Pooled sputum testing could halve cartridge use for tuberculosis diagnosis

A seven-country study found that testing sputum samples in pools used 51 per cent fewer molecular-test cartridges, with a small reduction in sensitivity compared with individual tests

Pooling sputum samples for tuberculosis testing could allow laboratories to test more people with the same number of molecular-test cartridges, according to the largest study of the approach to date. The multicountry findings helped inform World Health Organization (WHO) guidance for adolescents and adults in resource-constrained settings.

The study enrolled more than 14,000 people with tuberculosis symptoms in Bangladesh, Brazil, Cameroon, Kenya, Malawi, Nigeria and Vietnam between November 2023 and February 2025. Each person’s sputum was tested individually and in a pool of up to four samples using Xpert MTB/RIF Ultra, a rapid molecular assay for Mycobacterium tuberculosis (MTB) and rifampicin (RIF) resistance. The researchers compared results with culture, a reference method for tuberculosis diagnosis.

Pooled testing used just more than 7,000 cartridges, compared with about 14,000 for individual testing, a reduction of 51 per cent. The reported cost per test fell from nearly $8 to less than $4. Sensitivity was 3.1 per cent lower than with individual testing, mostly for samples with very low levels of tuberculosis bacteria.

“Pooling lets us test almost twice as many people for the same investment,” said Dr. Cyrille Mbuli, a study partner in Cameroon. The research team said the method requires no additional technology and could help extend access to rapid molecular testing where cartridge supplies and budgets are limited.

Tuberculosis remains a major cause of death, and many people with the disease are not diagnosed. The researchers are now examining how pooled testing could be combined with computer-aided chest X-ray triage and testing of tongue swabs. Local programmes will need to weigh the resource savings against the modest loss in sensitivity.

Pooling involves combining specimens from several people in a single molecular-test cartridge. If the pooled result is negative, the group can be cleared with one test; if it is positive, further testing is needed to identify which individual samples contain tuberculosis. This approach uses existing cartridges and equipment, but the study assessed a pool size of up to four sputum specimens.

The study was conducted through community screening and at primary health centres and district hospitals. It was the first to compare pooled testing with culture rather than only with individual molecular tests. The small sensitivity reduction occurred mainly in specimens with very low bacterial loads, so programmes will need to consider how pooling fits their local diagnostic pathways.

The work was conducted by Start4All, a four-year partnership funded by Unitaid and led by the Liverpool School of Tropical Medicine, in the UK, with the Stop TB Partnership based in Geneva, Switzerland and research groups in seven high-burden countries. The results were generated in routine laboratories and clinics, giving the study evidence from settings where pooled testing may be considered.

Tuberculosis causes about 1.3 million deaths each year, and only around half of people globally receive a WHO-recommended rapid molecular test as their initial test. The Start4All partnership said the results had informed WHO guidance on pooled testing. The Stop TB Partnership has also developed a toolkit for health professionals and national programmes, alongside a cost calculator to support decisions about where pooling may be useful.

The work is part of a broader set of studies on implementation, acceptability and accuracy in different settings. The next phase will combine pooled tests with computer-aided chest X-ray triage and assess tongue-swab samples as a simpler specimen option.

The researchers reported that the method reduced the cost per test by more than half while maintaining similar diagnostic performance for most samples. Because the sensitivity difference was concentrated among samples with very low bacterial loads, local programmes will need to consider the risk of missed cases alongside the increase in testing capacity.

For further reading please visit: 10.1016/j.langlo.2026.104051