Data Visualization for Decision Making in HIV Programs
- 1. Data Visualization that Works Facilitating HIV Program Targeting: Case examples and considerations Tara Nutley, Isabel Brodsky MEASURE Evaluation Stacey Berlow Project Balance MEASURE Evaluation Webinar April 27, 2016
- 2. MEASURE Evaluation D D D D Strategic objective: To strengthen health information systems–the capacity to gather, interpret, and use data–so countries can make better decisions and sustain good health outcomes over time.
- 3. How are data visualization tools being used to improve the use of data in HIV programs? What impact do they have on decision making? What are key elements of success?
- 4. Contacted 45 people from 17 PEPFAR implementing partner organizations
- 6. SouthAfrica Using proxies to map HIV services in Nkomazi District
- 8. HIV prevalence 18.8% National 14.1% Mpumalanga Province
- 9. We can improve access to health services if we know where PLHIV are located in relation to health facilities.
- 10. How can we fill data gaps through estimation and the development of proxy indicators?
- 11. Visualization methodology Estimate HIV positivity using PMTCT testing data at each health facility Use HIV positivity to estimate HIV prevalence at each facility Interpolate to estimate HIV prevalence for all areas between health facilities Map results using ArcGIS 1 2 3 4
- 13. Most PLHIV in Nkomazi are migrant farm workers. Many of them are undocumented and unwilling to access services due to fear of stigma and deportation. Photo source: TechnoServe
- 14. Prevalence decreased from 42.7% in 2012 to 40.5% in 2014.
- 15. Zambia MOH Dashboard Question: What kind of early warning system can be put into place to help the Zambian MOH to more effectively provide prevention of mother-to-child transmission (PMTCT)/Option B+ services across the country? Look at ANC/patient trends, commodity stock-outs and lab results
- 16. Collaborate with stakeholders • Understanding how the data will be used for decision making • Types of graphs and displays • User access
- 17. PMTCT dashboard architecture PMTCT dashboard database Three national databases PMTCT dashboard
- 18. Data aspects • Extracting the source data • Calculating indicators • Data quality • Refreshing the data
- 19. Software development team • Development team who knows how to develop dashboards • Excellent testing methodology
- 20. Lessons Learned
- 21. Collaborate with data users during visualization development.
- 22. Zambia: Collaboration with MOH ensured that developers understood what information was available and what ideally was needed to monitor the PMTCT program. A working group had previously identified a set of indicators that would help the ministry and implementing partners (IPs) manage the PMTCT program. South Africa: A partnership between implementing partners, local government, and geographic information system (GIS) specialists enabled programmatic questions to be answered and needs to be met.
- 23. Consider developing proxy indicators through estimation or modeling techniques
- 24. Zambia: Developed proxy PMTCT indicators for nine of the eleven core indicators on the dashboard in order to fill missing data gaps in the interim while data are collected South Africa: Developed methodology to estimate district-level HIV prevalence using proxy data
- 25. Include training on how to interpret visualizations
- 26. South Africa: District team members trained on map reading and interpretation to understand what the maps were showing and the implications of the maps in their day-to-day work Zambia: It was agreed that training to use the dashboard would be easy for users. It was also acknowledged that interpreting the graphs and data in tables will be more challenging and will likely require MOH policies. For example: if a target was missed by X percent, what action should the organization take?
- 28. Cleaning and matching data takes time. Even the most accurate data source may have errors or duplicates. Zambia: Multi-step data standardization process South Africa: Creating a map using a GIS requires a single, authoritative, and accurate master facility list
- 30. If users realize they cannot trust the data to help them make decisions, they may not use the visualization in the future. Zambia: Found a lot of missing data. The creation of a new dashboard for PMTCT gives developers and MOH the opportunity to ensure data are of good quality so that the dashboard is ultimately useful to decision makers. South Africa: Having up-to-date estimates of HIV prevalence prevalence data at the district level is crucial for decision making. If decision makers are using old data, they may not be making accurate targeting and program planning choices.
- 31. Consider sustainability when selecting GIS and dashboard software
- 32. Data visualizations vary in cost and levelof sophistication. Zambia: Dashboard created in free and open source software, which is a cost-effective option. However, it is generally not well documented and some code may not work. South Africa: The maps were created on ArcGIS, a sophisticated but proprietary software. If Nkomazi wants the maps to be updated in the future, the analysis must be re- and they would need access to the software.
- 33. MEASURE Evaluation is funded by the U.S. Agency for International Development (USAID) under terms of Cooperative Agreement AID-OAA-L-14-00004 and implemented by the Carolina Population Center, University of North Carolina at Chapel Hill in partnership with ICF International, John Snow, Inc., Management Sciences for Health, Palladium, and Tulane University. The views expressed in this presentation do not necessarily reflect the views of USAID or the United States government. www.measureevaluation.org