Symposium Program: https://www.dhis2.org/symposium2017-programme

Presentations Link:  https://drive.google.com/drive/folders/0B6ablhPQZu2bY1M5UXBHT0YyTWc

Notes by section

Key Follow ups :

• Follow up with Patrick on reviewing and participating in drafting use cases for the DHIS2 Implementer Guide
• Follow up with contacts in Guinea about the roadmap and OpenLMIS version 3 features

Day 1 - Thursday, March 23

Opening Remarks - Nicola Hobby

University of Oslo - National engagement in DHIS2 Projects - Knut Staring

Knut - provided a good background to the system, focus on integration with other systems, staying open. Referred to logistics systems as a system to integrate with and not do.

• Aggregate Data Exchange (ADX) for direct system to system integration
• OpenHIE (health information exchange) act as national backend resources for all 'front systems'. Recommends a gradual approach given the complexity. Essentially use one system as the facility registry until the full registry is implemented? Not sure each system needs a server and support team for each registry. There are ways to reach improved integration gradually.

Looking forward

• integrating surveillance (between social media and health systems) across the region, not just nationally
• integrate case level data in some places and aggregate in others
• How to sustain systems on the ground - encourage partners to support personal growth through:
  • Academies, mentoring, regional experts, academic scholarships 
  • Equally important to collaborate on human resources 
• Interoperability Implementer Guide: Supported by UNICEF/GAVI
  
  • Target group: Implementers, decision-makers 
  • Guidance on system choices
  • Success stories
  • Contributions to OpenHIE discussion 
• Online living implementation guide - not yet developed, may live in a wiki - haven't yet decided on format CONTRIBUTIONS WANTED! How do we get involved with this? Follow up. 
  • Concepts of integration
    • eg. OpenHIE
    • Integration vs. interoperability 
  • Step-by-step
  • Functional and disease specific topic pages 
• When asked on how to approach sharing data, Knut advises on first publishing meta data. Then slowly working towards sharing more and more data.
• Is there going to be one national database for HMIS data? Countries would like this but not NGOs.
  • Interoperability is one way of approaching that - allows organizations to keep their autonomy while still encouraging collaboration. 
• Question: Have you seen a leadership organization (ex. PEPFAR) having success in moving forward conversations on integration? 
  • One hurdle to changing national systems: Have printed many paper forms, spent $ on printing, distributing and training - to change that will take a few years. 
  • Digitizing and using tablets and mobile phones makes it technically feasible, but it goes back to what the organizations want to do. 
  • Have seen a few countries where key donors have managed to pull together MOH, WHO, and other key stakeholders (example, CHAI, malaria consortium) into a room and get them to collaborate. Slow process. 

FHI360 Nigerian national DHIS2 implementation

Nigeria FHI360 Office (570 million dollar budget)

• GF, USAID, BMGF, UNICEF are the donors
• have projects in all states within Nigeria

Actors

• Government is driving
• Implementers are navigators (direction from donors) but there to support the government. Big role in advocacy
• Donors
• UN agencies

Data governance over the HMIS falls to the statics department (DPRS) on paper but in practice it is more complicated. There are three national DHIS2 instances (see slides for the diagram). Great example of parallel reporting.

Reporting % are up in the 60 - 70%.

Key Challenges

• parallel reporting systems (drain on data capture) due to 
  • multiple stakeholders, 
  • multiple program/disease areas
  • multiple KPIs
  • varying reporting schedules PEPFAR is before National reporting process
  • ownership, want more instances to 'own' the entire process and control

How to solve it?

• Single platform
• Tool harmonization (included a 2 year review process)
• Migration to DHIS2 for all program (outside of HIV)
• 2013 focus multi level data reporting (using mobile)
• Committee to focus on data quality and integration

Roles of implementing partners

• usually where there are NGOs the reporting rates are higher

Data Use

• one focus is on reporting rates (data capture/operational) - usually the first step. Likes to see performance versus other districts to promote improvements.
• quarterly newsletters using charts and graphs

Future

1. integrated health data management
2. data quality initiatives
3. EMR/dhis2 interoperability (scale up EMR use to ease the burden of data capture)

E8 - Southern Africa Development Community - Malaria Elimination 8

EAC - East African Community 

ICAP Headquarter and Country Integration

ICAP
Laura Lincks

Why they chose DHIS2

• Community support
• Open source
• enhanced visualizations
• APIs
• Simplified organization of data elements  

DHIS2 Informs Management of Results-Based Financing in Tanzania

Broad Branch Academy
Jenna Wright

USAID 

Results Based Financing (RBF)
Cyclical process, verification of results determines payment
Different from input-based financing  

Public Sector Systems Strengthening (PS3) Activity 

Tanzania is rolling out a national RBF system

Still in development - will be in 9/30 regions within the next year 

Multiple actors, requires coordination, management, adaptability and DATA

RBF runs on data 

• number of services delivered at HFs (reported)
• number of services delivered at HFs (verified)
• Results of facility quality assessment 
• Results of council and regional performance assessment
• Invoices (derived) 

DHIS2 critical to successful roll out and management

TZ has adapted the DHIS2 system to meet the RBF functions 

DHIS2 developers in-country still working to adapt the system 

• Data quality still poor (incomplete, erroneous, manual entry) 
  
  • Ongoing data cleaning efforts. Unable to replicate payment invoices at a later date. 

Breakout 1: Data integration technical - DHIS2

Configuration of systems are often not harmonized. Things which can differ :

• Mismatch between national systems, NGO systems, donor systems 
• Metadata codes - indicators
• organizational hierarchy levels
• Age groups

Problem 1 - Metadata codes are not harmonized between systems 

DHIS2 - Identifier schemes. Can add multiple codes to an individual element 

Maintenance application - configuration 

Attributes 

Data Element: HIV_CARE

Changes parameter and code to CARE_CURR 

Q: Can all attribute IDs be in one JSON dump?
A: Yes 

Data Collection Frequencies

How to take monthly data and turn it into quarterly 

Analytics export as raw data, using the Analytics API

All data mapping happens in the source system 

Pivot table app - Data is monthly 

Using the analytics function in DHIS2, it aggregates the data from months to quarters 

Organizational Hierarchy Levels Vary Across Systems 

Repeat the process using Analytics 

Data element definitions are not equal across systems 

Data elements: used to store data

Indicators: Used to express data 

Program Indicators

To count individual records 

Data collected using Tracker. 

Connecting the Systems 

• Source system <--> Destination System
• Integration Driver - Automated data integration → Proprietary, developed by BAO 
  • Cloud hosted
  • Pull channel that makes a request to the source system - fetches and stores temporarily, then pushes into the destination system 
  • Allows to schedule, also allows logging 
  • Transformation, aggregation, mapping happens in DHIS2
  • Pull and Push channels - just enter URL for source and destination systems 

Using DHIS2 as a Business Intelligence Platform for a Health Insurance Provider 

NHIF - Kenya 

Great presentation on the theory of BI and approach to doing the business anslysis

Broad architecture of DHIS2 is a good.

Decided to use DHIS2 for pivot table, data sets, data elements and categories.

Pentaho Data Integration "Kettle" for ETL and Scheduling

• first column - Straight writing, direct reporting
• Independent Data warehouse, more structured data, improves performance, different cubes means you may find different results
• Bus architecture - you think hard about reports you need and only have cube
• EDW - operational data store (ODS) improves ability to update/change the cube

Cubes are dimensional data and all need to be represented in the cube

Business perspective

• Design of reports/dashboards
• Key figures
• Dimensions
• Hierarchies

Tech

• Design data modal
• Etl
• Etc….
• See slide

ETL process:

• source is the transaction data and updates DHIS2
• constantly checks for new organizations.
• Takes a full night (6pm to 5am) to upload data into DHIS2
• Extracts DHIS2 and Source data, transforms (compare old and new) - old data is removed, discarded
• Then loads back up into DHIS2 the subset of new valid records

DHIS2 starts acting weird or not working around 300,000 data associations, which is a downfall. Normally professional data warehouses can handle many more (millions).

Data analysis is mainly done in pivot tables and only 50% of needs being met.

Question : will an additional investment of X KSH justify an increase of Y% covered reporting needs? 

He would like to see DHIS2 as a tool for Enterprise Data Warehouse.

Challenges faced using open source:

• more software packages, the harder to maintain (different versions, etc)
• hard to have two maintain to open source systems
• wants ETL within DHIS2
• Enhance batch processing
• Enhance import of data (overwrite, aggregate info, delete)
• Enhance indicators (filter on attributes, re-usability)
• Enable mast er data
• Enhance use of dimensions 

Day 2 - Thursday, March 23

Tanzania MOH and partner integration

by Wilfred Senyoni from HISP TZ at Dar es Salaam

• End users found DHIS2 is easy to use and end up putting data into DHIS2 for dashboards. Sometimes facilities have their own instances and sometimes uses the national instance.
• Problems :
  • High reporting rates with DHIS2
  • District are committed to using HMIS but not really clear on how to use the data.
  • Data quality is quite low (40%). Data wasn't matching the paper registry or summary.
  • Limited data interpretation and planning (not using the data). 
• webportal launched hmisportal.moh.go.tz to promote transparency and innovation (for partners to access data). Story was a partner for a specific program ended up getting the entire national system web portal up and publishing quarterly. He appreciated that the partner decided to support the national program from focusing on their own programmatic needs.
• Strengthening HIS is a continuous process
• Cultivate local capacity
• strengthen existing local infrastructure
• partners have a major role in routine information use

OpenLMIS Breakout Session

Good turnout for the session. 

Business Cards & Contacts