Reducing the prevalence of tuberculosis in a highly endemic setting by community-wide active case finding


STUDY SITES: Ca Mau province

TIMELINE: 2013 –2019

PROJECTOR IMPLEMENTOR: National Lung Hospital / National TB Program

FUNDED BY: Woolcock Institute of Medical Research

CORE STAKEHOLDER: Ca Mau Center for Social Disease Prevention

PRIMARY OBJECTIVE: To determine the effectiveness of community-wide screening for TB using the Xpert MTB/RIF platform annually for four years, compared with usual TB control practices, on the TB prevalence in a rural province of Vietnam.


  1. Estimating the impact of the intervention on transmission of TB, measured as

    a. the prevalence of TB infection in children, and

    b. the transmission index estimated from molecular epidemiological investigation.

  2. Estimating the proportion of screened people with chest x-ray findings suspicious for pulmonary TB

    a. who are able to provide a sputum specimen, and

    b. who have a positive result of Xpert MTB/RIF.

  3. Measuring the cost-effectiveness, expressed as the cost per prevalent case of TB avoided;

  4. Estimating the sensitivity of a single sputum Xpert MTB/RIF test for culture-proven pulmonary TB in the context of active case finding in Vietnam;

  5. Identifying lessons for scaling up this approach;

  6. Identifying the conditions under which this approach is likely to be most effective, feasible and affordable.


All eligible and consented individuals aged 15 and above in 60 randomly selected sub-communes were asked to produce a single spontaneous sputum sample for Xpert MTB/RIF testing, regardless of symptoms. Participants whose sputum was Xpert MTB positive were invited to have a plain chest x-ray and to submit two additional sputum samples for confirmatory liquid culture. The screening will be conducted annually for four years.


We will use a two level, hierarchical generalised linear model with a Poisson error distribution to test the effect of treatment group allocation on number of prevalent cases of culture-proven pulmonary TB in the final year. Treatment group allocation will be the main (fixed) effect. Clusters will be treated as a random intercept term. Cluster population aged ≥ 15 years will be the offset. There will be no interim analyses and a p value < 0.05 will be treated as significant. No covariates will be included in the primary analysis. Proportions will be estimated with 95% confidence intervals, adjusted for clustering. The sensitivity of single sputum Xpert MTB/RIF will be estimated within the final year cross-sectional (prevalence) survey in the active intervention clusters. Positive sputum cultures for MTB are the “gold standard” for diagnosis. A cost-effectiveness analysis will be conducted by the end of the study.