Yield of household contact tracing for tuberculosis in rural South Africa

Kristen M. Little, Reginah Msandiwa, Neil Martinson, Jonathan Golub, Richard Chaisson, David Dowdy

Efficient and effective strategies for identifying cases of active tuberculosis (TB) in rural sub-Saharan Africa are lacking. Household contact tracing offers a potential approach to diagnose more TB cases, and to do so earlier in the disease course. Adults newly diagnosed with active TB were recruited from public clinics in Vhembe District, South Africa. Study staff visited index case households and collected sputum specimens for TB testing via smear microscopy and culture. We calculated the yield and the number of households needed to screen (NHNS) to find one additional case. Predictors of new TB among household contacts were evaluated using multilevel logistic regression. We recruited 130 index cases and 282 household contacts. We identified 11 previously undiagnosed cases of bacteriologically-confirmed TB, giving a prevalence of 3.9% (95% CI: 2.0-6.9%) among contacts, a yield of 8.5 per 100 (95% CI: 4.2-15.1) index cases traced, and NHNS of 12 (95% CI: 7-24). The majority of new TB cases (10/11, 90.9%) were smear negative, culture positive. The presence of TB symptoms was not associated with an increased odds of active TB (aOR: 0.3, 95% CI: 0.1-1.4). Household contacts of recently diagnosed TB patients in rural South Africa have high prevalence of TB and can be feasibly detected through contact tracing, but more sensitive tests than sputum smear are required. Symptom screening among household contacts had low sensitivity and specificity for active TB in this study.