Objective

To identify spatial clusters with unusually high levels of MDR-TB which are highly unlikely to have arisen by chance in Mpumalanga Province, South Africa.

Methods

Home addresses of all MDR-TB patients were collected from four MDR-TB facilities from 2013 to 2016. We mapped all addresses, linking them to the nearest ward with population estimates. A spatial analysis was conducted using kernel density in ArcGIS to estimate and map the distribution of the disease and used Gertis-Ord Gi to test for significant clustering.

Results

A total of 4,065 MDR-TB patients were mapped. Ten significant clusters (p-value < 0.05) were found across the province in six sub-districts: Mbombela, Nkomazi, Emalahleni, Govan Mbeki, Lekwa and Mkhondo. Mbombela has the highest number of significant clusters. The central region did not have any MDR-TB clusters.

Conclusion

There is clear evidence of MDR-TB clustering in Mpumalanga. This calls for concentrated TB prevention efforts and proper allocation of resources. Further investigations are needed to identify MDR-TB predictors.