Decide TB Publication in PLOS Medicine

Retrospective individual participant data analysis assesses diagnostic accuracy of WHO algorithms for childhood tuberculosis

In 2023, 10.8 million people fell ill with tuberculosis (TB) and 1.25 million died, confirming TB as the world’s leading infectious disease killer. Among them, an estimated 1.2 million were children, yet less than half received proper diagnosis and treatment.

To address this diagnostic gap, the World Health Organization (WHO) introduced in 2022 Treatment Decision Algorithms (TDAs) — simple, standardized clinical tools designed to help health workers identify children who should start TB treatment, especially in low-resource settings where access to advanced diagnostic tools, such as molecular or imaging tests, remains limited.

A new international study, published in PLOS Medicine, provides the first large-scale external validation of these WHO algorithms. The research was conducted within the Decide TB project, funded by the European & Developing Countries Clinical Trials Partnership (EDCTP), and coordinated by LMU Munich, Stellenbosch University, and the University of Bordeaux.

The analysis combined data from nearly 1,900 children with presumptive TB enrolled in four major paediatric studies across eleven countries in Africa and Asia (RaPaed-TB, Umoya, TB-Speed HIV, and TB-Speed Decentralization). The study assessed the diagnostic accuracy of two WHO TDAs—one including chest X-ray (Algorithm A) and one based solely on clinical information (Algorithm B).

Results show that both algorithms are highly sensitive, identifying most children who truly have TB (84% and 91%, respectively). However, their specificity—the ability to correctly rule out TB—remains moderate (51% and 31%), meaning that some children without TB would still be started on treatment. Performance was similar across key high-risk groups, including children under two years, those living with HIV, and those with severe malnutrition.

The study highlights the value of standardized clinical tools for rapid and equitable diagnosis of childhood TB, particularly at primary care level. It also emphasizes the need to improve specificity through the integration of novel diagnostic tools—such as biomarkers and artificial intelligence–assisted imaging—to ensure more accurate treatment decisions and reduce unnecessary treatment initiation.

=> Link to PLOS Medicine Publication