Background Tuberculosis (TB) disease remains undiagnosed or unreported in approximately 4.2 million people annually. To address these “missing millions”, we assessed the feasibility and impact of a scalable model for transmission-interrupting, community-based active case finding (ACF) in major cities of four African countries: Cape Town, South Africa, Lusaka, Zambia, Harare, Zimbabwe, and Maputo, Mozambique.

Methods In peri-urban informal settlements using a 1:1 randomised controlled trial (stratified by country), we performed ACF using a low-cost mobile clinic fitted with a portable 2-module Xpert system and compared a point-of-care (POC) Xpert to a standard-of-care centralised laboratory Xpert with sputum culture used as a reference standard. At the Cape Town site, those with microbiologically confirmed TB received infectiousness studies, including smear microscopy, chest x-ray (CXR) and cough aerosol sampling (CASS).

Results Of 4193 rapidly screened individuals, 1977 were identified as at risk for TB and received targeted screening and randomisation to either POC Xpert (n=988) or centralised Xpert (n=988). Across all sites, 97 (4.9%) of 1977 participants had microbiological confirmation of TB; 53/531 at the Cape Town site (10.0% TB positivity rate; 29/53 [54.7%] culture-positive). Xpert identified 75% (22/29) of all culture-positive samples. 40/53 participants had all infectiousness studies performed, and 26/40 were identified as probably infectious (defined as cavities on CXR [n=26] and/or smear-positivity [n=9] and/or CASS-positivity [n=3]). Xpert identified almost all probably infectious cases (25/26 [96.1%]).

Conclusion Community-based active case-finding using portable molecular-based diagnostic tools reliably detects probably infectious, minimally symptomatic TB patients. These data inform key elements of ACF strategies needed to bridge the gap to find, treat and end TB.