scroll

Although digitized data and information flows did not constitute the norm, they did contribute meaningfully to the Ebola outbreak response in specific instances, and introduced both quantitative and qualitative differences in data and information flows in the response when used. The following case studies[1] provide snapshots of the use and impact of digital technologies in the Ebola outbreak response. These case studies identify:

  • What: What was the tool, platform, or service used?
  • Why: In what context was the digital technology used, and for what purpose? 
  • Who: Who were the primary actors engaged in use of the technology or technologies?
  • How: How did the digital technology or technologies function?
  • Analysis: What outcomes and insights resulted?
  • Challenges: What challenges did actors encounter?

The case studies are organized by actor groups participating in the information and/or data exchange, including:

  • national governments,
  • traditional response and donor organizations (such as foreign donors, NGOs, and intergovernmental organizations like UN agencies),
  • frontline workers (such as community health workers or contact tracers),
  • citizens and affected populations, and
  • “remote” responders supporting the formal response (such as the digital humanitarians, members of diaspora groups, and context experts).
Connecting Frontline Workers and Governments

mHero: SMS between Frontline Health Workers and the Central Health Ministry in Liberia

What: The Mobile Health Worker Electronic Response and Outreach (dubbed mHero) system launched in pilot form in Liberia in November 2014 to meet the need for real-time data and information exchange between the central Ministry of Health and frontline health workers. It aimed to strengthen the government’s health information system with up-to-date information about health worker location and availability on the one hand, and to provide critical information to support health workers on the frontlines of the crisis on the other. In doing so it created a two-way flow of information between health workers and response organizers in real time.

Why: Health workers played a critical role on the frontlines of the Ebola outbreak response. For responding organizations, whether national or international, frontline health workers served as critical eyes and ears and identified cases as they emerged, and helped to treat the ill. In their work to combat the spread of Ebola, health workers of all kinds faced a significant risk of contracting the virus. Varying estimates calculated health workers' risk of Ebola exposure at between 21 percent and 42 percent higher than that of the average citizen in the three most-affected countries in West Africa[2]. Consequently, many health workers fell ill; others abandoned or moved their posts.

In many countries, including those most affected by the Ebola outbreak, governments lacked a clear picture of where health workers were located and what services they were providing to their clients, and where health workers were located. To fight Ebola, governments needed real-time, up-to-date information on the availability, location, and needs of their health workforce (e.g., did they have sufficient personal protective equipment?). Health workers, in turn, needed access to information about how to protect themselves against the virus and to share critical data about the disease’s spread.

Who:  In the Liberia deployment, mHero was launched by the national health ministry with a consortium of partners, including UNICEF, IntraHealth, and K4Health[3]. Prior to the outbreak, the Ministry of Health had been collaborating with IntraHealth and UNICEF on a digital health[4] worker database called iHRIS (a modified abbreviation of the open source “Human Resources Information Solutions” platform), and an SMS messaging system called RapidPro. mHero was planned and designed to enable interoperability between these two systems.

How: mHero is a mobile phone-based communication system that for the first time connected central ministry staff with frontline health workers via two-way text messages using the basic mobile phones that most health workers already owned. The platform allowed them to receive critical information in real time.

mHero integrated the existing iHRIS and RapidPro platforms, leveraging iHRIS to access the mobile phone numbers of government frontline health workers, and RapidPro SMS messaging to enable communication via text message that could be delivered to all or a targeted subset of those in the iHRIS database. It uses the principles and standards of OpenHIE4[5] to aggregate key data from human resource information systems and securely share and validate health worker information, and mHero can share data with the DHIS2 web-based health information management system. In Liberia, mHero enabled nuanced and tailored messaging to health workers based upon their qualifications, position, location, specialty, and how recently they had been trained. Another game-changing aspect of this innovation was its ability to enable health workers to initiate direct and real-time contact with health officials--allowing data and information to be “pushed” by health workers, in addition to “pulled” at the request of health officials.[6] This functionality is particularly critical in the context of detecting and monitoring a disease outbreak.

Analysis

  • The mHero pilot was used to validate frontline workers’ contact information and get a real-time picture of worker location and availability. By the pilot’s close in December of 2014, the health ministry had used mHero to reach nearly 500 health workers at facilities in four counties. The ministry used these SMS exchanges to validate health workers’ contact information, including their phone numbers, locations, job titles, supervisor information, and health facility association.
  • The Ministry of Health adapted mHero for over three dozen uses over the course of the response and recovery. Health ministry officials used the pilot to update personnel records in iHRIS, rapidly building the government’s capacity to engage real-time information with frontline health workers in the future. Between November 2014 and July 2016, over 40 mHero deployments[7] reached more than 7,000 health workers, showing increased use by both frontline health workers--including civil servant and community volunteer health workers--and central ministry staff.
  • The Ministry of Health adapted mHero to enable messages to be initiated by health workers, as well as by the central ministry. As the use of mHero evolved over the course of the response, the government began to lay plans to incorporate SMS messages originating with health workers. “I think it is about time we find a way the health workers can communicate directly to us. A health worker may have some burning issues to share, but we may [be] missing his or her concerns,” said Stephen Gbanyan, Director of Health Management Information Systems in the health ministry.[8]
  • The Ministry of Health has incorporated mHero into its draft digital health strategy. In December 2015, the Ministry of Health formally integrated mHero into Liberia’s draft Health Information System (HIS) and ICT Strategic Plan for 2016-2021, which upon formalization would make its use an official part of the government’s health strategy and health information system architecture.[9]

Challenges

The piloting and increased adoption of mHero by the Liberian Ministry of Health faced challenges, as did each of the other case studies profiled in this section. Here we provide a closer look at these constraints the lessons that resulted. In subsequent case studies, this section is condensed into a high-level summary.

CommCare Contact Tracing 2.0: Data Exchange between Contact Tracers and the Health Ministry in Guinea

What: In Guinea a consortium of partners, including the Guinean Ministry of Health, piloted and then expanded the use of the CommCare contact tracing tool to gather and share critical case data. An online dashboard provided by Tableau[10] was used to visualize contact tracing data in real time.

Why: Contact tracing was essential to containing the spread of Ebola. In most instances contact tracers used paper forms that were physically transported to a centralized location for manual digitization by a data entry clerk. This frequently resulted in a lag time between when contact tracers collected the case data, and when those data became available for decision-making. This was particularly the case in Guinea where many cases occurred in remote and transient communities.[11] 
 
Who: Partners included the Guinean government, the tech company Dimagi, the Earth Institute at Columbia University, and the UN Population Fund (UNFPA). Data gathered through the contact tracing app and visualized on Tableau dashboards were available for access by government response managers in prefecture-level offices and at the National Ebola Coordination Unit.[12]

A variety of international partners supported the project. Ericsson, the Paul Allen Foundation, and UNMEER donated smartphones; Solektra International donated solar chargers to power phones used in off-grid areas and panels to enable regular access to power in the government offices where CommCare data were used. UNFPA facilitated the purchase of SIM cards for a closed user group and for 500MB of data per month, and the Earth Institute provided computers and USB modems to government staff as necessary.[13]
 
How: A subset of government health staff who supervised contact tracing activities were selected by government and UN agency partners for training on the CommCare contact tracing software application. The app created digital records of the Ebola contact registration form, which captured information such as type of case (i.e., probable, suspected, confirmed), symptoms, GPS location of affected households, and alerts for cases in which individuals were registered with high temperatures or Ebola symptoms. The app enabled longitudinal tracking of individuals over time, an important feature given the need to trace contacts on a daily basis over the 21 days required to determine whether or not a contact had contracted Ebola.

Analysis

  • The CommCare Contact Tracing app was used in five of the eight prefectures in Guinea (Boffa, Conakry, Coyah, Dubreka, Forecariah), after first being piloted in December 2014.
  • ​​The app collected nearly 30,000 contact tracing forms. As of early 2016, approximately 440 contact tracers and supervisors had registered on the tool, of which 281 submitted approximately 19,000 contacts to be traced for location and signs of Ebola, using the CommCare contact tracing app. A total of approximately 280,000 forms were submitted documenting multiple contact tracing visits to the 19,000 contacts identified.
  • Improvements cited in the use of the CommCare contact tracing app over paper forms included: reduction of time and human error, and increased reliability, data verification, and contact tracer accountability. For example, in addition to increasing the rapidity with which frontline health care managers could see contact tracing data being collected, the use of the linked contact tracing app and Tableau data visualization platform supported accountability of contact tracers through the use of timestamps and GPS location on smartphones to enable real-time verification of household visits.[14]Contact tracers don’t visit people who [may be] sick because they are afraid,” one partner said.[15] The use of GPS enabled managers to monitor contact tracer compliance and to identify which contacts received daily visits. One district health manager reported: “The first thing I do every morning is check the dashboard on my computer. The GPS helps me know where my contact tracers have been and where they haven’t been, so I can see if there are issues and follow up.[16]
  • Contact tracers used the digital tool in conjunction with the paper forms; paper forms remained the official, required reporting mechanism during the outbreak.[17]

Challenges

  • A lack of standardized contact tracing protocols and metrics led to confusion about the type of information to be collected and measured. Based on learning from use in the Ebola outbreak response, Dimagi re-released the app as a version 2.0 for deployment in subsequent epidemiological outbreaks.
  • Because users did not have experience regularly using digitized data that appeared at a higher volume and velocity than paper-based data, data collected through the app was not always put to immediate use. Although the app enabled digital data collection, using these data represented a further challenge. An analysis of the tool found that, “While government staff members were especially enthusiastic about the wealth of data that could be used for supervision of contact tracers, actual use of the data was limited. Local government managers often had many competing interests and responsibilities, and thus relied instead on supervisors to report issues to them rather than exploring the data directly."[18]
  • An Earth Institute assessment of the pilot identified numerous challenges, including: availability of local staff to support phone configuration and hardware management; the time required to configure the Tableau dashboard, which was built by volunteer consultants; coordination and management challenges among the diverse array of project partners; limited use of the resulting data by government managers; unclear privacy regulations governing patient data; and the lack of a formal government health information system into which contact tracing data could be integrated.[19]
  • Difficulties around technical issues like battery life and mobile signal. required that contact tracers also use paper forms for back up. “Contact tracers in Port Loko, Sierra Leone, told us they preferred the mobile tool, and felt like the mobile submissions were communicated more quickly to decision-makers acting on the data. The mobile CommCare application also allowed for closer contact tracer supervision, and supervisors could be quickly alerted if a tracer was not making their rounds, or making fewer visits than expected,” said Courtney Kelly of Dimagi.[20]
  • The unique combination of volunteer efforts and donated materials would make it difficult to scale this particular model, although it did demonstrate within the constraints of an emergency response that digitized data could supplement existing paper-based contact tracing data flows.
Connecting Response Organizations and Extension Workers

Ebola Community Action Platform (ECAP): Using Open Data Kit (ODK) and Mobile Messaging to Communicate between Social Mobilizers and NGOs in Liberia

What: To capture and share this situational information in real time, the ECAP program collected KAP data from more than 2 million people in approximately 3,300 communities across Liberia. Over 800 social mobilizers collected KAP data using smartphones donated by the Paul Allen Foundation, and loaded with ODK and the WhatsApp peer-to-peer and group messaging app. To provide broad and open visibility into the changing landscape of community-level Ebola information needs, ECAP visualized aggregated KAP data through a public and easily accessible online platform.[21]

In addition to this digital component, ECAP adopted analog and word-of-mouth mechanisms to amplify and expand its messaging. ECAP used billboards and radio broadcasts across all 15 counties in Liberia, as well as posters, handouts, and person-to-person drama activities at the community level.

Why: Targeted social mobilization among at-risk communities was critical to bending the curve of new Ebola cases in West Africa. To better understand the appropriate formulation of behavior change messages required in Liberia, response organizations needed continuously updated information about community-level knowledge, attitudes, and practices (KAP) about the disease.

Who: ECAP was developed by Mercy Corps and Population Services International (PSI) with contributions from 77 partner organizations, many of which were Liberian NGOs. ECAP also partnered with IREX to reach community radio stations across the country.

How: In the digital portion of the ECAP program, social mobilizers using smartphones completed and submitted community-level KAP survey data on a monthly basis via ODK. In addition, social mobilizers used the WhatsApp platform for peer-to-peer messaging and group learning. In a sometimes unstable telecommunications environment, the WhatsApp platform enabled a layer of message delivery verification (i.e., small checks next to messages indicating whether messages were sent, delivered, and read). Because each of the project’s communities were geo-coded, the dashboard included up-to-date maps that included community KAP data.

For the analog elements of the program, ECAP partner PSI Liberia used billboards to reach people outside of the target communities in which ECAP worked to reinforce community-level messaging. The billboards featured health ministry-approved cartoons, which linked to specific ECAP mobilization topics. For radio, ECAP partner IREX Liberia worked directly with community radio stations to reach listeners in their local dialects, using familiar voices.

Analysis

  • The ECAP program had broad reach through a diverse group of national and international partners. It aggregated and visualized KAP data collected by over 800 social mobilizers who surveyed more than two million people in approximately 3,300 communities throughout Liberia.
  • The program strategically employed a hybrid communications approach that used both digital and low- or no-tech channels to reach the broadest possible audience. Early ECAP data confirmed radio was the most trusted source of information on the unfolding Ebola crisis, and highlighted the reach and impact of community radio stations in Liberia.[22] Billboards and local dramas organized to relay critical behavior change messages also were used to reinforce the community-level work of social mobilizers. Mercy Corps Monitoring, Evaluation, Research and Learning Manager Sophie Roden attributed the success of this multi-partner and multi-channel approach to its focus on a few key consistent messages that stayed the same regardless of partner or distribution channel.
  • The program used both open source and proprietary tools--open source (such as ODK) to serve basic data collection needs and proprietary messaging (such as WhatsApp) to reach a wide group of users at scale.
  • The program embraced a strategic approach to open data by sharing visualizations of aggregated KAP data on a publicly accessible platform for broad use by actors supporting the outbreak response. This created horizontal, peer-to-peer information flows among response actors to increase access to behavior change data and information.
  • The program used peer-to-peer information flows among social mobilizers and created feedback loops that unlocked new incentives. Use of peer-to-peer networking through the WhatsApp platform also produced the unintended benefit of serving as an important source of motivational encouragement for social mobilizers. Mercy Corps learning manager Sophie Roden described this benefit, saying, “Peer-to-peer networks like WhatsApp enabled young people to learn from each other, and encouraged staff to persevere in difficult contexts. Many [social mobilizers] were motivated to go the extra mile to hard-to-reach communities because they could share photos and their experience across the ECAP social mobilizer [network] and receive support from project staff.”
  • The program used peer-to-peer information flows among citizens to share and address common questions. Radio also created an important opportunity for dialogue and debate at the local level. Partner radio stations hosted regular call-in talk shows on Ebola-related topics, with experts—usually a medical professional or Ministry of Health representative—serving as guests. Community members would call in to have their questions and concerns addressed.
  • The program used feedback loops to adapt programming. In addition to gathering a near real-time assessment of critical and quickly changing community information needs, the targeted insights of the ECAP platform and network were used to refine behavior change messages and adapt programming. In the words of a representative of one national partner organization in Liberia, this enabled “discuss[ion] at the national level and among [partner organizations of] what is happening, what are the challenges, and [to] recommend a way forward.”[23]  Roden described how the real-time insights were used to adapt programming, saying, “The aim was to constantly change the messages to reflect what people knew. But we discovered that in the second round [of the learning surveys] that [the social mobilizers] didn’t fully understand the messages in the first round. So we had to reinforce the messages and then add some new material. The third round was better.”

Challenges

  • Despite the rich data that the ECAP program produced, it is not possible to know which aspects or channels of behavior change programming were most essential to turning the tide of new cases. Digital behavior change communications did enable real-time adjustments of behavior change messaging, however, based on changing sentiments and information.
  • As with other programs using digital technologies in the Ebola outbreak response, technical glitches did affect digital aspects of the programming. In particular, mobile data connections were sparse in rural areas resulting in delayed or undelivered messages--particularly in the southeast of the country, where mobile network connectivity gaps were larger.
  • RapidPro, the SMS platform the program initially had intended to use to send messages and reminders to social mobilizers faltered in the early stages of the response. According to Jeff Wishnie, Senior Director of Program Technology at Mercy Corps, “Out of 10 [text messages] sent only a few would get through.”
  • The WhatsApp platform also suffered setbacks, requiring staff time for workarounds. The software client installed on social mobilizers’ smartphones would periodically need to be upgraded, resulting in temporarily discontinuity for affected users. Updating Whatsapp content to receive new messages in rural areas became problematic due to limited connectivity. The WhatsApp message cache[24] would usually only update in populated areas with mobile broadband connectivity, leaving many social mobilizers working in rural environments disconnected from the platform.
  • The use of donated smartphones that were not commonly used locally meant there was a lack of locally available staff trained in IT troubleshooting. "For the first few months when a phone went down that was it," Roden said. In some cases ECAP staff had to troubleshoot themselves. In one instance, when a new version of WhatsApp had to be installed on all 800 smartphones, the team created a video instructing social mobilizers how to download and install the new version of WhatsApp, providing on-the-job-training.
  • The use of a data visualization platform combining data from multiple sources required a strong team of data analytics and visualization experts. This presented a challenge to international partners managing the data during the emergency phase of the response, and is a potential barrier to the program’s longer term success when it is handed over to national partners.
Naymote: Adapting Existing Networks for Community Mobilization

Naymote, an election monitoring organization in Liberia, partnered with Mercy Corps and PSI as part of the ECAP program. Naymote repurposed its network of community organizers, originally created to monitor elections, to support the community mobilization efforts. Staff from the Naymote call center called citizens already part of their network of supporters to inquire about the situation in local communities. They then shared this information with the local county or health offices to mobilize resources. For instance, when they received reports of dead bodies lying uncollected in the streets, they reported this to the Ministry of Health. When rumors were identified, such as one referring to Ebola as a government plot to get money, Naymote staff invited members of Parliament to get on the phone to answer questions. One staff member said, “Community organizing is important for health. Mobile phones enabled us to connect ordinary people who got to talk to people in positions of leadership. This was very helpful in dispelling rumors."[25]

Although phones played a key role in Naymote’s work--indeed, the team also passed along information about toll free Ebola hotlines provided by mobile network operators--the team also adopted multiple strategies to reach communities. For off-grid areas, Naymote held Ebola prevention trainings using a projector transported by truck between communities. Documentaries prepared specifically to support the response, including one with a clip from President Obama talking about the Ebola outbreak to help lend credence to the threat Ebola posed, were also translated into local dialects.

United Methodist Communications: SMS Campaign between Pastors and Bishops in Liberia and Sierra Leone

What: As the Ebola outbreak surged in Liberia and in Sierra Leone, pastors in local United Methodist Churches received “messages of hope”--inspirational Bible verses as well as information about the disease--via text messages sent from their national bishop. The text messages were sent using the SMS broadcast tool FrontlineSMS.[26]

Why: The Ebola crisis touched all parts of society, and required trusted messengers to counter prevailing rumors about the source and attributes of the disease. A pastor based in Sierra Leone explained, “Initially there was a lot of denial about Ebola.”[27]

Who: The bishops’ offices of the United Methodist Church in Sierra Leone and Liberia partnered with United Methodist Communications headquarters, based in Nashville, Tennessee, to send the messages. The partners used an online spreadsheet to craft, edit, and approve messages, which were then dispatched via SMS to pastors and community leaders in each country.

How: Prior to the Ebola outbreak, the church already had in place a list of its pastors’ mobile phone numbers, and the capacity to communicate with pastors via text message. As part of its Ebola outbreak response, the church complemented its SMS campaign engaging pastors with TV advertisements that reached urban audiences, and screened broadcasts using a projector transported by truck that reached off-grid and offline rural communities.

Analysis

  • Between August 2014 and October 2015, over 600 church and community leaders in Liberia and Sierra Leone received more than 550 messages.
  • In addition to providing facts about Ebola, the text messages played an important role in supporting and motivating pastors who played a key role in supporting their communities. One pastor in Liberia’s Kokoya district, which was placed under quarantine during the outbreak, recounted how the text messages gave him courage to persevere in dark moments. In a time when his community faced fear, sickness, loss, and restricted movement, he told a colleague, “When I feel depressed about what is happening around me as a result of the Ebola crisis, I pick up my phone and read the text messages from the very first text message to the end.”[28]
  • Those receiving text messages included pastors whose congregations reached into areas with limited or no access to mobile networks. Particularly in these cases, the pastors’ ability to weave messages into meetings of the congregation served as an important vehicle to deliver information to those individuals beyond the reach of digital networks and devices.
  • The program used a hybrid communications approach, integrating radio call-in shows to enable feedback loops at the community level. Through a partnership with Radio ELUM, a Methodist radio station based in Liberia’s capital city of Monrovia, the church aired select questions in radio broadcasts that featured local experts who addressed these questions. This created a feedback loop by which trusted, local authorities provided accurate information to communities.
  • Trust, and trusted messengers, played a critical role in delivering behavior change communications regardless of communications channel. Phileas Jusu, a Sierra Leone-based communicator employed by United Methodist Communications described the importance of trusted messengers and networks, saying, “We developed messages about avoiding physical contact and used aired broadcasts carried over national TV as a way to suggest a new form of greeting, with arms crossed over chest and bowing. This was a major cultural shift as people were used to touching. But when the bishop went on TV with other religious leaders doing it, it became more accepted.”[29]

Challenges

  • At first the program enabled only one-way, top-down communications. As the program evolved and it became clear that there was demand for two-way communications, the program integrated two-way messaging functionality that enabled pastors to pose questions to the bishop’s office. This revealed important local perceptions about Ebola. One response, for example, asked if Ebola was an American means of population control in Africa.
  • Particularly in a climate of fear and mistrust, pastors at first were wary of messages sent without attribution. Neelley Hicks, director of the ICT4D Church Initiatives program at United Methodist Communications in Nashville said, "We learned right away the importance of adding the bishop's name to the text messages so that the recipients could identify the source of information as a trusted contact."[30]
Connecting Response Organizations and Citizens

DeySay Rumor Tracking: SMS Exchange between Citizens and NGOs in Liberia

What: The DeySay[31]SMS platform in Liberia served to detect and manage rumors in near real time. At the community level, information, including rumors, moved by word-of-mouth, radio, and phone.

Why: Identifying and correcting rumors was a critical part of controlling the Ebola outbreak. In rural villages and urban centers alike, communications among citizens played a critical role in sharing both information and misinformation. Using trusted messengers and channels was essential to correcting misinformation.

Who: Internews, an NGO working to support local media and provide crucial news and information to communities, operated in partnership with the Liberian National Red Cross Society, UNICEF, and Project Concern International.

How:  The DeySay project sought first to understand the state of knowledge about Ebola, and second to improve access to accurate information for citizens by sharing regular updates on rumor tracking with the humanitarian community. DeySay collected information from hundreds of health workers, NGOs, and volunteers in Liberia through an SMS short code, which UNICEF provided free of charge. When those participating in the system became aware of a rumor, they submitted it via text message to the short code, which relayed the messages to a central coordination hub in Monrovia. There, the project team analyzed rumors for content and location trends, and developed new messages to counter misinformation and misunderstandings. These insights and new messages were shared with local media partners for rapid dissemination in the target area via radio, community engagement, and other channels. DeySay also produced a weekly newsletter updating the landscape of prevalent rumors and their geographic locations, offering official response organizations and local media alike a way to identify and respond to misinformation about Ebola at the community level.
 

DeSay Newsletter Screenshot

Screenshot of DeySay newsletter

Analysis

  • The digital component made it possible for organizations to quickly analyze and respond to rumors. One organization partnering with the DeySey network received several messages regarding the Ebola treatment units it operated. In one instance, the organization received feedback that community members living near a treatment unit were concerned about the smoke rising from an incinerator, giving rise to a rumor that bodies were being burned there. After learning of the rumor, the organization removed a tarp that had blocked the community’s view of the incinerator, enabling passersby to see that the organization was burning contaminated materials, not bodies. In other instances, community volunteers were able to investigate, understand, and then address rumors related to higher mortality rates in specific treatment units.[32]
  • The ability to rapidly adapt behavior change messaging in response to rumors represented another advantage of the program’s digital component. For example, during a vaccination program in schools a rumor in rural northern Liberia near the Guinean border broke out that the vaccinated children were taken by ambulance and hospitalized.[33] The rumor originated after a child in a public school in the area was diagnosed with a high fever. When the ambulance arrived to transport the sick child for care, students panicked and fled, and within the hour parents throughout the region were rushing to local schools and removing their children. To counteract the rumor, the false information was communicated to local radio stations, which broadcast the reported rumor and clarified the situation. In an added step, local health teams made school visits. As a result, parents returned their children to school and the rumor was stopped before it could spread and do further harm.[34]
  • Though it is not always possible to produce immediate and uniform behavior change, this case study provides a good example in which a hybridized approach combining the use of digital, radio, and word of mouth communications did work to quickly address a rumor by connecting to people through communications channels with broad reach.

Challenges

  • An initial attempt to establish an independent short code for the DeySay program was abandoned in favor of a partnership with UNICEF using a short code they had previously negotiated, in particular due to difficulties associated with creating a single short code that would be accessible to subscribers across multiple mobile network operators. The UNICEF partnership also granted the DeySay program use of their SMS aggregator (RapidPro). This did not enable direct access, however, which meant the DeySay project team could not themselves do troubleshooting or modify workflows in the system. Similarly, UNICEF held the relationship with mobile network operators, limiting the ability of DeySay program team to ask questions or request support.
  • When the DeySay project team arrived in Monrovia, rumors were rampant and trusted information was in short supply. “To determine the most trusted sources of information we used an information ecosystem model,” explained Anahi Ayala Iacucci, Internews Country Director in Liberia.[35] Through a partnership with Geopoll, the Internews team sent monthly SMS surveys to around 200 citizens in each of five regions. The surveys gathered regularly updated responses about the most trusted sources of information, whether government, media (including radio, TV, and print), civil society, or other channels--and showed that though trust in other channels varied, local radio remained one of the most trusted sources of information throughout the height of the outbreak. The surveys also asked people what information they needed. “Listening to what people need speaks to the principle of ‘design with the user,’” Mark Frohardt, Internews Senior Vice President for Design and Learning, said. “When dealing with complex issues like Ebola, you have to start with what information is most critical to the local population, not with what you want them to know about your service. A clear understanding of needs and context increases the amount of requested information and understanding of messaging from services providers,” he added.[36]
  • Many social mobilization efforts during the response used broadcast media as a one-way channel to blast information out, rather than partnering with local media to leverage radio programming’s ability to host a dialogue with the community. Internews maintained a behind-the-scenes role for the duration of the DeySay program, engaging 16 media partners in convenings that explained how the DeySay rumor tracking system worked, gathering journalists’ feedback, and then letting each media outlet decide whether and how to use the information. Three to four weeks later, most media partners were running dedicated radio programs using the DeySay information to address local rumors. The Internews team also provided journalists with available information that could be used to counter rumors. By the end of the program, nearly 200 media stations had opted to participate.
  • One challenge the program faced was how to discuss rumors without inadvertently further spreading them. Internews conducted trainings for journalists to address this, and to provide critical context to the disease. “We brought journalists to Monrovia and a training with the World Health Organization and the Ministry of Health to explain how diseases like Ebola work,” said Iacucci. “The journalists loved it; nobody had taken the time to explain it to them, yet they had the job of trying to explain to their audiences. The training was designed to try to take complex epidemiological and medical terms and break them down in a way that would be relatable and understandable,” she added.[37]

U-Report: SMS between Citizens and a UN Agency in Liberia

What: U-Report, an SMS platform designed to facilitate citizen engagement and community-led development, engages youth “U-Reporters” to provide information about their local context and, in return, provides information back to U-Reporters that they can use to work for change locally. The tool is designed to increase awareness of local needs, and transparency and accountability in development programming and services.[38]

Why: Lack of trusted information about Ebola was a major factor contributing to the disease’s transmission. To inform behavior change and social mobilization campaigns, response organizations needed to understand local knowledge, attitudes, and beliefs about the disease.

Who: The UN Children’s Fund (UNICEF) created the U-Report platform in 2011 and was using it at scale in five sub-Saharan African countries prior to the Ebola outbreak. When the outbreak struck, UNICEF leveraged an existing U-Report deployment in Nigeria where over 100,000 youth were already engaged, and set up new systems in Liberia and Sierra Leone.[39] In Liberia, UNICEF partnered with the Federations of Liberian Youth (FLY), active in all 15 counties in Liberia, to help recruit U-Reporters.

How: The U-Report platform used a short code provided by the Liberian Telecommunications Authority[40] and linked it to the open source RapidPro text messaging platform.[41]Young people opted in, and received polls and information via this free short code. Through U-Report, UNICEF aggregated real-time information about young people’s perceptions and concerns about Ebola and visualized responses on a public website to help inform behavior change efforts. U-Reporters were recruited in partnership with FLY through the use of social mobilizers, targeted campaigning, mobile vans, in-person meetings (e.g., town halls, churches, schools, mosques), as well as radio advertisements and promotion via call-in radio shows across 21 radio stations in Liberia.[42]

Analysis

  • Between November 2014, when the U-Report program was launched in Liberia, and June 2015, almost 51,000 youth had registered in Liberia to become “U-Reporters.”
  • In Monrovia, UNICEF used a human-centered design approach by partnering with representatives of the target audience, in this case teenage girls, to script survey questions querying knowledge, attitudes, and sentiments about Ebola. As a result of this collaboration, originally drafted survey questions such as “What has changed the most in your community because of Ebola?” were adapted to a more colloquial format used by texting teens (e.g., “wat bother U d most abt Ebola?”) In addition, a steering committee of youth representatives worked with UNICEF to identify weekly topics for the campaigns.
  • During the crisis, UNICEF’s Communication for Development behavior change teams used U-Report data to support community engagement work, in some cases to adapt and reposition some of UNICEF’s interventions and messages. The platform also was used to assess the effectiveness of radio campaigns to alert people to new outbreaks.[43]
  • In the recovery period, UNICEF used the platform to query sentiments and gather knowledge about related topics, such as school reopenings. The target audience was expanded to include teachers, women, and religious leaders.[44] Beyond the outbreak, U-Report remains an important platform for youth and citizen engagement. It is now being used in Guinea to address other public health issues, such as HIV/AIDS, as well as gender-based violence and women’s financial empowerment.[45]
  • UNICEF had experience using the U-Report platform during a prior Ebola outbreak, where it illustrated the important “sensing” value that citizen-led communications channels offered. In July 2012, an active U-Report deployment in Uganda began receiving messages from U-Reporters asking about that outbreak. Messages included:
    • “IN OUR AREA THERE’S A PARSON DIPHICATING BLOOD? FEELING FEVER ISN’T THIS SIGN OF EBOLA P’SE HELP”
    • “Our district Kibaale has been affected by an unknown disease but doctors are suspecting it to be Ebola disease”[46]

Challenges

  • Getting the U-Report short code up and running took time and multiple negotiations. Particularly in the context of the hectic early stages of the Ebola outbreak response, obtaining a short code for the U-Report project was difficult. Multiple organizations were making requests of mobile network operators, overwhelming the staff of those companies that already were strained by the toll of the outbreak and the demands of the response. UNICEF ultimately went to the Liberian health ministry to secure support for the short code.
  • Even once the short code was established, there were delays in rolling out the short code across multiple mobile network operators. To work around the varying timelines on which mobile operators were able to begin making the short code active on their networks, the UNICEF team took a phased approach and started operating the short code on networks as they became available, rather than waiting to launch the short code simultaneously across all networks.

Social Mobilisation Action Consortium (SMAC): Using Open Data Kit (ODK) and Mobile Messaging to Communicate between NGOs, Social Mobilizers, and Communities in Sierra Leone

What: The Social Mobilisation Action Consortium (SMAC) launched to collect and transmit information about the Ebola outbreak in ways that allowed communities to engage and ask questions. SMAC members recruited, trained, deployed and worked with community mobilizers, religious leaders, traditional healers, radio stations, and community leaders to provide targeted and accurate information to community members. In addition, SMAC members developed a process to receive information about sick cases and deaths in communities as well as other community-level information. The process evolved to run off mobile devices.

Why: As the outbreak escalated in Sierra Leone, it became apparent that a gap existed between coordination of the Ebola outbreak response and social mobilization interventions at the district level. This resulted in a lack of understanding about what was happening in affected communities, and what people were saying and doing about the disease.

Who: SMAC is an alliance of national and international NGOs and other partners, including GOAL, Restless Development, FOCUS 1000, BBC Media Action, and the CDC. All had been involved in the response before coming together as a consortium. For example, FOCUS 1000 had begun to conduct KAP surveys in August 2014. These surveys allowed greater understanding of the local perceptions and response to the Ebola outbreak, supported the design of more appropriate social mobilization interventions, and enabled tracking of knowledge, attitudes, and behavior changes across the response. FOCUS 1000, in partnership with CDC and UNICEF, conducted four KAP surveys during the course of the outbreak. The latter three were conducted using mobile devices.

Although all members collaborated on social mobilization activities, each organization brought niche specializations to the consortium’s work. GOAL and Restless Development led the work with community mobilizers (many of whom were also community health workers) using the Community-Led Ebola Action approach. FOCUS 1000 provided grassroots connections to influential community leaders and led the work with religious leaders and traditional healers. BBC Media Action linked SMAC to and led the work with local radio stations. The CDC provided technical support.

How: Although SMAC’s initial community-based reporting used paper forms, with funding from the Bill & Melinda Gates Foundation SMAC was able to move to a digital data collection system and expand the reach and type of data collected. The initial paper-based forms were streamlined into the digital system, with added questions and indicators. According to Mohammed Jalloh, former Senior Program Manager for FOCUS 1000 and lead on SMAC’s digital data collection system, “The digital work built off of what we had done. It was never about the new cool technologies. Many people find the cool software and then figure out how to use it. For us it was the opposite--these are challenges and this is how technology can help us solve them.”[47]

For the project, SMAC built a unique, customized Digital Data Collection System (DDCS) using a combination of software--ODK, KoboCollect, and Textit. The system allowed SMAC’s vast network of stakeholders to submit actionable Ebola-related data from thousands of targeted, sometimes remote, communities across the country and facilitated two-way daily communication with all phone users.

Social mobilizers (including community mobilizers, religious leaders, and radio station managers) used mobile phones to collect daily alerts about suspected Ebola cases and any deaths in the community. These data were submitted through SMS into the DDCS system, which generated an automated response with an appropriate course of action (e.g., notifying the nationwide “117” hotline or the local hotline for reporting cases) and allowed for follow-up (e.g., SMAC could follow up with the hotline or to report back to the community). The daily alerts were shared with district-level officials tracking the disease.

Second, the mobile phones allowed for ongoing reporting from social mobilizers who submitted weekly performance data about SMAC activities, community perceptions of Ebola, and information about burials and deaths in their communities. The digital technologies SMAC used allowed project staff to customize the surveys for each different group of social mobilizers.

Analysis

  • The SMAC project provided access to phones and training on the Digital Data Collection System to over 2,500 people. Fifty master trainers, 1,475 community mobilizers, 1,036 religious leaders, and 36 partner radio stations (in pairs or individually) received access to phones and were trained to use the DDCS and send real-time, community-based surveillance data, including information about deaths and suspected cases.
  • SMAC compiled and shared weekly reports with key local and national response actors that highlighted concerns and feedback from communities and provided updates on SMAC’s community engagement in the priority districts.
  • Accountability and reporting back to communities represented one major benefit of the daily alert process. SMAC staff suggested that prompt actions helped to save lives and strengthen trust between the communities and service providers. They reported on and used examples of actions that were delayed to advocate for service improvements. According to Katharine Owen, the SMAC Director, “We had concerns about establishing a parallel system with the daily alerts. But we discovered people might say they called 117 when actually they didn’t. But they still wanted to tell someone. Or we might find [the case] had been called in, and there was nothing happening. … Or where it had been reported, our people on the ground were able to report back to communities what was happening. … Whether the case was in the [formal case data collection] system or not, there is value in getting the information into the system or in being accountable and reporting back. This is important to maintain trust.”[48]
  • SMAC used their digitally connected network of social mobilizers and other community leaders to help assess the efficacy of messaging during a nationwide “stay-at-home,” during which citizens were encouraged not to leave their homes. During the stay-at-home, the government conducted a House-to-House campaign designed to educate people about Ebola and conduct disease surveillance. The second stay-at-home included a nationwide plan to distribute hand soap, but officials ran out of soap. Using its network, SMAC gathered and analyzed community perceptions (over 40,000 records) of the second lockdown and reported their findings to the NERC. According to Owen, “The data were able to demonstrate that if you promise everyone soap and not everyone gets soap, that is what they remember rather than the message given.”
  • Reusing and adapting survey software and analysis tools enabled significant savings in time and effort. FOCUS 1000, a SMAC member, conducted four KAP surveys in Sierra Leone, including three with mobile devices, that supported the overall Ebola response efforts. In addition to the KAP surveys in Sierra Leone, members of the FOCUS 1000 team trained another local NGO in Guinea, Sante Plus, with support from the CDC, to conduct a similar national KAP survey in Guinea in August 2015. Previous KAP work from Sierra Leone allowed the Guinea team to reuse and adapt the surveys and the analytical frameworks, saving time and effort. Using programmed tablets, the Guinea KAP data were collected from over 6,000 respondents across all eight administrative regions in Guinea within one month. The collected data were then swiftly analyzed with pre-programmed statistical analysis code, and key findings presented to the Guinea Ebola Coordination Center within three days after receiving the data.

Challenges

  • Despite the benefits, SMAC staff reported several challenges in deploying digital technologies. First, connectivity presented issues. The mobile phone network would break down, and the technical assistants for the mobile network operators were based outside of Sierra Leone. This caused delays for SMAC activities. It also took several months to set up the short code and functionality of the system to provide voice and data connections to all social mobilizers in the program. During the project, SMAC switched to a locally based provider to speed up the troubleshooting process, but then had to migrate the phones in a phased fashion to maintain services.
  • SMAC staff identified the “human element”--the training and support needed to take advantage of digital technologies--as another challenge. “We underestimated the human resources needed to do real-time data,” reported Jalloh. Staff indicated that despite the goal of using real-time data to inform decisions, this was difficult. Jalloh noted, “The use of the system during the House-to-House campaign to get feedback from communities was perhaps the best illustration of real-time usage of the system wherein findings were shared daily with the NERC to inform decision-making.”
  • Another issue was cost. SMAC’s activities required significant up-front financial resources to purchase the mobile phones and pay for recurrent costs associated with the data plans necessary for the surveys and the short code for the alert system. In addition, there were “hidden costs” in the form of hiring people to enter data and space to house staff members. Finally, SMAC staff spoke about the challenges of managing the devices. This included training people to use them, providing technical support when the devices broke, and decisions about what to do with the devices after the program ended.
  • SMAC staff reported that the amount of data represented a significant issue. For the daily alerts, they hired “alert interns” who helped to manage the data. The interns were able to follow up to collect more detailed information to put into a dashboard, which was shared with district-level officials.
Connecting Other Actors

Skype Information Management/Geographic Information Systems (IM/GIS Group): Online Coordination and Data Exchange among Response Workers (Global)

What: The Skype IM/GIS channel was established in late August 2014 by the Digital Humanitarian Network (DHN) to meet this need.[49] A variety of actors involved in health and humanitarian aspects of the Ebola response coordinated and delivered geospatial data sets to meet information management needs shared using this channel.

Why: Particularly during the early days of the 2014 Ebola outbreak, response organizations identified a need for detailed mapping of health, operational, and other information. Unfortunately, available or actionable data sets were limited, and there was no widely accepted, formal channel to request or share geolocation information.

Who: Members of the DHN, a network of the digital humanitarians who leverage digital skills to support humanitarian responses,[50] created the Skype IM/GIS channel to facilitate coordination between formal humanitarian aid actors and this network of digital experts. According to Roxanne Moore, who served as DHN Ebola coordinator from December 2014 through June 2015, “The Skype group was an online group that was created organically. Prior to that, there was not an open space to collaborate. … It was very chaotic.”[51] Participants in the Skype IM/GIS group included DHN members as well as staff of a variety of national and international organizations.

How: The channel served as a clearinghouse for requests for geographical and other information, and the tasking of creating new datasets and maps needed to support the response. This included mapping previously unmapped areas in Ebola-affected countries, and creating overlays such as: the location of Ebola treatment units and community care clinics; the population density of affected areas; border crossing points between affected countries and border crossing data; road blockades and quarantined areas where passage was restricted; and Ebola case and death counts by locality.[52] Many of these datasets were later shared on the Ebola GeoNode or HDX.

Analysis

  • By August 2015 over 230 individuals from approximately 100 organizations had joined the Skype IM/GIS Group, some in their official capacity and others as off-hours volunteers. Up until May 2015 the group continued to host regular data exchanges, although increasingly about broader health and other needs that accompanied the affected countries’ recovery after the crisis phase of the Ebola response.[53]
  • The IM/GIS channel supplemented existing, formal coordination mechanisms, and was used by many, including response workers in country, staff of international organizations working abroad, and members of the digital humanitarian community around the globe. This “open” model of ecosystem engagement enabled relevant actors to self-identify and self-organize around information needs largely outside of the context of formal mechanisms for information sharing by the response community.
  • One of the primary advantages of the open Skype group was the ability to connect a range of remote and operational responders, often across multiple sectors of activity (e.g., health, technology, GIS). According to Moore, “Having such a large Skype group is a huge deal. We’ve never had anything this large. You had the U.S. government talking to NGOs talking to VTCs [volunteer and technical communities] talking to UN agencies. The de-siloing of [information sharing among] organizations is exciting.” In this way, the Ebola response illustrated changes in how actors communicated.
  • DHN members engaged more with each other and with a broader range response actors than in prior crises in which they had been deployed, due in part to OCHA’s limited operationalization in the response, a result of the creation of UNMEER. This represents both a maturation of the digital humanitarian community and the longer timeframe of the Ebola response.[54] It also highlights the opportunity for the formal humanitarian and health communities to think more holistically about how to embrace open models for ecosystem engagement, including with the digital humanitarian community, in its work moving forward.

Challenges

  • Although numerous links existed between the IM/GIS group participants and formal response mechanisms and actors, a significant disconnect existed between formal and informal communications channels. This meant that some valuable information was lost.
  • The disconnect between communications on this channel and formal response mechanisms, and the lack of a widely accepted clearinghouse for operational data contributed to gaps in awareness of existing tools and duplication of effort. For example, several interviewees involved in the formal response were unaware of the mapping efforts of the digital humanitarians or of the existence of the Skype IM/GIS group.[55]
  • The health aspects of the Ebola response posed a new set of issues, particularly around terminology and data privacy. “A health response is difficult, since we needed to learn how to de-identify health information, learn new terminology, and account for time length [of the response],” reported Moore.
Digital Networks: Internet-Based Exchange of Real-Time Contextual Information

Context experts with deep and long-term expertise in the languages, cultural traditions, and social, economic, and political histories of Guinea, Liberia, and Sierra Leone and with expertise in public health and emergency response contributed substantively to the social engagement and behavior change efforts. These individuals included social scientists--particularly cultural and medical anthropologists--former Peace Corps volunteers, and members of the diaspora.

Many of these experts were based elsewhere or unable to travel to the affected countries yet wanted to contribute. They mobilized networks using email, websites, Skype, listservs, and other Internet-based platforms to provide advice from remote locations in a timely way and with the clear goal of supporting a contextually relevant response.

Notably, many UN agencies and NGOs embedded anthropologists in their emergency operations teams. UNMEER hired an anthropologist to work across the response and affected countries, and other agencies, including the CDC and MSF, hired behavioral and social scientists to work on social mobilization efforts.[56] In Guinea, at least four anthropologists were embedded in treatment units to assist with psychosocial efforts and to conduct research. The UK Department for International Development (DFID) funded research, policy-focused blogs, and a web-based platform--the Ebola Response Anthropology Platform--that aimed to facilitate a more effective, iterative, and adaptive response.[57] A similar network, the Emergency Ebola Anthropology initiative was established in the United States.

The expertise of these individuals was crucial in helping to ensure a contextually appropriate response and in countering the resistance that Ebola teams faced in affected communities. For example, in the early days of the response, burial teams used black body bags, whereas in Muslim communities in these countries the color white signified mourning.[58] The use of black body bags subsequently changed with feedback from these experts and affected communities, leading to the use of culturally appropriate white body bags. According to one health official, among other things the expertise of anthropologists and other behavioral scientists shifted understandings about the burial practices of different groups, and facilitated more clarity and precision in communications with affected communities. As a result, he said, “Messaging changed and became more adaptive and sensitive.”[59]

The contextual and qualitative knowledge supplemented much of the epidemiological data used in the response. Juliet Bedford, the anthropologist who was embedded with UNMEER and then UNICEF, explained how she engaged networks of anthropologists: “I was asked specific questions [from responders] and would get those out to the network. We crowdsourced information and, working with a few key people, created a rapid synthesis and turned it into two-page briefs. We provided focused papers with key considerations. … It was about getting information to the right people at the right time for strategy meetings.” In addition, members of the remote networks wrote to her with questions and suggestions for responders, based upon their contextual knowledge. For example, did responders consider the effect of the rains on social mobility, and hence the disease’s transmission, or the use of roadblocks in a post-war context? She would then flag these issues for the response’s senior leadership, at both the global and country levels. The network produced a series of short two-page papers in English and French on topics of relevance for responders (e.g., handling bodies, mobilizing youth, clinical trials, and community resistance).[60]

Humanitarian Data Exchange (HDX): Online Data Exchange among Response Organizations (Global)

What: The HDX, an open-source platform designed to serve as a central repository for both public and restricted humanitarian data sets, was under development when the Ebola outbreak struck.[61] Although the tool was new, it became a repository for many of the data sets collected during the response.

Why: Although many organizations collected extensive amounts of data as part of the Ebola outbreak response, many of these datasets were not shared or easily accessible. According to Sarah Telford, manager of the Humanitarian Data Exchange (HDX), the goal of the platform was to “try to change the culture of guardedness around information sharing ... and to solve the problem of making [response] data more immediately accessible.”[62]

Who: OCHA launched HDX in July 2014, just prior to the official WHO declaration of a Public Health Emergency of International Concern.

How: OCHA managed the HDX platform, inviting individuals to register through their organization to become “data contributors.” Contributors, in turn, were responsible for providing “metadata” (e.g., when and where data were collected, and by whom) and for keeping their datasets up to date. Due to organizational policies or other privacy concerns, some organizations used a privacy setting to enable data sharing only among their staff. UNMEER, for example, requested and used a closed portion of the platform. The majority of HDX data sets were available in machine-readable formats that enabled the aggregation, analysis, and visualization of the data, facilitating the use of data for making decisions.

Analysis

  • Although use of HDX occurred primarily in limited pockets, the platform did amass a significant number of datasets and a considerable usership. According to one official, “HDX has helped a lot as a very visible platform. People can see in one place, here’s the data.”[63]As of March 2016, HDX had nearly 60 public data sets, and 23 private data sets related to the Ebola response, contributed by 29 organizations. Those data sets have been downloaded nearly 18,000 times, and HDX users have downloaded the vast majority (72 percent) of data sets at least 10 times. Of those more commonly accessed data sets, 30 have been downloaded at least 100 times, and 3 have been downloaded over 1,000 times. Those downloaded over a thousand times were maintained by HDX, OCHA’s West and Central Africa office, and UNMEER. Two contained epidemiological data and the other contained data about Ebola treatment centers.
  • To fill the data gap about the location of Ebola treatment units, in August 2014 volunteers scanned the Internet looking for these data. Simon Johnson of the British Red Cross geo-coded these treatment units and visualized their locations on a dashboard.[64] Once these secondary data were posted to HDX as open data, other organizations provided feedback, including more precise location data. For Liberia, the U.S. Department of Defense added detailed plans of the proposed location and anticipated opening dates for new treatment units. In this way, opening these data and posting to a central location improved the accuracy of the data and prompted new uses for the data.

Snapshot of Ebola 3W dashboard, by Simon Johnson. 
Simon Johnson, “Ebola 3W Dashboard,” December 10, 2014, accessed September 9, 2016, https://simonbjohnson.github.io/Ebola-3W-Dashboard/.

  • The platform made it possible to gather datasets from a variety of sources and contributors. These included formal response organizations (UNMEER, WHO, OCHA’s West and Central Africa Offices, NGOs, and the U.S. Department of Defense) as well as digital humanitarians (the Humanitarian OpenStreetMap Team and Standby Task Force).[65]
  • A pilot program involving data collection about IPC in Guinea enabled tracking of IPC training efforts across response actors. OCHA and HDX, in conjunction with the Guinean Ebola coordination center, USAID, and IPC partners, launched the pilot in order to respond to a senior Guinean government official’s inquiry about how many individuals in Guinea had received training in IPC. Using a participatory process to identify key indicators, the group agreed on a common matrix using an Excel spreadsheet with pre-populated indicators and dropdown options (e.g., location data and health facilities) to minimize the time required to update the spreadsheet and data entry errors. The spreadsheet was posted on Dropbox, allowing broad access to all partners. Partners updated their information on a weekly basis and HDX analyzed and visualized the data.

Challenges

  • A lack of established standards made it difficult to compare and compile different data sets. “The lack of standards makes it so that you can’t create a common operational data picture based on multiple sources and types of data,” explained Telford. “You can only look at it data set by data set,” she said. In addition to needing data in common formats, response actors also needed data in a common language--such as by standardizing the use of whether a treatment unit was documented as a clinic, a hospital, or a mobile health facility. The Humanitarian Exchange Language (HXL), also created by OCHA, developed a shorthand version for language-based standards by adding a row of hashtags to help make terminology compatible across data sets (e.g., #gender to refer to a column of data about the gender of beneficiaries). Yet because HXL was not widely used by actors publishing data sets to HDX, it was difficult to merge multiple data sets, such as in adding a health facility layer to a map illustrating the availability of Internet coverage.
  • HDX had launched in beta form just prior to the Ebola outbreak, so it was not widely known or used at that time. “There is an exhaustion in our community [from seeing too many] unsustainable platforms that become data graveyards,” said Telford. “HDX was a new platform, so there was a little bit of hesitation [where people wondered if] this was another one-off. That prevented people from wanting to share their data,” she added. The HDX team sought to address this concern by being responsive to data providers’ needs, including helping data contributors ensure the metadata[66] was entered correctly, and in providing troubleshooting support in getting data sets uploaded to the platform.
  • There is not yet a widely established culture of data sharing in the humanitarian community. Since the practice of publishing machine-readable data sets to shared access platforms is still relatively new, the mechanisms or incentives did not yet exist for data producers to publish appropriately anonymized data sets.[67]Yet for those groups that did contribute data, there were multiple examples of shared, consumable, open data sets that created value. In one example, the New York Times used the machine-readable Ebola case and death data set (available on HDX) to create an interactive map that was featured on the homepage of the news outlet’s website for a day. This is an example, Telford believes, of data use by third parties to reach audiences that data contributors alone never could have reached.
  • A dearth of data literacy among humanitarians meant that in addition to streamlining requested data, HDX staff needed to provide capacity building by proactively engaging organizations to train them on how to use the platform as well as to prompt them to input their information, requiring significant staff time. “We’re a narrative-based culture, [so] this was a real challenge,” Telford said. According to another official who worked on a data collection effort posted on HDX, “I realized the problem is a capacity issue. The ones that were regularly contributing data had a focal point who is responsible and accountable.”[68] The need for stronger data literacy was especially acute among national staff in the most-affected countries in West Africa, creating a heavy reliance on international partners that slowed the use of data to support the response.
Missed Opportunities

The vast majority of data and information exchange in the Ebola outbreak response happened in the context of verbal, radio, and paper-based communications. In this context, how does one assess or test the value of digital technologies in the Ebola outbreak response? Providing evidence of the value of a missed opportunity would require measuring chains of information transmission that, due to a variety of constraints, simply did not exist in the response. Two sets of insights emerge from the response.

Rumor boxes in SL

These boxes, in Freetown, Sierra Leone, contain documentation of rumors related to the Ebola outbreak.

Insights from Paper-based Information Flows

To understand where information did not flow, it is useful to ask: Who did not receive information that could have been beneficial to the response? In this analysis, an obvious starting point is to consider the most widely used one-way vertical flow of information “up”: case data reporting.

As detailed in the Data Use and Digitization section, case data largely flowed one way and up from local to national and international decision-makers. An NGO official in Sierra Leone observed, “[We had] a very top-down, vertical approach. We can’t be just vertical.”[69] To realize the value of digitized information, vertical communications should be seen as the start, not the end, of an information exchange. Ideally communications should entail two-way information exchange (both “up” and “down”), as well as horizontal exchanges among peer groups, and benefit from communication among multiple nodes.

What if, for example, aggregate case counts, once gathered at the top of the information pyramid, had been regularly and deliberately shared back “down” to health workers reporting the data? “One thing we’re emphasizing is flipping the information flows so that they are community-driven rather than top-down,” one government official said of efforts to strengthen the response.[70] One can imagine how contextualized data returned to the point of data origin--such as a rise in Ebola caseload in a neighboring district--might have empowered health workers with important situational awareness that could have improved local preparation and decision-making. Where, for example, could behavior change communications efforts have been best focused to help stem the spread of the disease? And where might critical supplies like personal protective equipment, need to be pre-positioned? Although in some cases such exchange did happen, it was not widespread or consistent across the response.

Separately, many interviewees observed how the vertical, upward flow of information often occurred without clear understanding on the part of data collectors about how such data would be used. The example of the misuse of unique identifiers described above illustrates how a lack of contextualized understanding of data use negatively affected the quality of data and further affected the delivery of services. A lack of contextualization also reduced incentives for data collectors to understand the purpose and potential impact of the data they collected. One national official articulated how reversing the flow of information could help to create ownership and a culture of data use, beginning with those collecting the data: “If you want to engender a culture of data collection and reporting, and improv[e] the quality of the data, you must provide reasons why that data should be collected. ... Let [data collectors] understand that data is their data. … If you expect that person to report just counts--number of cholera cases, malaria, etc.--let them be able to compare if it was higher this month compared to last month. If it is higher, what’s the reason, what can I do in the facility to ensure it’s not high next time. ... Even at county level they’re just sending it forward.”[71]

This contextualization of data for data collectors, such as frontline and other health workers, would require a fundamental paradigm shift in the way health data are collected, analyzed, and used. Perhaps more aptly, it would require a paradigm shift in the larger practice of global public health for decision-making by expanding the role of local health workers from agents of local data collection and service delivery to agents of local decision-making, who are empowered with contextualized insights in real or near-real time.

Insights from Digital Information Flows Beyond the Formal Response

Even within the context of digitized data and information flows, there were missed opportunities to maximize the value of digitization. Some of these missed opportunities--including the lack of interoperability between digital systems, minimal use of machine-readable forms, and the lack of networked digital data-entry systems--were identified earlier in this report.

Other examples from outside the formal response delivered by governments, donors, and response organizations also show where digitized information flows added value, but were not fully leveraged in the response. Formal and informal responders adopted widely used communications enabling many-to-many communications--platforms like Google documents, Skype, and WhatsApp--to exchange information with a variety of individuals experiencing the crisis or participating in the Ebola outbreak response. Yet due to organizational and institutional restrictions on the use of communications platforms,[72] many did so acting in their individual capacity. The sharing of information within these channels, therefore, occurred informally and outside of official coordination channels.

Because knowledge about these channels spread via social networks and word of mouth, many response actors were unaware of the existence of these channels. The disconnect between communications on these channels and formal response mechanisms, and the lack of a widely accepted clearinghouse for operational data, contributed to gaps in awareness of existing tools and duplication of effort.

At the global level, staff members of government, UN agency, NGO, and other response actors joined a Skype chat group, often using personal accounts, to identify and share common geographic and other Ebola response information management needs. As detailed in the Skype IM/GIS Group case study, this group functioned as an informal (i.e., non-authorized) Skype chat group that created horizontal, and multi-directional information flow across a range of actors. At its peak, the Skype group included a total of 232 active users from 92 organizations. Requests for information ranged from GIS data on health facilities in the most-affected countries and Ebola case data to population distributions and movement restrictions or blockades.[73]

At the community level, United Methodist ministers communicated with one another and other members of their community directly through a self-organized WhatsApp channel. These horizontal, two-way information flows created feedback loops that enabled the immediate sharing of local and time-sensitive information. In at least one case this led to directly actionable insights that likely saved lives with the rerouting of an Ebola patient from an unprepared clinic to a hospital that was able to provide treatment.[74]

The use of these informal channels demonstrates the value of real-time, peer-to-peer horizontal information sharing among actor groups, such as responders and community leaders, and the breadth of the individuals’ networks sharing information relevant to the operational response. Where insights were exchanged on these channels outside of the context of the formal response, critical insights frequently were lost.

Lessons

The real-time information flows profiled above suggest a series of lessons from the Ebola outbreak response, and for health and humanitarian preparedness and response more generally.

  • The use of digital technologies in the response enabled access to more timely data across large distances and diverse actor groups. This was particularly critical in the context of gaining rapid access to case data and to behavior change communications both of which were critical to containing the disease.
  • Digital enabled peer-to-peer, horizontal messaging at scale, and the transfer of data and information “up” as well as “down” and “up and down” through feedback loops. Yet many digital programs did not use digital technology to its full potential in this respect, frequently because tools and platforms leveraging this functionality were launched just prior to the outbreak (e.g., mHero, HDX) and not used at scale during the course of the response.
  • Creating a two-way, real-time exchange between the central health ministry and its workforce enabled a significant transformation in the response. It enabled the government to benefit from a real-time sensing capacity, and to provide health workers with context to the data they produced, creating new insights, incentives, and accountability.
  • Health workers played a critical surveillance role in detecting disease outbreaks and their spread, and when empowered with digital technologies they could report these data in a timely way. Yet even where they had access to mobile networks, health workers lacked basic digital literacy or the trained enumerator skills necessary to gather survey data.
  • The introduction of digital technologies does not remove the need for human capacity and leadership; indeed, in many cases, it magnifies it. The Ebola crisis exposed critical gaps in the human capacity needed in the outbreak response. This capacity gap slowed the outbreak response and hindered the ability to rapidly gather, share, and use digitized data flows for decision-making.
  • Specifically, the data demands of the outbreak response highlighted the need for greater investments in a cadre of data experts with a wide variety of technical skills, including software engineering, data management, analytics, and visualization expertise, and workflow integration. This human capacity gap was most acute at the national level, but apparent among international responding organizations as well.
  • The demands of the response made it difficult to build national capacity while meeting operational needs, emphasizing the importance of building this capacity in preparedness for future emergencies.
  • In the context of largely paper-based health data systems at the national level, there was little established practice or demand for leveraging real-time insights from digital data. This meant that national actors frequently were unprepared or underprepared for the integration of digitized data both from a policy and a process (including workflows and decision-making) perspective.
  • The most-affected countries possessed fledgling digital health systems that, in most cases, were built to address specific diseases or aspects of public health delivery (e.g., HIV/AIDS or reproductive and maternal health) rather than an integrated health systems approach. This siloed and vertical orientation of information was a barrier to a holistic picture of public health data and the ability to make rapid, data-driven decisions during the outbreak response. This will continue to impair health outcomes until standalone systems are better integrated.
  • Some of the most successful uses of digital data and information flows were those that took a hybrid approach, incorporating word-of-mouth, paper-based, and analog channels alongside the use of digital technologies. Programs such as Naymote or DeySay were able to reach both online and offline communities, and yielded rich, locally relevant data.
  • Interoperability between digital systems can unlock powerful advantages, but takes time and requires careful attention to standards and terminology in addition to technical platform integrations.
  • Other challenges facing digitized data and information flows included a lack of standards, or lax adherence to standards in an emergency context, limited use of machine-readable data, and a lack of networked digital data entry systems.
  • Many digital health programs received a surge of attention and funding due to the resources unleashed with the declaration of a public health emergency; most times these resources are one-off. Ongoing program needs will require continued donor support and/or new business models to be sustainable over time. In some cases, the program was possible only due to significant donation of in-kind resources that are likely not replicable outside of the context of an emergency.

References

[1] Case studies explored in this report are those that surfaced through the interview process and, as mentioned in the Methodology section, are neither comprehensive nor fully representative of the diversity of tools used. These case studies in this section were selected because they illustrate different types and flows of data and information in a digitized, real-time or near real-time environment. We do not investigate paper-based or analog information flows in this section.
[2] In its report, Health Worker Ebola Infections in Guinea, Liberia and Sierra Leone ( WHO/EVD/SDS/REPORT/2015.1 (2015): 1-15. World Health Organization, 21 May 2015, http://www.who.int/hrh/documents/21may2015_web_final.pdf), WHO estimated the risk of infection at “between 21 to 32 times higher in health workers compared with non-health workers."  Of note, and as an example of the varying figures documenting the response, the CDC's October 2015, Morbidity and Mortality Weekly Report (MMWR) on the Ebola outbreak in Guinea indicated that "incidences of Ebola infection among healthcare workers [were] 42.2 times higher than among non-HCWs." See Margaret Grinnell et al, "Ebola Virus Disease in Health Care Workers--Guinea", 2014. Issue brief no. 38. Vol. 64. Morbidity and Mortality Weekly Report. Centers for Disease Control and Prevention, 02 Oct. 2015, accessed March 28, 2016, https://www.cdc.gov/mmwr/preview/mmwrhtml/mm6438a6.htm.
[3] IntraHealth and K4Health’s participation in the program was supported by USAID funding. In particular, IntraHealth’s accelerated work on mHero as part of the Ebola outbreak response won the support of USAID’s the Fighting Ebola Grand Challenge led by the Center for Accelerating Innovation and Impact (CII) in USAID’s Global Health Bureau and supported by USAID.
[4]
 Digital health is the use of any digital technologies to enable better collection and use of data, improved quality and reach of health service delivery, and better decision-making by governments, health workers and individuals. When used in accordance with best practice, such as that embodied in the Principles for Digital Development, digital health can help governments better understand and respond to public health needs, and empower individuals to make better choices to improve quality of life for themselves and their families. Digital health comprises the domains of eHealth and mHealth, and includes the adaptation, use, and support of digital technologies including wireless technologies (e.g., cellular, GPS, WiFi, Bluetooth), mobile devices (e.g., mobile phones, tablets, laptops, SIM-enabled devices, “smart” medical devices), sensors and embedded technologies (i.e. the Internet of things), as well as fixed digital devices, equipment, and infrastructure (e.g., desktops, servers, fixed broadband)
[5] See www.ohie.org for more information about Open Health Information Exchange.
[6] The iHRIS database was under development in Liberia and Guinea when the Ebola outbreak struck.
[7] These 40-plus deployments are in addition to the identification of 33 "use cases," or distinct purposes for mHero deployments.
[8]  As cited in Emily Nicholson, “Spotlight: Liberia’s Great Strides to Scale the mHero Platform,” IntraHealth, USAID, and UNICEF, February 2016, 3, http://www.intrahealth.org/files/media/spotlight-liberias-great-strides-to-scale-the-mhero-platform/mHeroSpotlight3.pdf.
[9] Emily Nicholson, “Spotlight: Liberia’s Great Strides to Scale the mHero Platform,” IntraHealth, USAID, and UNICEF, February 2016, 2-3, http://www.intrahealth.org/files/media/spotlight-liberias-great-strides-to-scale-the-mhero-platform/mHeroSpotlight3.pdf.
[10] For more information about Tableau and the Tableau Foundation, see http://www.tableaufoundation.org/.
[11] Jilian A Sacks et al., “Introduction of Mobile Health Tools to Support Ebola Surveillance and Contact Tracing in Guinea,” Global Health: Science and Practice 3, no. 4 (2015): 646, doi: 10.9745/GHSP-D-15-00207, http://www.ghspjournal.org/content/3/4/646.full.pdf+html 
[12] Sacks et al, Introduction of Mobile Health Tools to Support Ebola Surveillance and Contact Tracing in Guinea 3:647.
[13] Sacks et al, Introduction of Mobile Health Tools to Support Ebola Surveillance and Contact Tracing in Guinea 3:651-652.
[14] Sacks et al, Introduction of Mobile Health Tools to Support Ebola Surveillance and Contact Tracing in Guinea 3:646-659.
[15] Interview with international responder, May 2015.
[16] Conversation with Guinean district health manager, July 2015.
[17] Sacks et al, Introduction of Mobile Health Tools to Support Ebola Surveillance and Contact Tracing in Guinea 3:646-659.
[18] Sacks et al, Introduction of Mobile Health Tools to Support Ebola Surveillance and Contact Tracing in Guinea 3:646-659.
[19] Sacks et al, Introduction of Mobile Health Tools to Support Ebola Surveillance and Contact Tracing in Guinea 3:655-658.
[20] Email correspondence with authors, August 2016. As noted above, all interviewees are cited anonymously to protect confidentiality, except in relevant instances, such as these case studies that identify organizations thereby compromising anonymity. All case studies have been cross-checked with relevant actors, including permission to identify individuals by name.
[21] The ECAP case study is compiled from interviews with Sophie Roden, Jeff Wishnie, and Michael Catalano, April and May 2015. Also email correspondence with authors, July and August 2016 and remarks by Sophie Roden during the “OpenResponse ICT Webinar” hosted by Dimagi and examining lessons learned by eight partners providing ICT-based services in support of the Ebola outbreak response, and held online on December 3, 2015. See also http://www.ecapliberia.org/.
[22] Mercy Corps, “National Radio Reach Survey - A study on radio and ECAP 2 health messages in Liberia,” (Portland, OR: Mercy Corps, 2016), accessed September 21, 2016, http://www.ecapliberia.org/images/pdfs/USAID2015LiberiaECAPIIRadioSurvey.pdf.
[23] Interview with staff from national NGO, February 2016.
[24]  A cache is a temporary data storage area that enables Internet browsers to retrieve information about past data downloads from specific websites. This is particularly critical in instances in which users move between on- and offline environments.
[25] Interview with Naymote staff members, February 2016./h6>
[26]  For more information see http://www.frontlinesms.com/
[27]  Interview with Julu Swen, United Methodist Communications communicator in Liberia, September 2015.

[28] Remark by Reverend Benedict Greene (District Superintendent, Kokoya District) shared with Julu Swen, United Methodist Communications communicator in Liberia, and relayed via interview on September 17, 2015. Quote from Rev. Greene used with his approval.
[29] Phileas Jusu (Sierra Leone-based Communicator, United Methodists Communications) remarks at the Game Changers Summit in Nashville, September 17, 2015.
[30] Neelley Hicks, e-mail message to authors, May 31, 2016.
[31] The name “DeySay” is derived from colloquial Liberian English referring to how people speak about rumors.
[32] Interview with Anahi Ayala Iacucci, Senior Advisor for the Internews Center for Innovation and Learning, February 2016.
[33]  Anahi Ayala Iacucci, “Combating Rumors About Ebola: SMS Done Right,” Internews, March 26, 2015, accessed September 21, 2016, https://medium.com/local-voices-global-change/combatting-rumors-about-ebola-sms-done-right-da1da1b222e8#.1zqmx5y4i
[34] Anahi Ayala Iacucci, “Combating Rumors About Ebola: SMS Done Right,” Internews, March 26, 2015, accessed September 21, 2016, https://medium.com/local-voices-global-change/combatting-rumors-about-ebola-sms-done-right-da1da1b222e8#.1zqmx5y4i
[35] Interview with Anahi Ayala Iacucci and Mark Frohardt, Internews, September 2016.
[36] Interview with Anahi Ayala Iacucci and Mark Frohardt, Internews, September 2016.
[37] Interview with Anahi Ayala Iacucci and Mark Frohardt, Internews, September 2016.
[38] UNICEF, “Communication for Development: Responding to Ebola in Liberia,” accessed June 25, 2016, http://www.unicef.org/cbsc/index_73157.html.
[39]  Rebecca Levine et al., “mHealth Compendium,” African Strategies for Health 5 (June 2015): 20, https://www.msh.org/sites/msh.org/files/2015_08_msh_mhealth_compendium_volume5.pdf. For an example of U-Report’s use during the Ebola outbreak response in Sierra Leone, see http://sierraleone.ureport.in/poll/431.
[40] UNICEF Stories, “UNICEF develops critical Ebola data flow,” originally published in the New Dawn on October 27, 2014, http://www.unicefstories.org/2014/10/28/unicef-develops-critical-ebola-data-flow/. See also http://www.huffingtonpost.com/2015/03/05/u-report-liberia-ebola_n_6802048.html.
[41] "RapidPro: An Open Source Communication Platform,” UNICEF Innovation, accessed May 7, 2016, https://community.rapidpro.io/. RapidPro is an open source version of TextIT.
[42] UNICEF, “Communication for Development: Responding to Ebola in Liberia,” accessed June 25, 2016, http://www.unicef.org/cbsc/index_73157.html.
[43] Interview with Rafael Obregon, December 2015.
[44] UNICEF, “Communication for Development: Responding to Ebola in Liberia,” accessed June 25, 2016, http://www.unicef.org/cbsc/index_73157.html.
[45]  Interview with UNICEF officials, February 2016 and email communication to authors, July 2016.
[46]  Maria Luisa Sotomayor (Member of the U-Report coordination team), email communication to authors, May 6, 2015.
[47]  Interview with Mohammed Jalloh, February 2016.
[48]  Interview with Katharine Owen, February 2016
[49]  The DHN was launched in 2012 to support information sharing between formal aid actors and the informal digital humanitarian community. This was expanded in 2014 for the Ebola response through DFID support.
[50]  The DHN was created in 2012 by Andrej Verity of OCHA and Patrick Meier, a founder of the “crisis mappers” community. For more information, see “DH Network,” Digital Humanitarian Network, accessed June 3, 2016, http://digitalhumanitarians.com/.
[51]  Interview with Roxanne Moore, April 2015.
[52]  Information sourced from an analysis prepared in summer 2015 by interns working with Benson Funk Wilder, geographer/analyst at the U.S. Department of State.
[53]  A month-by-month analysis of data and information exchange on the Skype group is available in an unpublished Excel spreadsheet managed by Roxanne Moore of DHN.
[54]  Interview with Roxanne Moore, April 2015.[55] Interview with international responders, January and February 2016.[56] Interviews with anthropologists and NGO officials, November and December 2015, February 2016; see also World Health Organization, “Anthropologists Work with Ebola-affected Communities in Mali,” Features, January 2015, http://www.who.int/features/2015/anthropologists-ebola-mali/en/.
[57]  This platform, coordinated by anthropologists at the Institute for Development Studies, Universities of Sussex and Exeter, and the London School of Hygiene and Tropical Medicine. See “Ebola”, Institute of Development Studies, accessed May 31, 2016, http://www.ids.ac.uk/idsresearch/ebola and “Recent Addition,” Ebola Response Anthropology Platform, accessed May 31, 2016, http://www.ebola-anthropology.net/. This platform addressed the following issues: Identifying and diagnosing cases; management of the dead; caring for the sick; clinical trials; preparedness, and; communication and engagement.
[58]  Lisa Cobb, “Anthropology and Ebola Communication,” Ebola Communication Network, January 6, 2015, http://ebolacommunicationnetwork.org/anthropology-and-ebola-communication/.
[59]  Interview with health promotion official, November 2015.
[60]  Interview with Juliet Bedford, November 2015. The papers are downloadable from http://www.anthrologica.com/unmeer-west-africa.html
.[61] "The Humanitarian Data Exchange,” HDX, accessed May 31, 2016, https://data.humdata.org/
[62]  
Interview with Sarah Telford, January 2015. This case study is compiled from interviews with Telford in January 2015 and September 2016, and with David Megginson, also with HDX, in February 2015.
[63]  Interview with USG official, February 2015.
[64]  Simon Johnson, “Ebola 3W Dashboard,” December 10, 2014, accessed September 9, 2016, https://simonbjohnson.github.io/Ebola-3W-Dashboard/.
[65]  Email correspondence with Javier Teran (HDX), March 2016.
[66]  Metadata provides contextual information about other data, such as when those data were collected, by whom, and any dates on which data were modified.[67] HDX will not, for example, accept any data sets that include personally identifiable information.
[68]  Interview with UN official, February 2016.
[69]  Interview with NGO official in Sierra Leone, February 2016.
[70]  Interview with national official, April 2015.

[71]  Interview with national health official, May 2015.
[72]  Mikel Maron and Julia Kim, “Crisis data coordination: Lessons learned from the Ebola data calls hosted by the U.S. Government,” Unpublished report, 2015 (Washington: U.S. Department of State).
[73]  Roxanne Moore, “After action report: IM/GIS Ebola communications on Skype,” Digital Humanitarian Network, August 2015.
[74]  Interview with international responder, September 2015
.