronic data tools including Ona
(10), KoBoToolbox (11), SurveyCTO (12), CommCare (13), and
Enketo (14). Application Program Interfaces (APIs) exist which
can make ODK systems communicate with other common EDC
tools including REDCap (15), DHIS2 (16), R(17), and others.
In the wake of the 2013–2016 Ebola outbreak in West Africa,
which affected around 28,000 people (18), the World Health
Organization established the “R&D Blueprint” (19), bringing
together stakeholders in research and development activities
surrounding epidemics and health emergencies. The primary
goal of the R&D Blueprint was to facilitate rapid deployment
and evaluation of vaccines, tests, and therapeutics that could
be used to control epidemics and emergencies. This R&D
Blueprint includes provisions to assess the safety and efficacy
of experimental vaccines under expanded access (also known as
“compassionate use”) programmes. Because of the experimental
Abbreviations: APIs, Application Program Interfaces; CRFs, Case Report Forms;
CSV, Comma-Separated-Values; CLI, Command Line Interface; COVID-19,
Coronavirus Disease 2019; DRC, Democratic Republic of Congo; EVD, Ebola
Virus Disease; EDC, Electronic Data Collection; EDK, Emergency and Epidemic
Data Kit; FLWs, Front-Line Workers; GUI, Graphical User Interface; HCWs,
Healthcare Workers; LMICs, Low and Middle Income Countries; RSA, Rivest–
Shamir–Adleman; URL, Uniform Resource Locator; VM, Virtual Machine.
nature of unlicensed products, any expanded access must be
assessed in the context of research.
The Democratic Republic of Congo (DRC) has experienced
11 documented outbreaks of Ebola virus (2022) with the
most recent three having occurred in May–July 2018 (Équateur
Province), from July 2018 to July 2020 (North Kivu/Ituri
Provinces) and from May to November 2020. The North Kivu
epidemic is the second largest Ebola outbreak on record, with
more than 3,296 cases and 2,196 deaths having been reported by
late Dec 2019 (22). Control efforts targeted against the infection
were complicated by a number of factors which included regional
conflict, high population density, community mistrust of the
response and limited infrastructure for healthcare provision and
communications in the affected areas.
During both the Équateur (2018) and North Kivu (2018–
2020) outbreaks, the Ministère de la Santé RDC, and World
Health Organization attempted to use VSV-Zebov-GP (23), a live
replicating candidate Ebola, to halt the epidemic. The vaccine
was deployed using a ring vaccination strategy (24) wherein
the contacts and “contacts-of-contacts” of Ebola cases were
traced and offered vaccination. Ring vaccination aims to halt
the transmission of infection by providing a ring or “belt” of
resistant individuals around cases of infection. The success of
such approaches is highly dependent on good contact tracing
and high coverage vaccination. In the context of the R&D that
must accompany expanded access programmes, all participants
in a ring vaccination study must be followed-up for some period
(here at 30 min, 3 and 21 days post-vaccination) to assess the
safety of the product. Any cases of infection amongst vaccines
must also be linked to the vaccination data for efficacy estimates
(Figure 1). In the face of such complexity, there is a significant
need to collect, manage, and analyse large amounts of data
during a study such as this; particularly when the number
of participants grows to hundreds of thousands and includes
special/vulnerable groups such as pregnant women, infants, and
those with immune suppression.
At the time we began work on the current study, ODK
Collect was able to perform asymmetric encryption on records,
providing very high levels of security because once encrypted,
no field operator or malicious actor in control of a device
could decrypt or tamper with the data. Whilst Enketo provided
highly desirable options for browser-based data entry, it was
unable to perform encryption on records at the start of this
project. Neither platform had capacity to perform audit actions
in order to monitor enumerator behaviours during data entry
and modification, whilst options for biometric registration
of study participants were limited to a sophisticated but
Frontiers in Public Health | 2November 2021 | Volume 9 | Article 665584
Marks et al. ODK Upgrades for Vaccine Trials
FIGURE 1 | Data capture points within the Ebola Vaccination implementation. Individual CRFs captured information at differing points within the programme. Unique
participant IDs were recorded at each point to allow downstream data-linkage.
subscription-based fingerprint registration system offered by the
not-for-profit SimPrints project (25).
When the Équateur outbreak was declared in May 2018, the
partners of this study set out to develop the LSHTM Emergency
and Epidemic Data Kit (EDK), a specialist implementation of
ODK and other tools which encompassed an EDC, aggregation,
analysis, and monitoring system that (1) was scalable to
potentially millions of data form submissions, (2) could work
off-grid, for instance during long periods without internet
connexion, (3) was amenable to automation, (4) could facilitate
near-real-time monitoring and data sharing, (5) was fully
open-source, (6) had the capacity to register ISO/IEC 19794-
2 fingerprints, (7) could be replicated in the case of further
outbreaks or international spread, and (8) could optionally
generate an audit-trail for monitoring enumerator behaviours. In
this paper we report on the technical approach taken to rapidly
implement these changes and build an appropriate software suite
to support the vaccine programme roll-out in response to Ebola.
Approach to Platform Development
We followed Agile principles of platform development,
particularly with respect to (a) favouring the development
of working software over comprehensive documentation, (b)
involving our end-users and stakeholders in all stages and
deferring negotiations over roles, responsibilities, contracture,
and funding, and (c) allowing teams to self-organise and
adapt strategies in response to change. In practise we used
tools that were familiar to non-experts including WhatsApp,
Slack and GitHub to build a real-time development hub that
allowed academics, clinicians, computer scientists, field-workers,
WHO project-leads, and ministry staff to communicate and
contribute in real time to the development of the platform
whilst working in several countries, multiple time-zones and
hostile environments. During the early implementation phase,
we operated a 24-h working pattern, rotating work between
staff in order to have a working platform in place within the
first 10 days of the outbreak and in time for the first vaccine
doses to reach the field. Software developments to the ODK
ecosystem were developed and integrated into the EDK system
as and when they became available, with workarounds in place
in the interim.
To ensure that all software developments became available
to the widest possible user-base, we have implemented as many
software changes as possible to the core ODK and Enketo
systems, which is to say that the system we present should be
considered a specialised deployment of tools which continue
to be freely available through the parent projects ODK and
Enketo. New features and standards added to ODK for the
EDK system were reviewed by the ODK Technical Advisory
Board ( and made
available for comment on the community forum (https:// The open availability of all the
current developments of ODK contrasts with the approach
taken by several beneficiaries of the ODK ecosystem including
SurveyCTO, SimPrints and Ona; all of which control access to
some components of their software.
Frontiers in Public Health | 3November 2021 | Volume 9 | Article 665584
Marks et al. ODK Upgrades for Vaccine Trials
FIGURE 2 | System design of the LSHTM Emergency and Epidemic Data Kit. Devices in the field are used to carry out data collection (enumeration) activities on
browsers or an Android App. Those using Android have the option to register ISO/IEC 19794-2 fingerprint templates as part of the data collection. Encrypted records
are submitted to one of many parallel web accessible ODK Aggregate server front-ends. All servers lead to a single PostgreSQL database and backup system. All
backups contain only encrypted, non-human-readable form data. An Ubuntu virtual machine is scheduled to perform archiving (data pull, decrypt and export
activities, including optionally analysis of fingerprint templates) and triggers analysis (data tidying, analysis, and report generation) using R,Rmarkdown and other
open-source analytics tools. Outputs are securely copied or emailed to a workstation for end-user interactions. Areas highlighted in red show where data are stored or
transferred in encrypted formats and are non-human readable.
Conceptual Framework
The conceptual framework for the EDK system was (1) an
extensible project-oriented ODK server system that could
provide parallel server environments for multiple research
studies, (2) software changes to the ODK ecosystem which
could facilitate efficient management of millions of submissions
in near-real-time, (3) strengthening the security of Enketo
webforms via form-level asymmetric encryption, (4) an open-
source biometrics framework for registration of ISO/IEC 19794-
2 fingerprints via low-cost consumer hardware, (5) Audit trail
features for ODK, and (6) Automation of data management
and analysis using ODK Briefcase CLI, R, Rstudio, RMarkdown
and FlexDashboards. A schematic of the EDK system design is
provided in Figure 2.
Project Oriented ODK Servers
In order to make the system extensible, for instance to make
it possible to use an institutional installation of ODK servers
for multiple research studies, we designed a server system
infrastructure that allowed for the simultaneous operation of
a number of parallel activities (projects). Each project was
provisioned with its own dedicated ODK Aggregate server front
end, unique URL and configurable user privileges. These project
specific front-ends allowed for aggregation of data from EDC
devices in the field, and for end-user level management of case
report forms (CRFs) and individual project data entities in
isolation from other projects, their CRFs and data. Behind the
front-end, data from the many parallel projects were stored in
a unified format on a single PostgreSQL database which was
placed behind an institutional firewall and which was regularly
backed up for data protection and recovery from failure. For each
project, a data analysis pipeline was created on a virtual machine
(VM) which was able to call data from the PostgreSQL database
and to perform automated analysis, monitoring and reporting
functions (Figure 2). Between-project and meta-analytics were
also possible through this design. The addition of new projects
required three steps including (1) the creation of a new ODK
Aggregate front end, (2) design and deployment of project
specific CRFs, and (3) development of an analytics pipeline to
match the needs of the project.
Software Performance Developments in
the ODK Ecosystem
ODK Briefcase is a desktop Java application which contains an
application programming interface (API) that bridges the gap
between study data on a server and the downstream analysis
pipeline. It can both download individual data submissions (pull
actions) from the ODK Aggregate server database and also
parse, aggregate and export data to various formats, systems, and
backups. In the simplest terms, ODK Briefcase converts the many
individual data files collected from the field into a single data set
that is ready for analysis.
During the early development of EDK, we reached bottlenecks
in the form of the time taken to perform pull and export
actions. As the number of data submissions increased, so the
time taken to process the data came to exceed the 24-h analysis
and reporting cycle of the field deployment. In order to make
it possible to handle millions of CRF submissions to the EDK
system without impacting significantly on time taken for pull
and export operations, we introduced two performance related
features to Briefcase, including “smart append” and “resume from
last” controls.
The “smart append” feature speeds up exports of large datasets
by remembering the full date and time of the last submission
included in the most recent export for each form. By contrast
Frontiers in Public Health | 4November 2021 | Volume 9 | Article 665584
Marks et al. ODK Upgrades for Vaccine Trials
to the historical approach (where all existing records were
exported), the smart append feature exports only submissions
which are new since the last export operation completed,
appending these to the exported data from previous sessions.
The “resume from last” feature has an analogous function
for the pull operations and speeds up downloads of submissions
from ODK Aggregate by keeping track of information
representing the last downloaded block of submissions and
thereby only requesting new submissions in subsequent pulls.
Previously, all submissions were always requested and Briefcase
identified and discarded duplicates on download, leading to a
potentially very large number of redundant network requests
and database checks during each pull activity. The historical
approach became prohibitively slow when the submission count
reached the hundreds of thousands or more. Both of these
new features required storing new metadata in Briefcase and
augmenting the graphical and command line interfaces.
Strengthening Security of Enketo
A current limitation of ODK Collect is that it runs only on
Android based devices. Enketo is a suite of JavaScript tools which
is a part of the ODK ecosystem and which among other uses can
provide a web interface to ODK Aggregate servers. This web-
based interface allows offline caching and allows data collection
to take place through any modern device. A longstanding feature
of ODK Collect is its ability to asymmetrically protect CRF data
at the level of the individual form using a powerful cryptographic
process. This use of cryptography has particular value for
research studies that collect sensitive data, such as those for which
the EDK system was designed. This is especially valuable because
study data encrypted at the level of the form are archived in the
PostgreSQL database, in all web-facing servers and in all backups
in an encrypted, non-human-readable format. Historically the
Enketo system did provide functionality for secure end-to-end
data transfers, but had no capability to encrypt CRF data at the
form level.
The implementation of a Java-based encryption methodology
in JavaScript was a challenging task because internet browsers
have no native equivalent implementations of the algorithms
used. We analysed and documented the encryption methods
used in ODK Collect, before reverse engineering them for use
in Enketo. We then developed a robust process for asymmetric
encryption in Enketo where the form data are encrypted using
a random single-use symmetric encryption key, which is in
turn then asymmetrically encrypted using a public Rivest–
Shamir–Adleman (RSA) key that is inherited from the CRF.
The resulting asymmetrically-encrypted symmetric encryption
key is then passed to the server with the form submission
and the form data can then be decrypted in ODK Briefcase
using a private RSA key that is possessed only by authorised
users. After testing with ODK Briefcase, we openly published
an Enketo encryption algorithm that works across platforms
and on all modern browsers. This implementation can handle
and co-encrypt binary attachments, such as photos, movies, and
data from other sources including third party apps. We went
on to author a sub-specification of the encryption algorithms,
which has now been published as part of the open ODK XForms
specification (9). To facilitate the creation of an alternative ODK-
compatible encryption/decryption library or application in the
future, we separated our encryption implementation into its own
module within the Enketo code-base (26).
An Open-Source Biometrics Framework
for ODK
To provide a basis for biometric registration of study participants,
we developed an ODK Biometrics framework (27) of open-
source tools for capturing (through the “Keppel” Android app
and hardware scanner) and later matching (through a javascript
CLI) ISO/IEC 19794-2 fingerprint templates. The Keppel app is a
standalone project that is designed to interface with ODK Collect
and its derivatives. The app currently works in combination
with the ODK Collect app, which is able to call for delivery
of fingerprint template data from the Keppel app using an
Android “intent” (a software action which allows two apps to
communicate with one another). The primary purpose of the
Keppel biometrics framework is to (1) assist with the process of
linking separate forms that relate to a specific study participant
and (2) to confirm the identity of an individual seeking access to
their study data as part of their rights of access. Keppel does not
currently perform fingerprint matching processes on the Android
device or mobile app. The Keppel CLI runs on Linux-like systems
and is able to compare pairs of templates and to provide a score
for the strength of the match between each pair. End-users are
able to select thresholds for match/mismatch classification that
provide the appropriate level of sensitivity and specificity for their
work. The Keppel app currently works with the Mantra MFS100,
a low cost (US$35) optical fingerprint scanner manufactured by
Mantra Softech India PLC (
Audit Trail Features for ODK
Many research studies and clinical trials require that enumerator
behaviour during data collection can be fully audited by
managers, external observers and regulators. We implemented
a system in which ODK Collect is optionally able to generate
a customisable log of enumerator behaviour and meta-data
during data entry activities. If an ODK form is designed to
include an audit, ODK Collect now creates a comma-separated-
values (.CSV) audit file and appends data to this form as the
form is opened or closed and as data are entered, changed
or removed. The audit file is invisible to the end user during
data collection and is encrypted using the standard ODK
encryption protocols. The basic audit log file records a number
of data entities, including events,nodes,start/end timestamps,
coordinates (lat/lon), old-value, new-value, and current user
(Table 1). Events represent a particular user action such as
opening a form, saving a form, or displaying a question. The audit
system is able to optionally record the identity of the current
user, to request the user’s identity each time the form is opened
and also to log the current longitude and latitude of the device
when data entry/modification took place. The old-value and new-
value entities are used to record changes in question type events
(i.e., changes made to the research data) and the system can
Frontiers in Public Health | 5November 2021 | Volume 9 | Article 665584
Marks et al. ODK Upgrades for Vaccine Trials
TABLE 1 | Exemplar ODK audit trail with coordinates, current user, timestamps, and form events logged.
Event Node Start End Latitude Longitude Accuracy Old-value New-value User
Form start 1550615022663 Ernest
Location tracking enabled 1550615022671 Ernest
Question /data/name 1550615022682 1550615097082 37.4229983 122.084 14.08 John Ernest
Location permissions granted 1550615068610 Ernest
Location providers enabled 1550615068665 Ernest
Location tracking disabled 1550615095914 37.4229983 122.084 14.08 Ernest
Question /data/age 1550615097082 1550615097655 37.4229983 122.084 14.08 20 Ernest
Question /data/name 1550615097656 1550615102351 37.4229983 122.084 14.08 John John Smith Ernest
Location tracking enabled 1550615099271 37.4229983 122.084 14.08 Ernest
Question /data/age 1550615102351 1550615107630 37.4229983 122.084 14.08 Ernest
End screen 1550615107631 1550615109199 37.4229983 122.084 14.08 Ernest
Form save 1550615109199 37.4229983 122.084 14.08 Ernest
Form exit 1550615109199 37.4229983 122.084 14.08 Ernest
Form finalise 1550615109199 37.4229983 122.084 14.08 Ernest
TABLE 2 | Event types in ODK audit logs.
Event Description Node? Timestamps? Coordinates? Answers?
Form start Start filling in the form No start only If on/available No
Question View a question Yes Yes If on/available If enabled
Group questions View multiple questions on one
screen (field-list)
Yes Yes If on/available No
Jump View the jump screen No start only If on/available No
Add repeat Add a repeat Yes Yes If on/available No
Delete repeat Delete a repeat Yes Yes If on/available No
End screen View the end screen No Yes If on/available No
Form save Save the form No start only If on/available No
Form exit Exit the form No start only If on/available No
Form resume Resume the form No start only If on/available No
Form finalise Finalise the form No start only If on/available No
Save error Error trying to save No start only If on/available No
Finalise error Error trying to finalise the form
(probably encryption related)
No start only If on/available No
Constraint error Constraint or required error on finalise No start only If on/available No
Location tracking enabled/disabled Toggle location tracking in Collect No Yes If on/available No
Location providers enabled/disabled Toggle location providers in Android No Yes If on/available No
Location permissions granted/not
Toggle location permission in Android
optionally collect meta-data describing the reasons for changes
having been made during a form editing session. Types of audit
events are described in Table 2. The nodes audit entity describes
the data field that was affected by the event and timestamps
provide information on the time and duration of the event.
Relying on the time reported by the device for timestamps could
allow users or the network to change the device time and thereby
manipulate the correctness of the audit log. For this reason, we
only use device time for the form start timestamp. All subsequent
event timestamps are therefore the result of elapsed time (which
users cannot change) added to the form start timestamp. This
means that whilst the timestamps themselves may potentially be
inaccurate, the time elapsed within and between the timestamps
are always accurate within one form editing session.
Automation of Data Management and
To enable an automated system to manage the pull, decrypt, and
export actions of Briefcase and to then perform data analysis and
report generation steps, we set up an Ubuntu VM and scheduled
automated operations using the cron utility. Cron is a powerful
time-based job scheduler that is native to all Linux- and Unix-
like systems. It allows computer code to be run on a regular
Frontiers in Public Health | 6November 2021 | Volume 9 | Article 665584
Marks et al. ODK Upgrades for Vaccine Trials
basis and at predefined times. Cron requires very little computing
experience and should be accessible to most users with support
from an information technology team. On Windows systems it is
possible to use the Windows Task Scheduler to achieve the same
goal of automated pull, export, and decrypt actions, or to use the
recently released Windows Subsystem for Linux.
We conceptualised and implemented the data management
tasks as two separate domains of work which included both
archiving and analysis activities. Archiving consisted of the
management of the raw form submissions, along with the
maintenance of a set of up-to-date and human readable tables of
raw data in comma separated value (CSV) format. The outputs of
the archiving phase thereby represented the aggregated data from
each CRF which formed the basis for all work in the Analysis
phase. All activities in the archiving phase were automatically
managed by ODK Briefcase. We used cron to run programmes (in
the form of bash scripts) that were able to control ODK Briefcase
via the CLI and to perform regular pull, export, and decrypt
actions. In order to protect the integrity of the data archive from
human errors, we treated the CSV files in the archive as volatile
entities that were subject to corruption if accessed by software
other than ODK Briefcase. In order to ensure the integrity of the
files, we isolated the archive from the analysis pipeline and used
only copies of the CSV data files in downstream analysis.
The completion of a cron-scheduled archive process (pull,
export, decrypt) triggered a series of Rscripts using a Linux “pipe”
and leading to Rs native Rscript CLI command. On initiation of
the Ranalysis, the first step was to make a working copy of the
most up to date CSV files (from the archiving phase) in a system
folder outside the ODK Briefcase managed archive folder.
Analysis was conducted primarily using R, a widely used open-
source statistical software package. The analysis of data included
the use of both Rand Rmarkdown scripts, which eventually
generated a large number of reports, charts, tables, line lists and
other outcomes that had been conceptualised by the field and
vaccination teams. We favoured the use of analysis tools that
were both simple to use and openly available. We used primarily
ggplot2 (28), plotly/ggplotly (29), leaflet (30), and flexdashboard
(31) to allow us to create interactive data visualisations that could
be easily modified by future users with minimal need for coding.
Because of the operational need to provide different teams with
daily line-lists, a number of reports were automatically formatted
as Microsoft Excel spreadsheets because Excel remains the default
tool for many teams working with lists or tables. Figure 3
provides a schematic representation of some of the outputs of
the system.
Automation of Data Management and
We demonstrated the utility of a system for electronic data
management during large-scale emergency clinical research
settings through our case study of platform deployment during
the response to the 2018–2020 North Kivu Ebola epidemic. Our
need for near-real-time reporting during this work highlighted
certain software behaviours that represented bottlenecks on the
time taken for data management activities when using ODK.
The large volumes of data that were being produced by the
intense field operations of ring vaccination for Ebola virus control
soon meant that the time taken for ODK Briefcase to download
>500,000 form submissions from the server and then to export
them to files for analysis was exceeding 24 h. We solved this
problem by implementing two new commands in Briefcase,
firstly the “resume-from-last” pull operation (ODK Briefcase
v.1.14.0) and later the “smart-append” export operation (ODK
Briefcase v.1.17.0). Through the inclusion of a JavaScript Object
Notation (.json) file within the app’s data storage directory, ODK
Briefcase now stores the progress of the last pull and export and
decrypt operations and since ODK Briefcase v.1.17.0 is not only
able to resume from the last positions, but also to intelligently
resume the position from restored backups and across mirrors or
forks of the system.
The use of metadata files to store the position was found
to be preferable over storing this information in a system level
preferences store as this change facilitated retention of the last
pull/export positions in backups, thereby eliminating the need to
start pulls and exports from the first submission after any system
failure and/or recovery from backup. This work highlighted
limitations of the Aggregate submission download API and has
fed into the design of a replacement API.
Between July 2018 and May 2020, data from more than
280,000 eligible study participants were recorded using the
system. In our hands and when working with more than 1.75
million form submissions on the server, the time taken to
perform a daily or ad hoc pull and export operation was reduced
from hours-or-days to seconds-or-minutes, with the time taken
now only dependent on the number of new submissions received
since the last pull/export action and not on the cumulative
number of submissions.
Project Oriented ODK Servers, Scalability,
and Audit Trails
The system established an extensibility model by which the
platform could be rapidly parallelised for use in other activities
and projects. The effectiveness of this project-centred approach
became clear in November 2018, when a small number of
Ebola Virus transmission chains were traced into Uganda (which
borders Eastern DRC) and where ring vaccination needed to start
the next day. Whilst the design and delivery of ring vaccination
activities undertaken in Uganda were identical to those in
DRC, the work came under the jurisdiction of the Ugandan
Ministry of Health and required separate administration and
data management. We took advantage of our extensibility
model to replicate the Ugandan ring vaccination system as
a new “project” and within 1 h of the first reports of cases
having been imported to the country, we had established a
fully operational system dedicated to work in that country. In
studies undertaken alongside the ring vaccination work, the
VSV-ZEBOV-GP vaccine was also used in a programme of
prophylactic vaccination studies that targeted healthcare workers
(HCWs) and “front-line” workers (FLWs) such as ambulance
Frontiers in Public Health | 7November 2021 | Volume 9 | Article 665584
Marks et al. ODK Upgrades for Vaccine Trials
FIGURE 3 | Fully customizable automated data flows. The time-based scheduler cron calls ODK Briefcase to perform data archiving and parsing of data from
individual forms to aggregated tables of data in comma separated variable (CSV) format. Cron calls to Rcontrol R markdown documents that perform statistical,
geospatial and demographic analysis, along with data manipulation to create line lists, audit documentation, interactive monitoring dashboards (based on open tools
such as Flexdashboard, LeafletJS, etc), interactive maps and other outputs that are shared to workstations of partners in the field, senior academic team, and
internal/external monitors.
drivers, porters, burial teams, and other people working in roles
with high risk of exposure. These studies took place not only in
DRC, but also in several neighbouring countries including South
Sudan, Uganda, Burundi, and Rwanda. Parallel EDK projects
were used by teams led by each of the local ministries of
health and the segregation of the project data between different
jurisdictions had the additional benefit that we were able to
comply with national and international data laws and best-
practises, and also to ensure that each country had total authority
over their own data. It was also possible to customise CRFs for
local study requirements, as well as to change the language(s)
used in the forms. Audit logs were implemented in ODK Collect
v1.25.0. As with the biometrics framework, the audit trail feature
is only available on the ODK Collect App and derivatives. No
provision for audit in Enketo webforms is currently available.
After testing with ODK Briefcase, we 228 openly published an
Enketo encryption algorithm that works across platforms and
on all modern 229 browsers. This implementation can handle
and co-encrypt binary attachments, such as photos, movies 230,
and data from other sources including third party apps. We
went on to author a sub-specification of the 231 encryption
algorithms, which has now been published as part of the open
ODK XForms specification 232 (9). To facilitate the creation
of an alternative ODK-compatible encryption/decryption library
or 233 application in the future, we separated our encryption
implementation into its own module within the 234 Enketo
code-base (26).
Real-World Implementation and
The international VSV-ZEBOV-GP prophylactic vaccination
programme ultimately saw the vaccination of around 40,000
FLWs/HCWs in DRC (n24,000), Burundi (n4,000), Rwanda
(n3,500), South Sudan (n3,000), and Uganda (n7,000). In
addition to our work during the North Kivu Ebola outbreak, we
have also used the EDK system to provide data collection systems
for research on other EVD research and in coronavirus disease
2019 (COVID-19) surveillance and vaccine/therapeutics trials;
as well as in more than 200 non-emergency research projects.
We have demonstrated the stability of the system by maintaining
more than 100 parallel projects over a period of 2 years. The new
audit trail and biometrics features became available towards the
end of the VSV-ZEBOV-GP campaign and were not used in the
field during that work. These features are however freely available
to all users of ODK and have subsequently been used in research
taking place during the COVID-19 pandemic.
We present tools for automation of data management and
reporting and open-source upgrades to the ODK ecosystem.
These improvements implement data audit-trails, biometric
participant registration, enhanced security for webforms and
multiple performance upgrades that facilitate scaling and
automation. These developments were driven by needs emerging
in real time during the complex and challenging response to
the North Kivu Ebola Virus Disease (EVD) epidemic. ODK
Frontiers in Public Health | 8November 2021 | Volume 9 | Article 665584
Marks et al. ODK Upgrades for Vaccine Trials
and Enketo were selected for use in this work primarily
because they provide the archetype for all ODK-related tools
(Ona, KoboToolbox, etc.) and so developments made here
affect functionality in a substantial proportion of modern EDC
platforms and tools. ODK-based tools are also highly attractive
to researchers and non-programmers because of the relative
simplicity of CRF design in the ODK ecosystem (which is based
around the use of Microsoft Excel spreadsheets). Further to this,
ODK is amenable to scaling and has been used in some of the
largest epidemiological studies ever undertaken, including the
Global Trachoma Mapping Programme (5).
In order to maximise potential impact and accessibility to the
outputs of this work, we ensured that all improvements made to
ODK and Enketo were integrated with the open-source code-
base of the main projects (26,32). Parties wishing to establish
a functionally analogous system to the one described here are
therefore encouraged to start with an installation of the most
up to date releases of ODK’s core suite of tools. A large global
community of ODK and Enketo users, along with users of related
systems including Ona, KoBoToolbox, SurveyCTO, and others
will benefit equitably from these developments and these toolkits
now represent a more complete data collection and management
solution for robust clinical research studies.
The performance improvements made during our work on
EDK have contributed to the design of a new, more performant
ODK server system, ODK Central (32), which is designed to
replace ODK Aggregate. ODK Central overcomes many of the
performance limitations of Aggregate when submission counts
get very large and also has native features that allow multiple
projects to be managed from a single server environment; thereby
removing the need to establish the more complex system of
parallel server front-ends used in the work described above.
The additional code we provide is sufficient to implement
the biometrics framework (27) and to replicate the core
infrastructure of an automated archiving and analysis system
(33), although users are expected to provide their own CRFs
and Rscripts and we provide only exemplar CRFs and an
analysis script which demonstrates the basic function of the
automated analysis pipeline. The CRFs and analysis scripts used
in vaccination activities surrounding the North Kivu epidemic
contain sensitive materials and parties interested in these should
contact the authors directly.
The use of webforms makes it possible to deploy ODK and
other tools without the need for any software installation, or for
any access to specific device types (i.e., Android). As a result
of our work, the Enketo system can now securely collect data
using any device which has a modern browser such as Chrome,
Firefox, or Safari. The use of Enketo is therefore a simple means
by which to increase the range of devices that can submit secure
data to an ODK server to now include iOS devices, smart
TVs, eReaders and both desktop and laptop PCs. Because they
facilitate the collection of data through a simple URL, webforms
are particularly useful when collecting data remotely in crowd-
sourcing, electronic online surveys, clinics, laboratories, and
many other settings that are less well-suited to the use of mobile
apps. The novel JavaScript-based method of encryption that
we have developed for Enketo is functionally analogous to the
encryption system used by ODK Collect and was implemented
in Enketo Express v1.72.0. The Enketo encryption system was
recently deployed in an online survey that studied the effects of
COVID-19 on health and well-being in 10,000 UK participants
(34). This system works both online and off-grid because Enketo
can cache completed (encrypted) forms in the browser until
an internet connexion is found. Enketo forms are also a built-
in feature of ODK Central, which further simplifies the work
required to set up a system that includes secure web-forms as part
of the data management solution.
Whilst the developments we present substantially improve the
utility of ODK tools in public health research, this ecosystem
is not without outstanding issues that represent barriers to
more flexible use in this context. Primary amongst these is
that data flow is unidirectional from devices towards the server
and complex workflows are currently required to filter data
back to the field via the server and back-end services. In
the study this was managed by unique study identifiers being
entered at each stage to facilitate down-stream data linkage.
Options for synchronisation of data between devices and the
server would greatly simplify the process of longitudinal study
by making data from earlier time-points and activities more
accessible to (and shareable between) enumerators in the field.
The ability to open and edit, or to add to forms previously
collected on a different device would also increase the range
of tasks to which the tools could be applied and would
reduce or remove the need to link different forms together
in downstream analysis. A key challenge to implementation
of both such capabilities comes from the need to be able to
maintain options for both online and off-grid working, though
requirements for off-grid capabilities are perhaps diminishing in
many low and middle income countries as mobile connectivity
becomes more accessible globally. Biometric frameworks for
not only registering, but also recognising study participants
would strengthen data collection by confirming the identity of
participants at key stages in data collection, in safety monitoring
and in upholding rights of participant access and data security.
Whilst SimPrints (25) already provides this type of functionality,
no free-for-use biometric framework for ODK is currently
available and future development of our open-source biometrics
framework will seek to implement both on-device fingerprint
matching/recognition and compatibility with a wider range of
hardware devices. The ability to extensively view, manage, search,
edit, and audit changes to the PostgreSQL database from within
the ODK Central server environment would further increase the
range of applications of ODK tools in clinical research studies;
as would tools for study randomisation. The combination of
biometric registration, functional audit-trails, and an auditable
data management interface on the server would combine to
make these open tools a more complete and attractive option
for use clinical research; and in particular in GCP compliant
clinical trials.
Publicly available datasets were analysed in this study. This
data can be found here: Project name: ODK; Project home
page:; Operating system(s): Collect: Android,
Briefcase, Aggregate: any; Programming language: Java;
Frontiers in Public Health | 9November 2021 | Volume 9 | Article 665584
Marks et al. ODK Upgrades for Vaccine Trials
Other requirements: e.g., Java; Licence: Apache 2.0; Any
restrictions to use by non-academics: None. Project name:
Enketo; Project home page:; Operating
system(s): Platform independent; Programming language:
JavaScript, XSL, HTML, CSS; Other requirements: N/A;
Licence: Apache 2.0; Any restrictions to use by non-academics:
None. Project name: ODK Biometrics/Keppel; Project home
Operating systems: Android (App), Platform Independent
(CLI); Programming language: Kotlin, JavaScript;
Other requirements: e.g., Licence: MIT Licence; Any
restrictions to use by non-academics: None. Project
name: EDK Automation tools; Project home page: https://; Operating
system(s): Platform Independent (CLI); Programming language:
Bash, R; Other requirements: e.g., Licence: MIT License; Any
restrictions to use by non-academics: None.
The studies involving human participants were reviewed and
approved by the software platform developments required no
specific ethics permission. Vaccination activities described in the
case study were approved by the Democratic Republic of Congo
(DRC) Ministry of health and DRC Public Health School Ethics
Committee or by the National Research Ethics Committees of
Rwanda, South Sudan, Uganda and Burundi. These activities
were coordinated in collaboration with the World Health
Organization and were carried out in accordance with GCP
standards and with the Declaration of Helsinki. All vaccination
activities were reviewed and monitored by an independent Data
and Safety Monitoring Board (DSMB). Written informed consent
to participate in this study was provided by the participants’ legal
guardian/next of kin.
MMar and CR: conceptualised the study. MR, HB, CS, GA,
SL, MMac, SG, YA, and HM: software/hardware engineering
and code. MR, GA, CS, HM, and YA: software development
on Enketo and ODK. CS: software development on ODK
biometrics framework. CR and MMar: software development
for automation and Ranalysis. PS-G, A-MR, AD, AT, MMar,
SL, HB, CH, CW, PK, and CR: platform development and field
deployment of the Ebola data monitoring system. All authors
wrote, reviewed, and approved the manuscript.
This research was funded by the Department of Health and
Social Care using UK Aid funding and is managed by the
NIHR (PR-OD-1017-20001).
We would like to express our profound appreciation and respect
to the people of North Kivu and Ituri, to thank our partners in
the WHO teams and Ministries of Health of Uganda, Rwanda,
Burundi, and South Sudan and also the field operatives who
allowed us to support them during the Équateur (2018) and
North Kivu (2018–2020) Ebola outbreaks. We are also grateful
for the support of the members of the ODK community
forum ( and to Esther Amon and
Eleanor Martins.
1. ODK. Available online at: (Accessed June 16, 2020).
2. Phiri TB, Kaunda-Khangamwa BN, Bauleni A, Chimuna T, Melody D,
Kalengamaliro H, et al. Feasibility, acceptability and impact of integrating
malaria rapid diagnostic tests and pre-referral rectal artesunate into the
integrated community case management programme. A pilot study in Mchinji
district, Malawi. Malar J. (2016) 15:177. doi: 10.1186/s12936-016-1237-2
3. Fornace KM, Surendra H, Abidin TR, Reyes R, Macalinao MLM, Stresman
G, et al. Use of mobile technology-based participatory mapping approaches
to geolocate health facility attendees for disease surveillance in low resource
settings. Int J Health Geogr. (2018) 17:21. doi: 10.1186/s12942-018-0141-0
4. Rajput ZA, Mbugua S, Amadi D, Chepngeno V, Saleem JJ, Anokwa Y,
et al. Evaluation of an android-based MHealth system for population
surveillance in developing countries. J Am Med Inform Assoc. (2012) 19:655–
9. doi: 10.1136/amiajnl-2011-000476
5. Solomon AW, Pavluck AL, Courtright P, Aboe A, Adamu L, Alemayehu
W, et al. The global trachoma mapping project: methodology of a 34-
country population-based study. Ophthalmic Epidemiol. (2015) 22:214–
25. doi: 10.3109/09286586.2015.1037401
6. Dixon J, MacPherson E, Manyau S, Nayiga S, Khine Zaw Y, Kayendeke
M, et al. The ‘Drug Bag’ method: lessons from anthropological studies of
antibiotic use in Africa and South-East Asia. Glob Health Action. (2019)
12:1639388. doi: 10.1080/16549716.2019.1639388
7. Habtamu E, Wondie T, Aweke S, Tadesse Z, Zerihun M, Zewudie Z, et al.
Posterior lamellar versus bilamellar tarsal rotation surgery for trachomatous
trichiasis in Ethiopia: a randomised controlled trial. Lancet Glob Health.
(2016) 4:E175–84. doi: 10.1016/S2214-109X(15)00299-5
8. Tom-Aba D, Olaleye A, Olayinka AT, Nguku P, Waziri N, Adewuyi
P, et al. Innovative technological approach to ebola virus disease
outbreak response in Nigeria using the open data kit and form hub
technology. PLoS ONE. (2015) 10:e0131000. doi: 10.1371/journal.pone.
9. ODK. XForms Specification. Available online at: https://GetodkGithubio/
Xforms-Spec/#Encryption (accessed June 16, 2020).
10. Ona. Available online at: (accessed June 16, 2020).
11. Kobo Toolbox. Available online at: (accessed June 16,
12. Survey CTO. Available online at: (accessed June
16, 2020).
13. CommCare. Available online at: (accessed
June 16, 2020).
14. Enketo. Available online at: (accessed June 16, 2020).
15. REDCap: Research Electronic Data Capture. Available online at: https:// Accessed 16 Jun 2020.
16. DHIS2. Available online at: (accessed 16 June 16,
17. R Core Team. R: A Language and Environment for Statistical Computing. R
Foundation for Statistical Computing. (2014).
18. World Health Organization. Ebola Situation Report - 30 March 2016. Ebola
(2016). Available online at:
ebola-situation- report-30-march-2016 (accessed June 16, 2020).
Frontiers in Public Health | 10 November 2021 | Volume 9 | Article 665584
Marks et al. ODK Upgrades for Vaccine Trials
19. World Health Organization. An R&D Blueprint For Action To Prevent
Epidemics. Plan of Action. (2016). Available online at: https://www.who.
int/Blueprint/About/r_d_Blueprint_Plan_of_Action.pdf (accessed June 16,
20. Nsio J, Kapetshi J, Makiala S, Raymond F, Tshapenda G, Boucher N, et al. 2017
outbreak of ebola virus disease in northern Democratic Republic of Congo. J
Infect Dis. (2019) 221:701–706. doi: 10.1093/infdis/jiz107
21. Rosello A, Mossoko M, Flasche S, Van Hoek AJ, Mbala P, Camacho A, et al.
Ebola virus disease in the Democratic Republic of the Congo, 1976-2014.
ELife. (2015) 4:e09015. doi: 10.7554/eLife.09015
22. Aruna A, Mbala P, Minikulu L, Mukadi D, Bulemfu D, Edidi F, et al.
Ebola virus disease outbreak - Democratic Republic of the Congo,
August 2018-November 2019. Morb Mortal Wkly Rep. (2019) 68:1162–
5. doi: 10.15585/mmwr.mm6850a3
23. Henao-Restrepo AM, Camacho A, Longini IM, Watson CH, Edmunds WJ,
Egger M, et al. Efficacy and effectiveness of an RVSV-vectored vaccine
in preventing ebola virus disease: final results from the guinea ring
vaccination, open-label, cluster-randomised trial (Ebola Ça Suffit!). Lancet.
(2017) 389:505–18. doi: 10.1016/S0140-6736(16)32621-6
24. Ebola ça Suffit Ring Vaccination Trial Consortium. The ring vaccination
trial: a novel cluster randomised controlled trial design to evaluate vaccine
efficacy and effectiveness during outbreaks, with special reference to ebola.
BMJ. (2015) 351:H3740. doi: 10.1136/bmj.h3740
25. Storisteanu DML, Norman TL, Grigore A, Norman TL. Biometric fingerprint
system to enable rapid and accurate identification of beneficiaries. Glob Health
Sci Pract. (2015) 3:135–7. doi: 10.9745/GHSP-D-15-00010
26. Enketo LLC. Official Enketo Libraries and Apps. (2020). Available online at: (accessed June 16, 2020).
27. LSHTM Global Health Analytics Group. LSHTM-ORK/ODK_Biometrics.
Kotlin (2020). Available online at:
Biometrics (accessed June 16, 2020).
28. Wickham H. Ggplot2: Elegant Graphics for Data Analysis. New York, NY:
Springer-Verlag. (2009). doi: 10.1007/978-0-387-98141-3
29. Sievert C. Plotly for R. (2018). Available online at:
(accessed May 11, 2021).
30. Cheng J, Karambelkar B, Xie Y. Leaflet: Create Interactive Web Maps With the
JavaScript “Leaflet”. (2019). Available online at:
package=leaflet (accessed May 11, 2021).
31. Iannone R, Allaire JJ, Borges B. Flexdashboard: R Markdown
Format for Flexible Dashboards. (2018). Available online at: https:// (accessed May 11,
32. Nafundi. ODK. Code Repository. Available online at:
getodk (accessed June 2020).
33. LSHTM Global Health Analytics Group. LSHTM-
ORK/EDK_Automation_Tools. (2020). Available online at: https:// (accessed June
16, 2020).
34. Rogers NT, Waterlow N, Brindle HE, Enria L, Eggo RM, Lees
S, et al. Behavioural change towards reduced intensity physical
activity is disproportionately prevalent among adults with serious
health issues or self-perception of high risk during the UK
COVID-19 lockdown. MedRxiv [Preprint]. doi: 10.1101/2020.05.12.
Author Disclaimer: The views expressed in this publication are those of the
authors and not necessarily those of the Department of Health and Social Care.
Conflict of Interest: CS, GA, HM, and YA work for Nafundi LLC, a company that
maintains ODK. MR is the lead developer, founder and owner of Enketo LLC.
The remaining authors declare that the research was conducted in the absence of
any commercial or financial relationships that could be construed as a potential
conflict of interest.
Publisher’s Note: All claims expressed in this article are solely those of the authors
and do not necessarily represent those of their affiliated organizations, or those of
the publisher, the editors and the reviewers. Any product that may be evaluated in
this article, or claim that may be made by its manufacturer, is not guaranteed or
endorsed by the publisher.
Copyright © 2021 Marks, Lal, Brindle, Gsell, MacGregor, Stott, van de Rijdt,
Almazor, Golia, Watson, Diallo, Toure, Houlihan, Keating, Martin, Restrepo,
Anokwa and Roberts. This is an open-access article distributed under the terms
of the Creative Commons Attribution License (CC BY). The use, distribution or
reproduction in other forums is permitted, provided the original author(s) and the
copyright owner(s) are credited and that the original publication in this journal
is cited, in accordance with accepted academic practice. No use, distribution or
reproduction is permitted which does not comply with these terms.
Frontiers in Public Health | 11 November 2021 | Volume 9 | Article 665584
... A raft of EDC tools aimed specifically at work in LMICs have recently become available. These predominantly app-based tools, which include ODK (7), Ona Data (8), KoBoToolbox (9), SurveyCTO (10), CommCare by Dimagi (11) Enketo (12), REDCap (13), OpenClinica (14) and DHIS2 (15) all have the capability to function without reliable internet connections and can be used to generate digitised data sets of very high quality in even the most inhospitable and complex settings (16). Several of these platforms (ODK, SurveyCTO, KoBoToolbox, Ona and CommCare) are derived from a common code-base (ODK) and have overlapping functions, but none of these software platforms currently provide a robust built-in method for capturing verifiable biometric proof of identity data that can be used to later [1] link CRFs that were collected at different times or [2] confirm the identity of a participant requesting deletion of their data. ...
... Data were collected on encrypted Android handsets using ODK Collect (16) and were hosted at institutional data centres. We recruited volunteers to provide fingerprints for a real-world evaluation of the system. ...
... Scant literature addresses topics of participant (mis)identification in research and trials, but in clinical care the subject is better explored (23,24) and highlights that patient mis-identification is common even in high income countries (HICs) with well-developed health information systems. In resource-restricted settings, challenges are magnified, particularly when complex longitudinal data collection centres around the use of experimental medical products, or the surveillance of infectious diseases in remote locales (16). ...
Full-text available
The availability of low-cost biometric hardware sensors and software makes it possible to rapidly, affordably and securely sample and store a unique and invariant biological signature (or biometric “template”) of participants in research and trials. This has applications in consent, linkage of case reporting forms collected at different times, and in confirmation of participant identity for purposes of safety monitoring and adherence to international data laws. The use of mobile electronic data collection software has recently become commonplace in clinical trials and research. A raft of tools based on the open-source ODK project now provide diverse options for data management that work consistently in resource-restricted settings, but none have built-in functionality for capturing biometric templates. In this study, we report the development and testing of a novel open-source app and associated method for capturing and matching biometric fingerprint templates during data collection with the popular data platforms ODK, KoBoToolbox, SurveyCTO, Ona and CommCare. Using data from more than 1000 fingers, we show that fingerprint templates can be used to link data forms with high accuracy and that this accuracy increases with the addition of multiple fingerprints on each data form. By focussing on publishing open-source code and documentation, and by using an affordable (<£50) and mass-produced model of fingerprint sensor, we are able to make this platform freely available to the large global user community that utilises ODK and related data collection systems.
... 10,11,13 To improve health programs in many resource-poor nations, the open data kit (ODK), opensource mobile software has been used by healthcare workers to rapidly collect, collate, and support the manipulation of intricate and large volumes of data types (text, location, images, audio, video, barcodes). [14][15][16] The ODK is the most vastly utilized, as it reinforces the function of many other electronic data collection (EDC) tools such as; Ona, Kobo Toolbox, Survey CTO, COMM Care, Enketo, and triangulates with Redcap, DHIS2, and R through application program interfaces (APIs). 14,17 ODK permits vast deployment and evaluation of the diagnostic test, therapeutics, and access to the compassionate use of safe and efficacious vaccines to curb the spread of this dangerous viral haemorrhagic disease. ...
Full-text available
Background: The proper management of healthcare data is fundamental to the health system processes; artificial intelligence has proven its value in these processes. Artificial intelligence can simplify the management of information, improve data security, and automate data flow. It is also useful in the analysis and interpretation of big data. Hence, it has the possibility of screening and diagnosing diseases, categorizing disease severity, detecting therapeutic agents, and forecasting outbreak spots.Methods: A data analysis and validation engine was developed to perform data quality control checks, classify addresses, and generate epidemiology numbers using the index and parse command on the command-line interface of DAVE.Results: DAVE correctly formatted data and created a local copy of the datastore and the index. It also returned previous EPID numbers to each entry and assigned a new EPID number to missed entries. DAVE imported the entries into the data template of the existing data management tool and generated a sample manifest that is then sent to the Laboratory. The data flow from the point of collection to storage and reporting was assessed as 100% accurate without errors and in real-time; there was also the ability to roll back if any error occurred.Conclusions: DAVE is a semi-autonomous system that operates with minimal human intervention; it is automatically faster as it leverages computing power to parse, store, and retrieve data while practically eliminating the need for manual data quality assessment. The DAVE functionality can be extended to incorporate additional features like forecasting outbreaks of emerging/re-emerging diseases, categorizing the severity of diseases and analysis of data in our setting.
... We collect data using electronic case record forms (eCRFs) developed using Open Data Kit (ODK) software on password protected tablets, and on standardised paper CRFs when eCRF use is not possible. 28 We designed a modular system of ODK forms in English and French to collect data on different aspects of the study, including consent, vaccinations and errors on eCRF completion. Prior to starting the study, we provided and reviewed guidelines for CRF completion with study site personnel. ...
Full-text available
Introduction Ebola virus disease (EVD) continues to be a significant public health problem in sub-Saharan Africa, especially in the Democratic Republic of the Congo (DRC). Large-scale vaccination during outbreaks may reduce virus transmission. We established a large population-based clinical trial of a heterologous, two-dose prophylactic vaccine during an outbreak in eastern DRC to determine vaccine effectiveness. Methods and analysis This open-label, non-randomised, population-based trial enrolled eligible adults and children aged 1 year and above. Participants were offered the two-dose candidate EVD vaccine regimen VAC52150 (Ad26.ZEBOV, Modified Vaccinia Ankara (MVA)-BN-Filo), with the doses being given 56 days apart. After vaccination, serious adverse events (SAEs) were passively recorded until 1 month post dose 2. 1000 safety subset participants were telephoned at 1 month post dose 2 to collect SAEs. 500 pregnancy subset participants were contacted to collect SAEs at D7 and D21 post dose 1 and at D7, 1 month, 3 months and 6 months post dose 2, unless delivery was before these time points. The first 100 infants born to these women were given a clinical examination 3 months post delivery. Due to COVID-19 and temporary suspension of dose 2 vaccinations, at least 50 paediatric and 50 adult participants were enrolled into an immunogenicity subset to examine immune responses following a delayed second dose. Samples collected predose 2 and at 21 days post dose 2 will be tested using the Ebola viruses glycoprotein Filovirus Animal Non-Clinical Group ELISA. For qualitative research, in-depth interviews and focus group discussions were being conducted with participants or parents/care providers of paediatric participants. Ethics and dissemination Approved by Comité National d’Ethique et de la Santé du Ministère de la santé de RDC, Comité d'Ethique de l'Ecole de Santé Publique de l’Université de Kinshasa, the LSHTM Ethics Committee and the MSF Ethics Review Board. Findings will be presented to stakeholders and conferences. Study data will be made available for open access. Trial registration number NCT04152486 .
Full-text available
Background: An interventional study named Malakit was implemented between April 2018 and March 2020 to address malaria in gold mining areas in French Guiana, in collaboration with Suriname and Brazil. This innovative intervention relied on the distribution of kits for self-diagnosis and self-treatment to gold miners after training by health mediators, referred to in the project as facilitators. Objective: This paper aims to describe the process by which the information system was designed, developed, and implemented to achieve the monitoring and evaluation of the Malakit intervention. Methods: The intervention was implemented in challenging conditions at five cross-border distribution sites, which imposed strong logistical constraints for the design of the information system: isolation in the Amazon rainforest, tropical climate, and lack of reliable electricity supply and internet connection. Additional constraints originated from the interaction of the multicultural players involved in the study. The Malakit information system was developed as a patchwork of existing open-source software, commercial services, and tools developed in-house. Facilitators collected data from participants using Android tablets with ODK (Open Data Kit) Collect. A custom R package and a dashboard web app were developed to retrieve, decrypt, aggregate, monitor, and clean data according to feedback from facilitators and supervision visits on the field. Results: Between April 2018 and March 2020, nine facilitators generated a total of 4863 form records, corresponding to an average of 202 records per month. Facilitators’ feedback was essential for adapting and improving mobile data collection and monitoring. Few technical issues were reported. The median duration of data capture was 5 (IQR 3-7) minutes, suggesting that electronic data capture was not taking more time from participants, and it decreased over the course of the study as facilitators become more experienced. The quality of data collected by facilitators was satisfactory, with only 3.03% (147/4849) of form records requiring correction. Conclusions: The development of the information system for the Malakit project was a source of innovation that mirrored the inventiveness of the intervention itself. Our experience confirms that even in a challenging environment, it is possible to produce good-quality data and evaluate a complex health intervention by carefully adapting tools to field constraints and health mediators’ experience. Trial Registration: NCT03695770;
Full-text available
Background: Approval for the use of COVID-19 vaccines has been granted in a number of countries but there are concerns that vaccine uptake may be low amongst certain groups. Methods: This study used a mixed methods approach based on online survey and an embedded quantitative/qualitative design to explore perceptions and attitudes that were associated with intention to either accept or refuse offers of vaccination in different demographic groups during the early stages of the UK's mass COVID-19 vaccination programme (December 2020). Analysis used multivariate logistic regression, structural text modeling and anthropological assessments. Results: Of 4,535 respondents, 85% ( n = 3,859) were willing to have a COVID-19 vaccine. The rapidity of vaccine development and uncertainties about safety were common reasons for COVID-19 vaccine hesitancy. There was no evidence for the widespread influence of mis-information, although broader vaccine hesitancy was associated with intentions to refuse COVID-19 vaccines (OR 20.60, 95% CI 14.20–30.30, p < 0.001). Low levels of trust in the decision-making (OR 1.63, 95% CI 1.08, 2.48, p = 0.021) and truthfulness (OR 8.76, 95% CI 4.15–19.90, p < 0.001) of the UK government were independently associated with higher odds of refusing COVID-19 vaccines. Compared to political centrists, conservatives and liberals were, respectively, more (OR 2.05, 95%CI 1.51–2.80, p < 0.001) and less (OR 0.30, 95% CI 0.22–0.41, p < 0.001) likely to refuse offered vaccines. Those who were willing to be vaccinated cited both personal and public protection as reasons, with some alluding to having a sense of collective responsibility. Conclusion: Dominant narratives of COVID-19 vaccine hesitancy are misconceived as primarily being driven by misinformation. Key indicators of UK vaccine acceptance include prior behaviors, transparency of the scientific process of vaccine development, mistrust in science and leadership and individual political views. Vaccine programmes should leverage the sense of altruism, citizenship and collective responsibility that motivated many participants to get vaccinated.
Full-text available
Objectives: We assessed whether lockdown had a disproportionate impact on physical activity behavior in groups who were, or who perceived themselves to be, at heightened risk from COVID-19. Methods: Physical activity intensity (none, mild, moderate, or vigorous) before and during the UK COVID-19 lockdown was self-reported by 9,190 adults between 2020-04-06 and 2020-04-22. Physician-diagnosed health conditions and topic composition of open-ended text on participants' coping strategies were tested for associations with changes in physical activity. Results: Most (63.9%) participants maintained their normal physical activity intensity during lockdown, 25.0% changed toward less intensive activity and 11.1% were doing more. Doing less intensive physical activity was associated with obesity (OR 1.25, 95% CI 1.08–1.42), hypertension (OR 1.25, 1.10–1.40), lung disease (OR 1.23, 1.08–1.38), depression (OR 2.05, 1.89–2.21), and disability (OR 2.13, 1.87–2.39). Being female (OR 1.25, 1.12–1.38), living alone (OR 1.20, 1.05–1.34), or without access to a garden (OR 1.74, 1.56–1.91) were also associated with doing less intensive physical activity, but being in the highest income group (OR 1.73, 1.37–2.09) or having school-age children (OR 1.29, 1.10–1.49) were associated with doing more. Younger adults were more likely to change their PA behavior compared to older adults. Structural topic modeling of narratives on coping strategies revealed associations between changes in physical activity and perceptions of personal or familial risks at work or at home. Conclusions: Policies on maintaining or improving physical activity intensity during lockdowns should consider (1) vulnerable groups of adults including those with chronic diseases or self-perceptions of being at risk and (2) the importance of access to green or open spaces in which to exercise.
Full-text available
Importance: There are growing concerns that the UK COVID-19 lockdown has reduced opportunities to maintain health through physical activity, placing individuals at higher risk of chronic disease and leaving them more vulnerable to severe sequelae of COVID-19. Objective: To examine whether the UK's lockdown measures have had disproportionate impacts on intensity of physical activity in groups who are, or who perceive themselves to be, at heightened risk from COVID-19. Designs, Setting, Participants: UK-wide survey of adults aged over 20, data collected between 2020-04-06 and 2020-04-22. Exposures: Self-reported doctor-diagnosed obesity, hypertension, type I/II diabetes, lung disease, cancer, stroke, heart disease. Self-reported disabilities and depression. Sex, gender, educational qualifications, household income, caring for school-age children. Narrative data on coping strategies. Main Outcomes and Measures: Change in physical activity intensity after implementation of UK COVID-19 lockdown (self-reported). Results: Most (60%) participants achieved the same level of intensity of physical activity during the lockdown as before the epidemic. Doing less intensive physical activity during the lockdown was associated with obesity (OR 1.21, 95% CI 1.02-1.41), hypertension (OR 1.52, 1.33-1.71), lung disease (OR 1.31,1.13-1.49), depression (OR 2.02, 1.82-2.22) and disability (OR 2.34, 1.99-2.69). Participants who reduced their physical activity intensity also had higher odds of being female, living alone or having no garden, and more commonly expressed sentiments about personal or household risks in narratives on coping. Conclusions and relevance: Groups who reduced physical activity intensity included disproportionate numbers of people with either heightened objective clinical risks or greater tendency to express subjective perceptions of risk. Policy on exercise for health during lockdowns should include strategies to facilitate health promoting levels of physical activity in vulnerable groups, including those with both objective and subjective risks.
Full-text available
On August 1, 2018, the Democratic Republic of the Congo Ministry of Health (DRC MoH) declared the tenth outbreak of Ebola virus disease (Ebola) in DRC, in the North Kivu province in eastern DRC on the border with Uganda, 8 days after another Ebola outbreak was declared over in northwest Équateur province. During mid- to late-July 2018, a cluster of 26 cases of acute hemorrhagic fever, including 20 deaths, was reported in North Kivu province.* Blood specimens from six patients hospitalized in the Mabalako health zone and sent to the Institut National de Recherche Biomédicale (National Biomedical Research Institute) in Kinshasa tested positive for Ebola virus. Genetic sequencing confirmed that the outbreaks in North Kivu and Équateur provinces were unrelated. From North Kivu province, the outbreak spread north to Ituri province, and south to South Kivu province (1). On July 17, 2019, the World Health Organization designated the North Kivu and Ituri outbreak a public health emergency of international concern, based on the geographic spread of the disease to Goma, the capital of North Kivu province, and to Uganda and the challenges to implementing prevention and control measures specific to this region (2). This report describes the outbreak in the North Kivu and Ituri provinces. As of November 17, 2019, a total of 3,296 Ebola cases and 2,196 (67%) deaths were reported, making this the second largest documented outbreak after the 2014-2016 epidemic in West Africa, which resulted in 28,600 cases and 11,325 deaths.† Since August 2018, DRC MoH has been collaborating with partners, including the World Health Organization, the United Nations Children's Fund, the United Nations Office for the Coordination of Humanitarian Affairs, the International Organization of Migration, The Alliance for International Medical Action (ALIMA), Médecins Sans Frontières, DRC Red Cross National Society, and CDC, to control the outbreak. Enhanced communication and effective community engagement, timing of interventions during periods of relative stability, and intensive training of local residents to manage response activities with periodic supervision by national and international personnel are needed to end the outbreak.
Full-text available
Understanding the prevalence and types of antibiotics used in a given human and/or animal population is important for informing stewardship strategies. Methods used to capture such data often rely on verbal elicitation of reported use that tend to assume shared medical terminology. Studies have shown the category ‘antibiotic’ does not translate well linguistically or conceptually, which limits the accuracy of these reports. This article presents a ‘Drug Bag’ method to study antibiotic use (ABU) in households and on farms, which involves using physical samples of all the antibiotics available within a given study site. We present the conceptual underpinnings of the method, and our experiences of using this method to produce data about antibiotic recognition, use and accessibility in the context of anthropological research in Africa and South-East Asia. We illustrate the kinds of qualitative and quantitative data the method can produce, comparing and contrasting our experiences in different settings. The Drug Bag method produce accurate antibiotic use data as well as provide a talking point for participants to discuss antibiotic experiences. We propose it can help improve our understanding of antibiotic use in peoples’ everyday lives across different contexts, and our reflections add to a growing conversation around methods to study ABU beyond prescriber settings, where data gaps are currently substantial.
Full-text available
Background: In 2017, the Democratic Republic of the Congo (DRC) recorded its eighth Ebola virus disease (EVD) outbreak, approximately 3 years after the previous outbreak. Methods: Suspect cases of EVD were identified on the basis of clinical and epidemiological information. Reverse transcription-polymerase chain reaction (RT-PCR) analysis or serological testing was used to confirm Ebola virus infection in suspected cases. The causative virus was later sequenced from a RT-PCR-positive individual and assessed using phylogenetic analysis. Results: Three probable and 5 laboratory-confirmed cases of EVD were recorded between 27 March and 1 July 2017 in the DRC. Fifty percent of cases died from the infection. EVD cases were detected in 4 separate areas, resulting in > 270 contacts monitored. The complete genome of the causative agent, a variant from the Zaireebolavirus species, denoted Ebola virus Muyembe, was obtained using next-generation sequencing. This variant is genetically closest, with 98.73% homology, to the Ebola virus Mayinga variant isolated from the first DRC outbreaks in 1976-1977. Conclusion: A single spillover event into the human population is responsible for this DRC outbreak. Human-to-human transmission resulted in limited dissemination of the causative agent, a novel Ebola virus variant closely related to the initial Mayinga variant isolated in 1976-1977 in the DRC.
Full-text available
Background: Identifying fine-scale spatial patterns of disease is essential for effective disease control and elimination programmes. In low resource areas without formal addresses, novel strategies are needed to locate residences of individuals attending health facilities in order to efficiently map disease patterns. We aimed to assess the use of Android tablet-based applications containing high resolution maps to geolocate individual residences, whilst comparing the functionality, usability and cost of three software packages designed to collect spatial information. Results: Using Open Data Kit GeoODK, we designed and piloted an electronic questionnaire for rolling cross sectional surveys of health facility attendees as part of a malaria elimination campaign in two predominantly rural sites in the Rizal, Palawan, the Philippines and Kulon Progo Regency, Yogyakarta, Indonesia. The majority of health workers were able to use the tablets effectively, including locating participant households on electronic maps. For all households sampled (n = 603), health facility workers were able to retrospectively find the participant household using the Global Positioning System (GPS) coordinates and data collected by tablet computers. Median distance between actual house locations and points collected on the tablet was 116 m (IQR 42-368) in Rizal and 493 m (IQR 258-886) in Kulon Progo Regency. Accuracy varied between health facilities and decreased in less populated areas with fewer prominent landmarks. Conclusions: Results demonstrate the utility of this approach to develop real-time high-resolution maps of disease in resource-poor environments. This method provides an attractive approach for quickly obtaining spatial information on individuals presenting at health facilities in resource poor areas where formal addresses are unavailable and internet connectivity is limited. Further research is needed on how to integrate these with other health data management systems and implement in a wider operational context.
Full-text available
Background: rVSV-ZEBOV is a recombinant, replication competent vesicular stomatitis virus-based candidate vaccine expressing a surface glycoprotein of Zaire Ebolavirus. We tested the effect of rVSV-ZEBOV in preventing Ebola virus disease in contacts and contacts of contacts of recently confirmed cases in Guinea, west Africa. Methods: We did an open-label, cluster-randomised ring vaccination trial (Ebola ça Suffit!) in the communities of Conakry and eight surrounding prefectures in the Basse-Guinée region of Guinea, and in Tomkolili and Bombali in Sierra Leone. We assessed the efficacy of a single intramuscular dose of rVSV-ZEBOV (2×107plaque-forming units administered in the deltoid muscle) in the prevention of laboratory confirmed Ebola virus disease. After confirmation of a case of Ebola virus disease, we definitively enumerated on a list a ring (cluster) of all their contacts and contacts of contacts including named contacts and contacts of contacts who were absent at the time of the trial team visit. The list was archived, then we randomly assigned clusters (1:1) to either immediate vaccination or delayed vaccination (21 days later) of all eligible individuals (eg, those aged ≥18 years and not pregnant, breastfeeding, or severely ill). An independent statistician generated the assignment sequence using block randomisation with randomly varying blocks, stratified by location (urban vs rural) and size of rings (≤20 individuals vs >20 individuals). Ebola response teams and laboratory workers were unaware of assignments. After a recommendation by an independent data and safety monitoring board, randomisation was stopped and immediate vaccination was also offered to children aged 6-17 years and all identified rings. The prespecified primary outcome was a laboratory confirmed case of Ebola virus disease with onset 10 days or more from randomisation. The primary analysis compared the incidence of Ebola virus disease in eligible and vaccinated individuals assigned to immediate vaccination versus eligible contacts and contacts of contacts assigned to delayed vaccination. This trial is registered with the Pan African Clinical Trials Registry, number PACTR201503001057193. Findings: In the randomised part of the trial we identified 4539 contacts and contacts of contacts in 51 clusters randomly assigned to immediate vaccination (of whom 3232 were eligible, 2151 consented, and 2119 were immediately vaccinated) and 4557 contacts and contacts of contacts in 47 clusters randomly assigned to delayed vaccination (of whom 3096 were eligible, 2539 consented, and 2041 were vaccinated 21 days after randomisation). No cases of Ebola virus disease occurred 10 days or more after randomisation among randomly assigned contacts and contacts of contacts vaccinated in immediate clusters versus 16 cases (7 clusters affected) among all eligible individuals in delayed clusters. Vaccine efficacy was 100% (95% CI 68·9-100·0, p=0·0045), and the calculated intraclass correlation coefficient was 0·035. Additionally, we defined 19 non-randomised clusters in which we enumerated 2745 contacts and contacts of contacts, 2006 of whom were eligible and 1677 were immediately vaccinated, including 194 children. The evidence from all 117 clusters showed that no cases of Ebola virus disease occurred 10 days or more after randomisation among all immediately vaccinated contacts and contacts of contacts versus 23 cases (11 clusters affected) among all eligible contacts and contacts of contacts in delayed plus all eligible contacts and contacts of contacts never vaccinated in immediate clusters. The estimated vaccine efficacy here was 100% (95% CI 79·3-100·0, p=0·0033). 52% of contacts and contacts of contacts assigned to immediate vaccination and in non-randomised clusters received the vaccine immediately; vaccination protected both vaccinated and unvaccinated people in those clusters. 5837 individuals in total received the vaccine (5643 adults and 194 children), and all vaccinees were followed up for 84 days. 3149 (53·9%) of 5837 individuals reported at least one adverse event in the 14 days after vaccination; these were typically mild (87·5% of all 7211 adverse events). Headache (1832 [25·4%]), fatigue (1361 [18·9%]), and muscle pain (942 [13·1%]) were the most commonly reported adverse events in this period across all age groups. 80 serious adverse events were identified, of which two were judged to be related to vaccination (one febrile reaction and one anaphylaxis) and one possibly related (influenza-like illness); all three recovered without sequelae. Interpretation: The results add weight to the interim assessment that rVSV-ZEBOV offers substantial protection against Ebola virus disease, with no cases among vaccinated individuals from day 10 after vaccination in both randomised and non-randomised clusters. Funding: WHO, UK Wellcome Trust, the UK Government through the Department of International Development, Médecins Sans Frontières, Norwegian Ministry of Foreign Affairs (through the Research Council of Norway's GLOBVAC programme), and the Canadian Government (through the Public Health Agency of Canada, Canadian Institutes of Health Research, International Development Research Centre and Department of Foreign Affairs, Trade and Development).
Full-text available
Background The World Health Organization recommends that persons of all ages suspected of malaria should receive a parasitological confirmation of malaria by use of malaria rapid diagnostic test (RDT) at community level, and that rectal artesunate should be used as a pre-referral treatment for severe malaria to rapidly reduce parasitaemia. This paper reports on findings from a pilot study that assessed the feasibility, acceptability and effects of integrating RDTs and pre-referral rectal artesunate into the integrated Community Case Management programme in Malawi. Methods This study used mixed methods to collect information for this survey. Pre- and post-intervention, cross-sectional, household surveys were carried out. A review of integrated community case management reports, including supervision checklists was conducted. Quantitative data were collected in tablets running on open data kit software, and then data were transferred to STATA version 12 for analysis. For key indicators, proportions were calculated at 95 % confidence intervals. Qualitative data were recorded onto digital recorders, translated into English and transcribed for analysis. Results Out of 86 observed RDT performances, a total of 83 (97 %) were performed correctly with a proper disposal of sharps and biohazard wastes. Only two (2 %) febrile children who had an RDT negative result were treated with artemether–lumefantrine, contrary to malaria treatment guidelines. Utilization of community health workers (CHWs) as a first source of care increased from (33.9 %) (95 % CI; 25.5–42.3) at baseline to (89.7 %) (95 % CI; 83.5–95.5) at end line in the intervention villages. There was a corresponding decrease in the proportion of caregivers that first sought care from informal sources from 12.9 % (95 % CI; 6.9–18.9) to 1.9 % (95 % CI; 0.9–4.4) in the intervention villages. Acceptability of the use of RDTs and pre-referral rectal artesunate at the community level was relatively high. Conclusion Integration of RDTs and pre-referral rectal at artesunate community level is both feasible and acceptable. The strategy has the potential to increase and improve utilization of child health services at community level. However, this depends on the CHWs’ skills and their availability in remote areas.
The Democratic Republic of the Congo has experienced the most outbreaks of Ebola virus disease since the virus' discovery in 1976. This article provides for the first time a description and a line list for all outbreaks in this country, comprising 996 cases. Compared to patients over 15 years old, the odds of dying were significantly lower in patients aged 5 to 15 and higher in children under five (with 100% mortality in those under 2 years old). The odds of dying increased by 11% per day that a patient was not hospitalised. Outbreaks with an initially high reproduction number, R (>3), were rapidly brought under control, whilst outbreaks with a lower initial R caused longer and generally larger outbreaks. These findings can inform the choice of target age groups for interventions and highlight the importance of both reducing the delay between symptom onset and hospitalisation and rapid national and international response.