SPATIAL PATTERNS OF MULTIDRUG-RESISTANT TUBERCULOSIS DISTRIBUTION IN CAPE TOWN AND PORT ELIZABETH, SOUTH AFRICA

Abstract

Objectives: To carry out a spatial clustering analysis of multidrug-resistant tuberculosis incidence (MDR TB) in South Africa on the global and regional levels and to use multiple linear regression to find socioeconomic predictors of the distribution using data from the Cape Town and Nelson Mandela Bay (Port Elizabeth) local municipalities.

Methods: National Health Laboratory Service data between 2009 and 2011 were used along with South African Census 2011 data to find spatial distribution of MDR TB incidence. The Getis-Ord Gi* statistic tool of ArcGIS® 10.2 was used to compute the location of significant hot and cold spots on the level of South Africa, Cape Town and Port Elizabeth (Nelson Mandela Bay municipality). Lastly, the regional datasets were examined by the means of a non-spatial multiple linear regression in STATA 13.

Results: MDR TB incidence data points were significantly clustered on thenational as well as regional levels. The Getis-Ord Gi*statistic yielded significantresults, showing the hot and cold spots of the disease on the level of South Africaas well as the Cape Town (CT) and Nelson Mandela Bay (NMB) loca lmunicipalities. A multiple linear regressions selection revealed that in the CT municipality the MDR TB incidence variable was significantly negatively correlated with education (p=0.002) and positively with income (p=0.015) and TB incidence (p<0.001). Port Elizabeth MDR TB incidence showed a significant positive relationship with TB incidence (p<0.001).

Conclusions: The distribution of multidrug-resistant tuberculosis is significantly non-random on at least two levels of resolution, the national and the municipal. In general, the spatial distribution of MDR TB and socioeconomic variables suggests higher incidence in economically disadvantaged areas, which was confirmed by analysis of data from CT and NMB. Nevertheless, further research is required to understand the socioeconomic predictors of the disease and to find additional variables that may be associated with the spatial distribution of MDR TB both locally and globally.