2016 Global Digital Health Forum, December 13-14

Interactive Session: Data Visualization Tools for CHW Supervision to the Last Mile

The Digital Health Knowledge Base - Content, and Taxonomies, and Platforms, Oh My!

OpenHIE: Global to Local

Session presenters and description

Session focused on Tanzania’s OpenHIE work.

Customizing global solutions to address challenges which have developed organically.

Governance will be needed for each registry to maintain on an ongoing basis.

Types of registries

Immunization registry

Client shared registry

Facility registry

Terminology service (associates/maps codes for vaccines/products)

HMIS (DHIS2)

Health Worker Registery (not clear if this was there)

Data capture

• Tablet (at immunization date)

• Paper scan (at immunization date)

• SMS (at birth)

• Website (stock mgmt)

Henry Mwanyika, PATH

Tanzania - want to build on the current initiatives, solve duplicate efforts (entry and sources), quick decision making from different levels. What is the future? To have information at our fingertips and integrate existing systems.

Ed at INSTEDD will talk about the facility registry

Health facility portal (made by a local team).  Can download the data (made it public)

Hfrportal(facility portal)

Planwise (is used for planning supply replenishment)

Carl Leitner, IntraHealth

Immunization record, client registry

Connectathon with Dlab in country to build capacity of local developers. Sounded like a really good idea for creating buyin and understanding.

Derek Ritz

• BID design work with the immunization record

• Centralized the security measures - PKI certificates (only authenticating services)

• Centralized audit of messages moving between systems

Role playing and acting out a baby's birth.

1. Register the village elder (uses facility register

2. Register the child - Rapid Pro registration

• Once the village elder is registered, then all 'children' born there take on that information

• Needs to be entered in a specific way for appropriate registration

1. Vaccination nurse uses tablet based application to record the immunization (Tanzania health record with a barcode)

1. Scans the barcode to indicate the vaccine was given

2. Enter weight (if in range, ok, if not give guidance)

3. Enter all vaccines

4. Other information (deworming, etc.)

Rebuilt paper forms so a machine can scan and record transactions (versus typing into a tablet)

Scanning saves time and allows for the immunization registry to be used for care management versus recording what happens.

Brian, talking about VIMS

Transaction traffic is standards based…

Interoperability layer (need to check with team on status of VIMS)

• DHIS2 led for ADX standards

• Care service discovery

• Mobile alert message standards

Ministry of health of ICT created the health facility registry

New Resources from the World Health Organization (WHO)/Johns Hopkins University Global mHealth Initiative (GMI) Partnership

Session presenters and description

Dykki Settle from PATH elaborated on “A Global Good” — a tool used by numerous countries and funded by numerous donors; easy to implement; adaptable to country-specific needs; well-documented; “packaged for accelerated re-use”; easy to scale; interoperable with other Global Goods. He cited DHIS2 as an exemplary global good. If the tools you are using are not supported, not global goods, then your investment is at risk.

Audience examples: OpenMRS, OpenLMIS, master facility lists, the MAPS Toolkit, the OpenHIE blueprint, DHIS2 (especially its ability to scale)

Garrett Mehl from WHO unveiled a new resource, the Digital Health Atlas. It is a global atlas of digital health implementations around the globe. For VillageReach and our OpenLMIS work, it offers a way to monitor the trends in different digital tools. In the countries where VillageReach works, it may help us improve coordination and find new partnerships. The Atlas is a new resource that is starting out empty and it is now open and ready for anyone to log in and post the work they are doing. 
Next Steps 
→ VillageReach and OpenLMIS should consider posting the implementations/projects we have underway into the Digital Health Atlas.

Alain Labrique from Johns Hopkins shared a new joint publication Monitoring and Evaluating Digital Health Interventions. Tigest Tamrat also shared in-progress WHO Guidelines. And Garrett Mehl promoted a taxonomy.

Broadband Commission Digital Health Report

See session presenters and description

Ann Aerts from the Novartis Foundation shared a preview of the upcoming report:

• Not enough countries have a national eHealth/digital health strategy (and not enough have electronic medical records or a telehealth strategy either)

• Country Case Studies will include Nigeria and Mali, Canada and others

• Findings: (1) it requires sustained senior government leadership and financing; (2) having effective cooperation with clearly defined roles is critical; and (3) a strategy for ICT.

• The report will be released around March 1, 2017 at the Mobile World Congress.

Enabling Interoperability

See session presenters and description

Katherine (Katie) Healy from Johns Hopkins presented a pilot of mobile health interventions for infant vaccine reminders and scheduling. They also created a digital vaccination register that resembled the previous paper registers and piloted these in Bangladesh. They chose OpenSRP to scale this system to the national level. They also leveraged DHIS2, Tableau, CouchDB, and other tools. They created a standardized “concept dictionary” to store standard terminology and allow data sharing. They created this concept dictionary in OpenMRS.

MVP-CIEL is the default dictionary in OpenMRS. It has 60,000 existing concepts and maps to SNOMED, RxNorm and other standards. MVP-CIEL handles concept requests and additions. It allows translations to be gathered.

Hermes Sotter R___ from the Tanzania Ministry of Health ICT team spoke in stead of the JSI person who could not attend. He described the health landscape in Tanzania and showed a map of 120 different software systems that the Ministry is currently using! But less than 1% of their 7000+ health facilities have electronic medical records. Hermes described their main challenges: Fragmented ICT pilots and numerous HIS silos; a lack of interoperability and data sharing; and of course fragmented business processes and challenges with stakeholder coordination. A JSI colleague, Francois, talked about eLMIS as well.

Alex Little from Digital Campus moderated the questions. Key take-aways:

• There are major challenges around different ID numbers in different systems, EG in electronic medical records versus records in a national birth-and-death system.

• The transition from paper to electronic records is difficult, but health workers do see that it saves time, reduces duplication and improves organization and use of the information.

Mapping a Data Use Culture

OpenLMIS Session

Audience Examples of Customization/Extension Points:

• We talk to our External ERP using a web service instead of using FTP.

• In Zambia there are 2 organizations that manage warehouses. Both have different warehouse management systems.

• Rather than sending an order to a central warehouse, perhaps other facilities nearby would provide those.

• If the requirements of the facility are not in a standard pack size, does that logic happen at the requisitioning facility or at the warehouse? Answer: Mary Jo talked about pack size, global product codes, trade items and the ReferenceData service.

• An outside vendor to the Ministry of Health provides only two products, and that outside vendor has certain providers of those products. How does OpenLMIS support that scenario? Mary Jo described. He asked if that would be overwhelming with hundreds of products sourced from different vendors? Mary Jo explained there has not yet been a need for this.

Audience Questions:

• What is the tech stack? Answer: Open source using Java and Postgres.

• Moving to micro-services gives you extensibility but adds sophistication (CI/CD, testing, etc). How are you planning to address those challenges? Answer: We have set up tooling and environments for ourselves, had the technical committee identify risks, and document it well.

• Where are on-hand balances (stock balances) in the diagram? Answer: Mary Jo explained that service is not yet built. Others in the audience explained that Stock On Hand data is also gathered during the requisitioning process.

• This is a process that is prompted by external stimuli—so is there any support for predictive ordering based on trends? Answer: There are calculations for average consumption that can help in this regard.

• The process looks like the Supply Chain Operations Reference Model (the SCOR). SCOR is a body of knowledge around supply chain. Ashraf explained that OpenLMIS does follow aspects of the fulfillment process from that model, but it also requires fitting it to the context of the low- and middle-income countries. Another JSI member explained that each country has different Standard Operating Procedures.

• If a country is not following best practices, it’s useful to understand that. If we bake the best practices into the tool (OpenLMIS) it would help to educate and inform around the best practices.

• A PWC/JSI person also explained how they did use SCOR.

• Wendy with JSI explained how the Coca-Cola model did not apply at the conclusion of a process/study about supply chain. She also explained the goals of eLMIS to identify in-country best practices and move from paper to electronic automation. This allows continuous improvement that is not possible on paper.

Lessons Learned for Sustainability

Session presenters and description

Jeanne Koepsell from Save The Children (and also a conference organizer) introduced the session.

Julio Pires from Jhpiego Mozambique and Manuel Macebe from the Mozambique Ministry of Health presented lessons from a Mozambique HRIS implementation. They call their implementation eSIP-Saude. It covers health worker staff records and connects with the payroll system. They had to work across ministries (MoH with civil service and finance ministries). They have a national state employee registry called eCAF. They added health-specific requirements to eCAF. They host the eSIP-Saude in a data center within the country. They showcased dashboards for decision-making and BI.

Julio stressed why they pursued collaboration with other ministries rather than doing their own system: they really wanted to leverage existing infrastructure and ensure long-term ownership of the project and the involvement of key players. Collaboration was definitely harder—there were different priorities and different “work rhythms” and methodologies to bridge—but hopefully the long term sustainability makes it worthwhile. He credits their success overcoming this to “win-win approaches”, clear SLAs and NDAs, and alignment with the objectives and interests of all actors. Plus four helpings of patience, according to Julio. He cited the advantages including a faster national roll-out and decreased operational costs. Julio suggests beginning the collaboration at the technical level, but formalizing it at the higher level (inter/ministerial).

Erica Layer from D-Tree International presented about the Safer Deliveries program in Zanzibar which works on maternal and child health. Key to program success was closely supporting community health workers and their supervisors.

• The program registered pregnant mothers and allowed community health workers to easily pay for their transportation to get to a health facility when they went into labor.

• Technology tools were built on CommCare for a simple mobile app.

• But there were challenges in maintenance and sustainability of the program. It has been hard to maintain high levels of engagement when the program scaled up.

• They switched to a new technology platform, MangoLogic, so their non-technical staff can easily maintain it without engaging expert programmers. They also simplified the UI.

• Erica showed a sophisticated mentorship and pre-training system so that “champion” health workers get trained first and help other health workers get on board. The champions also help target CHWs who need extra training and in-person follow-ups.

• The team also created a supervisory system within the mobile app that tracks attendance at monthly meetings, with GPS coordinates and timestamps from each of the staff plus the supervisor. It also provides reports to see activity of each CHW and of the supervisor.

• The team also created a pay-for-performance incentive for the CHWs and supervisors.

• Now they are transferring responsibility to the MoH and reducing the NGO involvement. They have an MOU to make this official. They are also connecting the citizen feedback survey data that feeds the pay-for-performance metrics that give more money to the facilities that have better feedback and results. Wow!

• They are also now creating a program to help the pregnant mothers save money during their pregnancy for transport and for other delivery costs. Wow!!

Telling Stories with Data Visualizations

see session presenters and description

Aly Azhar from VaxTrac shared a Benin case study. VaxTrax is a non-profit with projects in Benin, Nepal and Sierra Leone. Their project, VaxTrac Monitor, created a data dashboard that merges data from clinics up to the district level. Aly showed a Google Map-based dashboard that color-codes each facility to identify which have submitted immunization records recently (green) versus which have not (yellow or red). This map helps MoH staff and supervisors target their follow-up. Aly encourages iterative testing and feedback of dashboards to tune and refine for the needs of users. It was also a challenge to get users to incorporate it into their regular workflow; even though they were initially excited, usage fell off and follow-up was required.

Technology-wise, they use an Android app on tablets at health clinics. Data syncs to CommCare.

Jeff Bernson from PATH spoke about how visualization of data improves use of data. He works in malaria elimination, and is working in southern Zambia on an initiative called “Visualize No Malaria” with Tableau and the Zambian government. He says “Surveillance is the next malaria vaccine”. Their dashboards help to show what facilities might have missing data. They also created a map visualization to show how community health workers are connected to their community health facilities. Showing this to people on a map helped them see missing data and data errors—it helped all stakeholders realize the importance of engaging district staff to manage their master data. Currently they are experimenting with weather data and maps of where water flows and ponds. By working with MapBox they have started extracting structures from maps and imagery and overlaying all this together. These layers and model help the district staff begin to think about resource allocation.

Technology-wise, data is collected on feature phones and flows into DHIS2. They have worked with Tableau to make dashboards and also used Twilio to automate SMS alerts. “Does an SMS count as a data visualization?” For him, it was just as important to get SMS messages out, because the dashboards themselves would not be seen by the specific workers where it mattered.

Mandy Dube from PATH also shared stories from the BID initiative about why it is important to make the health worker the hero.

Data Moves at the Speed of Trust

Current landscape:

• Technology changes

• No simple rules to follow

• Culture change around data and digital development

• Real perceived privacy & security risks

• Current practices are ad hoc, highly variable, and not standardized

• Different evolving legal landscapes (how to deal with non-european/usa citizens where there aren't any frameworks)

• Limited guidelines from donors

Data lifecycle (see photo)

• Data capture (integrate outside data sets)

• Data processing

• Data distribution and release

• Data archiving

• Data analysis

Risks (see photo)

Understand, Apply, and Expand on DHIS2

Session presenters & agenda

Nicola Hobby - DHIS2

DHIS2 Demo site

Shows full range of available DHIS2 offerings

play.dhis2.org

Login: admin

PW: district

Types of data collected: Aggregate data, events, tracking (equipment, drugs → ?)

Why is DHIS2 successful?

• Developed IN the developing world (thought it was developed at U of Oslo?)

• Low IT literacy skills needed (thought it was somewhat complicated to set up?)

• Rapid scalability

• Supported by grassroots community building efforts via academies (Steffen said academies are being restructured)

• Flexibility

• Open source (Nicola mentioned that this may/may not contribute to success. Customization, training, long-term and ongoing maintenance have costs). Initiation barrier is low - no fee, free to download)

• Adoption at critical mass

Data integration driver

Can have separate instances of DHIS2 (example DRC MSH, MOH, MSH headquarters), integration driver will pull data from source to destination automatically and in a scheduled fashion.

DHIS2 Symposium

March 23-24, 2017

US-Based, University of Olso-sponsored

Bringing together variety of use cases. Used to be NGO-specific, why not open to country use-cases. Interoperability as theme. In DC, FHI360 center.

CommCare Case Studies

See session presenters and description

Lisa Noguchi from Jhpiego showed off a mobile app project to create a tool, powered by CommCare, that helps collect data points to estimate the gestational age in order to better estimate the date of delivery for babies.

Sarah Hodsdon from Dimagi showed a project in 5 provinces of Mozambique. Their goal was to identify patients and keep them in monthly treatment for 6 months. Their workflow connects CHWs with case managers at the clinic level. The app is built with CommCare and includes data collection and also the delivery of video content.

Building on Local Capacity and Sustainability

See session description (presenters are wrong there)

Presenters: Kathy Gettelfinger, ThoughtWorks; Sean McDonald, FrontlineSMS; Darlene Irby, Palladium; Merrick Schaefer, USAID; Steven Wanyee Macharia, IntelliSOFT; Rebecca Saxton-Fox, USAID.

Rebecca Saxton-Fox from USAID announced the creation of WAHIT, the West Africa Health Informatics Team. They are starting a local hiring process now so they can begin to support local HIS tools starting in 2017.

→ We should explore if we can collaborate with them in an OpenLMIS deployment to get local dev staff.

Merrick Schaefer, USAID, described his experience trying to hire local IT/software talent in-country in Africa. He felt it was a “seller’s market”. He also argued that hard skills totally mattered over soft. He also described a large network of Tech Hubs in Zambia, Lagos, and more.

Steven Wanyee Macharia, IntelliSOFT, described boot camps and partnerships with universities that are helping to develop more talent.

→ Connect with Steven by email to get introduced to some of the University of Dar folks, and UW and UCSF folks he is working with. They are implementing Bahmni for EMR needs in Zambia and work in Tanzania and other countries in the region.

Sean McDonald, FrontlineSMS, has run a program for 6 years in Nairobi. He argues for a long-term strategy. They offer paid internships and work closely within the community.

Successes:

• Long-term strategy with a local on-the-ground presence (see Sean’s comments above)

• Encourage staff to go in the field and see the users they are impacting

Failures:

• Hire talent and then lose them quickly. Had multiple developers lose interest (even good developers who contributed modules to OpenMRS).

• Hire the wrong mix of hard skills and soft skills (with the wrong expectations for the person)

• A mailing list is not enough to connect the developers to the global community

• Hard to find good UX and Design skills

Lessons:

• Be clear on what skills you want. Just tech skills? Or communication abilities? Kathy suggests that it is easier to teach the hard skills than teaching the soft skills. Darlene agreed that personality and curiosity are critical, and no arrogance. It’s also important to make sure candidates are willing and interested in traveling to rural areas.

• It really depends on where the pipeline of work is coming from. Is there existing funding or are you expecting the in-country staff to help advocate for the value proposition?

• Pay fair and competitive wages.

• Culture within the office is a huge incentive. It’s also not costly to do it well.

Hindsight is 20/20: Lessons from Mobile Technology for Community Health (MOTECH) program in Ghana

Session presenters and description

Anitha Moorthy, Grameen Ghana, described the MOTECH project with its messaging (Mobile Midwife) and data collection components.

David Hutchful, Grameen Ghana, described the technology infrastructure. Once a nurse and client meet, data is captured on a feature phone to enroll the client into MOTECH. They may be added to campaigns that provide SMS or Voice messages each week to the client. There are appointment reminders and IVR (interactive voice response). Data feeds into OpenMRS to store all of the actual health and clinical data.

Amnesty LeFevre and Larissa Jennings from Johns Hopkins presented about the assessment findings and lessons learned.

Tim Wood, now with the Gates Foundation, explained that this project started in 2008 and faced many challenges that are familiar as the global health landscape has evolved: technology, governance, sustainability and more. He specifically noted the lack of early buy-in from Ghana Health Service (the MoH).

Issues:

• Technological: limited capacity of the phone lines (2 lines for calling 42,000 patients, mostly between 5-6pm!)

• Less than 2/3rds of expected messages were received by intended recipients!

• Active listening (to more than half the message length) decreased over time. Why!?