In August 2017, a data analysis workshop was organised in Dhaka, Bangladesh, during which stakeholders from various Bangladeshi institutions – including BRAC, MSH and ICDDR Bangladesh – worked together to generate new hypotheses about where missing cases of TB are found, whom they may be, why they may be missed, and which solutions are needed to reach and care for them. Building on the extensive knowledge of the KIT CASE team, maps were produced and presented to assess these questions. This followed KIT’s Bangkok workshop on its MATCH Approach to tuberculosis care in 2017.
The maps demonstrated that each district in Bangladesh operates in a unique environment: with different TB burdens, levels of programme quality, socioeconomic conditions, and access to facility services. Comparing these different pieces of information by data triangulation can be useful to detect commonalities and inconsistencies in the data which in turn can be used for hypothesis generation.
Data triangulation & mapping
Triangulation is a powerful analytical technique which facilitates validation of data through cross verification of information from two or more analytical methods or data sources. By combining multiple data sources and filters through which we observe epidemiological processes, we can validate assumptions about possible interventions to improve prevention and care. An example of how triangulation is used to compare and validate assumptions regarding TB case detection in Bangladesh is shown in the maps above.
The maps on the top show the number of people tested for TB within the total population as well as the number of diagnosed TB patients being reported. These maps provide information regarding how many TB suspects are being identified and diagnosed. Since these outcomes depend largely on the coverage and access to services, assumptions regarding service availability need to be validated. The maps to the lower left show indicators of general health service delivery (immunisation) and coverage (microscopic diagnostics available). Finally the number of people who are at risk for TB and are accessing health care is strongly influenced by their socioeconomic status. The maps to the lower right show the literacy and poverty of populations in Bangladesh. Brought together these three thematic conceptualisations provide a practical and concise picture of localised TB epidemics and programme responses.
In addition to triangulating data, spatial statistics are used to identify patterns and anomalies in these data to improve the rigour of these analyses. For more insights into said spatial analyses, please refer to our MATCH Manual.