PP63 Household assets and Tuberculosis infection in northern KwaZulu-Natal – data from a tuberculin school survey

Abstract

Background Analyses of the 2010 ZAMSTAR Tuberculosis (TB) prevalence survey show that TB disease remains associated with poverty in Southern Africa. The association between TB infection and poverty is less clear, with a Zambian study finding higher rates of TB infection in residents of wealthier households (Am J Trop Med Hyg 2009; 80(6): 1004–1011). We examined the association between household wealth and TB infection in children in northern KwaZulu-Natal. Methods We obtained parental consent to test 6–8 year old children for TB infection using the Mantoux tuberculin skin test. These children were registered in the Africa Centre household surveillance programme and attending local schools. Household wealth quintiles were derived using Principal Component Analysis from data on assets. In STATA v13, logistic regression was used to assess the association between TB infection and household wealth. To account for clustering, school was fitted as a random effect. These analyses were adjusted for age, sex and household location (urban/periurban/rural). Two sensitivity analyses were conducted. In one, the analysis was restricted to children with a Bacillus Calmette–Guerin vaccine (BCG) scar. In the second, untested children reported to be receiving TB treatment were assumed to have TB infection. Results School visits were conducted July to December 2013. We obtained Mantoux readings on 1258 of 3699 resident children with 14.5% positive at ≥10 mm. A further 28 children were reported to be receiving TB treatment. 90% of tested children had a BCG scar. We were less likely to have Mantoux readings for children in wealthier households, rural children and older children. Thirty-eight children for whom we had no data on household wealth were excluded from the analysis. There was little association between household wealth and TB infection (p = 0.53), with the adjusted odds of infection in the wealthiest quintile 0.78 times lower than in the poorest quintile (n = 1220; 95% CI 0.42–1.47). When analysis was restricted to children with a BCG scar, there was again little association between household wealth and TB infection. Including untested children reported to be receiving TB treatment did not alter estimates. Conclusion In this setting, relative wealth does not appear to be associated with TB infection. The association between TB infection and household wealth may vary across Southern Africa. This may be a result of differences in the associations between socio-economic position and household structure or social contact patterns in different settings.