Mapping health data flows and use in rural Tanzania

Does the information that health facilities report “upstream” ever flow back to the local decision makers who need it most? Two new studies unveil opportunities to make health data more relevant and accessible at the hyperlocal level.

This blog post was published as part of the Data Zetu project. Data Zetu is now an initiative of the Tanzania dLab, a local NGO that promotes innovation and data literacy through a premier center of excellence. For more information about the dLab, visit www.dlab.or.tz. For more information about the Data Zetu project, visit www.irex.org.

Two new studies — informed by Data Zetu’s two years of activities, existing research, and interviews with 31 health facilities — unveil new insights into how and where health data flows across Tanzania and at the hyperlocal level in a rural district.

Understanding problem-relevant data

Governments, companies and research institutes are collecting more data than ever before. But is all this data making any difference to the lives of ordinary women, men and children in local communities seemingly far removed from global and national centres of power? To put it differently: Does the data have any relevance or use in relation to the daily challenges faced by relatively disconnected local communities?

To answer these questions, Data Zetu needed at least two critical pieces of information. First, it needed to know the actual problems faced by local communities. Listening campaigns were held at fourteen wards across Tanzania to give communities the opportunity to articulate their most pervasive and pressing problems.

At this Listening Campaign in Mwaya ward, Kyela District, community members articulated and prioritized their most pressing challenges in sectors including health and gender equality. How does information and data about these priorities flow to this hyperlocal level?

Second, Data Zetu acknowledged that the availability of problem-relevant data at the hyperlocal level is important. It would therefore be useful to establish how data flows between global agencies, national and local government agencies, civil society organisations, research institutes and hyperlocal communities. It is not only local communities who stand to benefit from access to data relevant to the problems they face; donor agencies such as PEPFAR rely on accurate and relevant local data on health service delivery to meet their goals. Understanding multi-directional and multi-level data flows allows impediments to the flow of data to be identified and recommendations for the improvement of data flows to be made.

How do we map problem-relevant data?

In Kyela District, located in the south-western part of Tanzania, one of the most pressing issues raised at Listening Campaigns was the prevalence of cholera during the rainy season. So, Data Zetu undertook research to understand better how data flows in the Tanzanian health system — especially at the hyperlocal level where people and health care actors are in the most direct danger of contact with cholera.

Our team scoured the web and the extant literature (such as Development Gateway’s 2016 analysis of Tanzania’s health and agriculture results data ecosystem) for relevant data sources, interviewed key experts and presented preliminary versions of the data flow maps for comment and improvement, including at the 2018 Open Data Research Symposium in Buenos Aires. We also interviewed key personnel in Kyela’s local government authority, including the District Medical Officer, and visited every one of the 31 public hospitals, dispensaries, clinics and health centres in Kyela District to establish first-hand how health data flows at the hyperlocal level.

All 31 health facilities in Kyla District were visited by the Data Zetu research team. [Source: Data Zetu]

Our findings: Upstream data, downstream blockages

The two research reports published present many interesting findings and useful recommendations. But probably the most unique and insightful findings from the research are the two data flow maps themselves. The first map shows the flow of health between multiple actors and at all levels of the data system.

Data flows in the Tanzanian health system. [Source: Van Schalkwyk & Silaa (2018) Connecting Flows and Places: Making Data Useful to Hyperlocal Communities in Tanzania]

Insights that emerge from the above data flow map include:

1. Data flows are more definitively outwards and upwards from local communities and their health facilities than they are inwards or downwards.
2. There are several data flows that terminate, nor are there sufficient feedback loops in the flow of health data. Terminations stagnate the flow of data in the system by precluding the creation of feedback loops.

The absence of functional feedback loops limits opportunities for improving the quality of health data which was found to be of concern. A lack of feedback loops in the system also limits the ability of system to improve on the relevance of the data collected to hyperlocal communities.

The second map zooms in on the flow of health data between public health facilities and the local government authority in Kyela District.

The flow of health data at the hyperlocal level. [Source: Silaa & Van Schalkwyk (2018) Mapping Hyperlocal Health Data Flows: The Case of Kyela District, Tanzania]

The map on health data flows at the district level reveals that:

1. There are no formal systems/guides that allow for horizontal data flows between health facilities in the district. Health facilities do, however, exchange health information informally indicating a demand for data from neighboring health facilities in the district.
2. There is no formal systems/guides for sharing of facility-level health data directly with communities, or vice versa. While formal structures in the form of the Health Facility Governance Committee and ward and village committees are in place to connect communities to their health facilities, the committees do not function effectively in a complex and hierarchical communication system. There is no evidence of a functional mechanism for capturing extra-facility health data on events such as births and deaths in local communities.
3. Health facilities work closely with NGOs which, in turn, work closely with local communities. However, NGOs’ data are not effectively integrated into the health data ecosystem.

It is clear from both data flow maps that the flow of problem-relevant data to hyperlocal communities can be improved. Such improvements in the flow of data will place communities in a better position to use data effectively in their efforts to improve their health and well-being.

The insights gained from the research and recommendations put forward by Data Zetu, combined with the Subnational Data Roadmap, will be taken forward by the Tanzanian dLab as it works to leverage data to drive better policies and decision-making. And while a broad cross-section of relevant stakeholders were engaged to produce these documents, we invite for any person who will be able to develop the second version if seems needed And further research is already underway in Cote d’Ivoire that will build on the groundbreaking work done by Data Zetu as we continue to develop a better understanding of how data can be made useful to hyperlocal communities around the globe.

At this meetup in western Tanzania, the Data Zetu team meets with local community members to share data about health priorities of their fellow citizens, collected months earlier at a Listening Campaign.