Estimating the tuberculosis burden in resource-limited countries: a capture-recapture study in Yemen.

BACKGROUND: The lack of applicable population-based methods to measure tuberculosis (TB) incidence rates directly at country level emphasises the global need to generate robust TB surveillance data to ascertain trends in disease burden and to assess the performance of TB control programmes in the context of the United Nations Millenium Development Goals and World Health Organization targets for TB control. OBJECTIVE: To estimate the incidence of TB cases (all forms) and sputum smear-positive disease, and the level of under-reporting of TB in Yemen in 2010. METHODS: Record-linkage and three-source capture-recapture analysis of data collected through active prospective longitudinal surveillance within the public and private non-National Tuberculosis Programme sector in twelve Yemeni governorates, selected by stratified cluster random sampling. RESULTS: For all TB cases, the estimated ratio of notified to incident cases and completeness of case ascertainment after record linkage, i.e., the ratio of detected to incident cases, was respectively 71% (95%CI 64-80) and 75% (95%CI 68-85). For sputum smear-positive TB cases, these ratios were respectively 67% (95%CI 58-75) and 76% (95%CI 66-84). CONCLUSION: We estimate that there were 13 082 (95%CI 11 610-14 513) TB cases in Yemen in 2010. Under-reporting of TB in Yemen is estimated at 29% (95%CI 20-36).