Seeing TB More Clearly: Using Subnational Data to Support Targeted Case-Finding in South Africa
South Africa continues to carry a high burden of tuberculosis (TB), but that burden is not evenly distributed. National, provincial and even district-level estimates can mask substantial variation between communities, raising an important programmatic question: Where should limited TB case-finding resources be prioritized?
Through the Subnational TB and HIV Modelling Project in South Africa, funded by the Gates Foundation and implemented in support of the National Department of Health’s TB programme, EPCON is exploring how routinely available health data, spatial modelling and geospatial tools can provide a more granular picture of TB burden and support more targeted programme planning.
Looking beyond observed TB positivity
Routine TB testing data provide valuable information on where TB is being diagnosed, but observed positivity does not necessarily reflect underlying burden. Areas differ in how much testing takes place and in who is reached by testing services.
To account for this, EPCON combines routine molecular TB testing data with demographic, socioeconomic, HIV, healthcare-access and spatial information to estimate TB burden at health sub-district level. The modelling approach adjusts for differences in testing intensity and calibrates the resulting estimates against national epidemiological estimates from the Thembisa model.
This allows us to move beyond simply asking “Where are more positive TB tests being recorded?” towards asking “Where might the underlying TB burden be concentrated once differences in testing effort are considered?”
Identifying where people with TB may be missed
The project also examines the TB detection gap by comparing modelled burden with observed TB notifications. This can highlight areas where estimated burden exceeds the number of people currently being diagnosed and notified.
This information has important programmatic potential. It can help identify areas where additional testing or active case-finding could be considered and where existing services may not be reaching everyone with TB.
The analysis is taken further through an Enumeration Area-level vulnerability index, incorporating factors such as population density and informal dwelling patterns. Rather than representing directly observed TB incidence, this provides a more granular prioritisation layer to identify communities that may warrant closer attention.
From modelling to action
These outputs are brought together in the EPCON Epi-control platform, where users can interactively explore estimated TB burden, testing patterns, detection gaps and local vulnerability.
Importantly, the work is already moving beyond model development towards practical application. EPCON has worked with implementation partners, including the Aurum Institute, AHRI and Aquity Innovations, to generate geographically targeted recommendations that can inform the prioritisation of TB case-finding activities in their implementation areas. This provides an opportunity to assess how model-generated recommendations can complement local programme knowledge and support operational planning.
The intention is not to replace existing surveillance systems or programme expertise, but to provide an additional decision-support tool that helps translate complex data into geographically targeted action.
