Perceptions of cough-audio-based tuberculosis screening and its integration into routine care: a qualitative study in South Africa and Uganda

Abstract Tuberculosis (TB) remains a leading cause of morbidity and mortality, requiring timely case identification for transmission control. This requires rapid screening to identify individuals eligible for molecular testing. While artificial intelligence-enabled cough-audio classifiers show promise, little is known about how they are perceived or how they may be integrated into health services in high-burden settings. This exploratory study was conducted as part of the clinical evaluation of the Cough Audio triaGE for TB ((4)-CAGE-TB) application. Data were collected in 2023–2024 across ten health facilities in Kampala, Uganda, and Cape Town, South Africa. In-depth interviews were conducted with healthcare providers (n = 109), clients (denoting individuals receiving care) (n = 57), and TB programme stakeholders (n = 14), complemented by facility observations. Data were analysed thematically to examine TB screening practices, experiences with digital health technologies, and perceptions of cough-audio–based screening. We found that TB screening was clearly articulated in national policy but inconsistently implemented; it was often reactive, driven by symptoms or clinical suspicion despite a policy shift towards risk-based screening. Clients did not recognise screening as a distinct moment, limiting opportunities for health education. Cough-audio screening was perceived as a potentially valuable innovation for improving objectivity and prioritisation for confirmatory testing. Participants suggested it could enhance engagement, strengthen recognition of TB risk, and improve linkage to testing. Enthusiasm was tempered by concerns about accuracy, workflow integration, and infection prevention. Trust in the technology was closely tied to trust in providers and the health system, underscoring its relational nature, and its positioning as a complement to clinical judgement. Overall, cough-audio TB screening holds promise not as a standalone diagnostic tool, but as a means of reshaping screening within routine care. Its impact will depend on context-sensitive integration into workflows, diagnostic cascades, and care relationships. Qualitative evidence is essential to guide implementation beyond algorithmic performance. Author Summary In many countries, TB remains a major cause of illness and death. A key challenge is identifying people early enough so they can be tested and treated. New digital tools that can analyse cough sounds are being developed to identify people who might have TB, but we know little about people’s attitudes to these tools and how they could be used in health facilities. Through interviews with healthcare staff, people attending health facilities, and TB programme staff in Uganda and South Africa, we explored their experiences of TB screening and how they viewed using cough sounds to identify people with TB. We found that although TB screening is part of national policy, it is not consistently implemented in healthcare facilities, with limited opportunities to engage with people seeking care and explain TB screening. Using cough sounds for TB screening was seen as promising because it could make screening more structured, objective, and facilitate linkage to confirmatory testing. However, participants raised concerns about accuracy and how the tool would fit into busy routines. Overall, our findings show that these technologies are not just technical tools, but part of how care is organised and experienced.