The Global Health Dashboard Epidemic
Series: Communicating Data for Health Impact
“We have an epidemic of dashboards…we have a dashboard of dashboards.”
In November 2018, I attended a meeting where a government health policy maker from West Africa shared this honest description of global health data visualization in his country. The audience — predominantly public health professionals — laughed.
His comment (and clever use of public health jargon) has stuck with me. I think it’s an accurate reflection of global health data visualization in many low- and middle-income countries (LMICs). Dashboards and exploratory data tools dominate data visualization in global health and are frequently treated as go-to solutions to encourage evidence-based decision-making. What led to their dominance and is this unwelcome tyranny?
What triggered the epidemic?
There is a need to measure progress.
Over the past few decades, demand to measure progress towards global accountability frameworks (like the Millennium Development Goals and Sustainable Development Goals), country-level health strategic plans, and program impact has elevated the importance of robust monitoring & evaluation plans. Monitoring & evaluation frequently requires data that represent different aspects of health service delivery, which can come from multiple sources. Donors need to see if their investments are translating into measurable health outcomes and impact. Sometimes money is linked to performance, which makes analyzing and visualizing relevant data very important. This has helped encourage research and funding towards improving global health data collection, analysis, and visualization.
There is a lot of global health data.
There are two main sources of global health data — household surveys and routine health management information systems — which have separate strengths/weaknesses and can serve different functions.
Nationally-representative household surveys, including the Demographic and Health Surveys (DHS) (led by USAID) and Multiple Indicator Cluster Surveys (MICS) (led by UNICEF), are the most commonly used sources of global health data and are conducted in about 100 LMICs. Each country’s survey results are summarized in lengthy reports, which feature data visualization types (tables, bar charts, line graphs, pie charts, and maps) that have not radically changed over time.
However, DHS now more frequently produces infographic posters to accompany their reports, which is an easier way to digest the 500+ page reports. I can attest to seeing these posters hung up in government offices or people taking these posters (let’s assume they had good intentions to hang them in their office). DHS and MICS data also feeds into the DHS Program’s STATcompiler and UNICEF Data, which are exploratory data tools, and global health dashboards. DHS and MICS are conducted in LMICs every three to five years, and raw datasets are available for public download here.
Health management information system data is collected at public/private health facilities and community health institutions on health status and services delivered. DHIS2, an open-source platform, is used by more than 60 countries for their national HMIS. DHIS2 functions as an exploratory data tool and allow users to make dashboards of favorite graphs/health indicators. DHIS2 data visualization options are relatively standard (bar graphs, line graphs, area charts, pie charts, radar graphs, gauge charts).
So how many global health dashboards exist?
The volume of global health dashboards is daunting. A dashboard seemingly exists for every aspect of global health, and in some cases, there are dashboards that display similar data. Dashboards predominately are either country-specific (e.g. different dimensions of health in one country) or topic-specific (e.g. multiple different indicators related to a topic) and represent diverse data sources (including DHS and MICS data).
Here is a non-exhaustive list of dashboards listed by global health topic. Keep in mind that these are only dashboards that are publicly available; countries, donors, LMIC governments, and non-government organizations commonly have their own private digital and paper-based dashboards. Happy scrolling!
General/Maternal, newborn, and child health
- Countdown to 2030 Country Profiles
- SDG Index and Dashboards
- UNICEF Progress for Every Child in the SDG Era Country Profiles
- World Bank Health
- OCHA Humanitarian InSight
- WHO World Health Statistics — Mortality Due to Natural Disasters
- World Food Program — Emergencies
- PEPFAR Panorama
- United to Combat Neglected Tropical Diseases
- WHO World Health Statistics — HIV, Malaria, Tuberculosis
Mental health/Non-communicable diseases
- Global Nutrition Report Profiles
- WHO-UNICEF-World Bank Joint Child Malnutrition Estimates
- WHO-UNICEF Global Breastfeeding Collective
Water, sanitation, and hygiene (WASH)
Woah. Are there only dashboards?
STATcompiler and UNICEF Data, which I mentioned above, are data exploratory tools widely used globally by donor, non-governmental organization, and LMIC audiences. Institute for Health Metrics and Evaluation (IHME) data visualization tools are also popular, but in my opinion, may be less accessible to audiences in LMICs.
So are dashboards terrible?
In public health, “epidemic” has a negative connotation. A dashboard epidemic? It’s not that bad. The dashboard trend has empowered individuals to use global health data.
However, dashboards should not be considered default magical solutions for visualizing global health data. There is actually very limited evidence on how existing global health dashboards are used and if they actually influence decision-making. Theoretically, we hope that improved data availability encourages evidence-based programs and policies. In reality, the relationship between data and decision-making is tenuous in global health. Dashboards ≠ data used for decision-making.
Want to help support better dashboards in global health?
Here are some suggestions:
- Decide on a clear data story. Figure out if a similar dashboard already exists and whether it can be improved upon before making your own.
- Ensure that the target audience is engaged throughout the development process.
- Use indicators that are appropriate to your data story and can be measured relatively reliably and accurately.
- Design dashboards with accessibility in mind. Only choose data visualization types that can be accurately interpreted by the target audience. Use color-blind friendly aesthetic elements. Avoid too much flash content, which can be difficult to load in areas with poor internet connections.