Tuberculosis control and molecular epidemiology in a South African gold-mining community

Background Gold miners have very high rates of tuberculosis. The contribution of infections imported into mining communities versus transmission within them is not known and has implications for control strategies. Methods We did a prospective, population-based molecular and conventional epidemiological study of pulmonary tuberculosis in a group of goldminers. Clusters were defined as groups of patients with Mycobacterium tuberculosis isolates with identical IS6110 DNA fingerprints. We compared the frequency of possible risk factors in the clustered and nonclustered patients whose isolates had fingerprints with more than four bands, and re-interviewed members of 45 clusters. Findings Of 448 patients, ten were excluded because they had false-positive cultures. Fingerprints were made in 419 of 438, of which 371 had more than four bands. 248 of 371 were categorised into 62 clusters. At least 50% of tuberculosis cases were due to transmission within the community. Patients who had failed treatment at entry to the study were more likely to be in clusters (adjusted odds ratio 3·41 [95% CI 1·25–9·27]). Patients with multidrug-resistant isolates were more likely to have failed treatment but were less likely to be clustered than those with a sensitive strain (0·27 [0·09–0·83]). HIV infection was common (177 of 370 tested) but not associated with clustering. Interpretation Despite a control programme that cures 86% of new cases, most tuberculosis in this mining community is due to ongoing transmission. Persistently infectious individuals who have previously failed treatment may be responsible for one third of tuberculosis cases. WHO targets for cure rates are not sufficient to interrupt transmission of tuberculosis in this setting. Indicators that are more closely linked to the rate of ongoing transmission are needed.