BACKGROUND: Active tuberculosis case-finding in communities is crucial for early disease detection and disruption of transmission; however, it is complex and costly. A thorough analysis of tuberculosis screening strategies is essential, particularly because the health systems affected often have limited resources. We aimed to compare two community screening approaches in terms of tuberculosis case yield and cost. METHODS: This pragmatic community trial, with a paired screen-positive design (ie, within-person comparison), was conducted in the Butha-Buthe District in Lesotho and the uMgungundlovu District in South Africa. All household members aged at least 18 years were eligible to participate, except those who were seriously ill, had a condition that prevented their safe and full participation, were receiving tuberculosis treatment, or were pregnant (Lesotho only). We compared two screening approaches: the use of computer-aided detection (CAD) software on digital chest x-rays alone (the CAD4TBv7 approach) and the use of CAD followed by a C-reactive protein (CRP) test if the CAD score was in a particular range (the CAD4TBv7-CRP approach). A confirmatory Xpert MTB/RIF Ultra test was conducted when required by the algorithm of one or both approaches. Coprimary outcomes were to establish whether the CAD4TBv7-CRP approach was non-inferior to the CAD4TBv7 approach (non-inferiority margin -10%) and to conduct a comparative cost analysis between the two approaches. FINDINGS: Among 20 023 enrolled participants, the median age was 42 years (IQR 29-60) ; 12 387 (61·9%) participants were female, 7625 (38·1%) were male, and 11 (0·1%) were intersex. 4534 (22·6%) of 20 023 participants were living with HIV, 1547 (7·7%) had a history of tuberculosis, and 1545 (7·7%) reported at least one of the four main symptoms of tuberculosis. The primary outcome set, defined as participants who were identified as having tuberculosis according to at least one of the two approaches and had complete data for both approaches, comprised 73 participants: the CAD4TBv7 approach identified 69 (94·5%) of these cases of tuberculosis and the CAD4TBv7-CRP approach identified 60 (82·2%) cases. The difference of -12·3% (95% CI -23·0 to -1·6) indicates that the non-inferiority criterion was not met. The cost per detected case of tuberculosis was US5454 (95% uncertainty interval 4294-6777) for the CAD4TBv7 approach and 7486 (5948-9246) for the CAD4TBv7-CRP approach, making the CAD4TBv7-CRP approach 37·3% more expensive. INTERPRETATION: In active tuberculosis case-finding in our community setting, the combination of CAD followed by CRP testing offered no advantage over CAD alone owing to its inferior case yield and higher cost. CAD alone, however, has the potential to be an effective and economically viable tuberculosis screening strategy in areas with high disease burden. FUNDING: The European and Developing Countries Clinical Trials Partnership.
Reither et al. (Wed,) studied this question.