Sputum scarcity and associated factors in people undergoing tuberculosis testing in South Africa, Uganda, India, and the Philippines: an analysis of cross-sectional observational data

Authors

• Welile V. Nwamba, Stellenbosch University, Faculty of Medicine and Health Sciences
• Anna O. Okunola, Stellenbosch University, Faculty of Medicine and Health Sciences
• Rebecca Crowder, University of California, San Francisco
• Devasahayam J. Christopher, Christian Medical College, Vellore
• William Worodria, World Alliance for Lung and Intensive Care Medicine in Uganda
• Charles Yu, De La Salle Medical and Health Sciences Institute
• Zaida Palmer, Stellenbosch University, Faculty of Medicine and Health Sciences
• Gcobisa Ndlangalavu, Stellenbosch University, Faculty of Medicine and Health Sciences
• Daphne Naidoo, Stellenbosch University, Faculty of Medicine and Health Sciences
• Ellen M.H. Mitchell, Prins Leopold Instituut voor Tropische Geneeskunde
• Florian M. Marx, Universität Heidelberg
• Adithya Cattamanchi, University of California, San Francisco
• Grant Theron, Stellenbosch University, Faculty of Medicine and Health Sciences

Publication Date

5-1-2026

Document Type

Article

Publication Title

Eclinicalmedicine

Abstract

Background: Tuberculosis (TB) testing often excludes people who cannot expectorate sputum. Non-sputum tests are a priority; however, there are limited data on the frequency and characteristics of sputum scarcity, which will influence the impact of non-sputum tests. Methods: We analysed adults with presumptive pulmonary TB recruited between February 2018 and February 2024 from a multi-country study (South Africa, Uganda, India, The Philippines; Cohort A, n = 2224), a South African cohort (Cohort B, n = 2179), and a South African cohort of people living with HIV (PLHIV) initiating antiretroviral therapy (ART) without syndromic selection (Cohort C, n = 297). Participants provided sputum for Xpert MTB/RIF Ultra (Ultra) and MGIT960 culture. Individuals unable to spontaneously produce sputum (non-salivary, ≥1 ml) underwent induction. We performed multivariable regression analysis and present adjusted ORs from LASSO to describe the relationship of variables with sputum scarcity. Separate sub-analysis was conducted for variables with >10% missing data. Findings: Overall, 13% (618/4700) people had sputum scarcity. Female sex [adjusted odds ratio [(aOR) 1.58 (95% confidence interval (CI) 1.31, 1.91)], and culture-negativity [1.45 (1.02, 2.08)] were positively associated with sputum scarcity. In Cohort A and B, country was also positively associated with sputum scarcity: India [1.43 (1.07, 1.91) and The Philippines [7.16 (5.34, 9.61] relative to South Africa. Smoking [0.61 (0.43, 0.88)], TB symptoms: cough [0.17 (0.13, 025)] and night sweats [0.71 (0.58, 0.86)] were negatively associated with sputum scarcity as well as previous TB [0.63 (0.47, 0.85)]. In people with TB, the only additional factor associated with sputum scarcity was, among PLHIV, being on ART [3.70 (1.19, 11.46)] were positively associated with sputum scarcity. Ultra sensitivity did not differ between induced and expectorated sputum [84% (64, 95) vs. 89% (84, 93); p = 0.218]. Interpretation: People with sputum scarcity are more likely to be female, do not smoke, have few symptoms and test culture-negative. Sputum scarcity also shows geographic variation. Among people with TB, those on ART are more likely to be sputum-scare. Sputum induction did not compromise Ultra sensitivity compared to expectorated sputum. These findings highlight people who may benefit most from non-sputum tests and the importance of such tests in people with milder disease. Funding: R2D2 TB Network and EDCTP programmes supported by the EU.

APA Citation

Nwamba, W., Okunola, A., Crowder, R., Christopher, D., Worodria, W., Yu, C., Palmer, Z., Ndlangalavu, G., Naidoo, D., Mitchell, E., Marx, F., Cattamanchi, A., & Theron, G. (5-1-2026). Sputum scarcity and associated factors in people undergoing tuberculosis testing in South Africa, Uganda, India, and the Philippines: an analysis of cross-sectional observational data. Faculty Research and Scholarly Works.
DOI:10.1016/j.eclinm.2026.103977