MARATTO

preprint · medRxiv

Predicting radiological severity of pulmonary tuberculosis in children: an assessment of the WHO-criteria and novel prediction scores on an individual participant dataset

2026Open accessUniversity of Zambia

Abstract

Background: The World Health Organization (WHO) recommends 4-month treatment for children with non-severe pulmonary tuberculosis, outlining eligibility criteria for settings with and without chest X-ray (CXR). We evaluated the diagnostic accuracy of the WHO eligibility criteria in settings without CXR (WHO-criteria) and developed clinical scores to support disease classification. Methods: Using data from an individual participant dataset (IPD; Decide TB) of children with confirmed/unconfirmed tuberculosis from four diagnostic studies (RaPaed-TB, Umoya, TB-Speed HIV, TB-Speed Decentralisation), we assessed the diagnostic accuracy of the WHO-criteria (with/without bacteriological testing) using expert CXR interpretation as a reference. We developed two multivariable logistic regression models with (Score 1) and without (Score 2) bacteriological testing, converted coefficients into integer scores with a threshold of >10 corresponding to a sensitivity ≥70%. Results: Of 2,383 children in the Decide TB IPD, 633 (26.6%) met the eligibility criteria for a 4-month regimen, of whom 116 (18.3%) had radiologically severe disease. With and without bacteriological testing, the WHO-criteria had sensitivities of 30.1% (95%CI: 20.3%-40.2%) and 21.7% (95%CI: 10.4%-34.5%), and specificities of 83.4% (95%CI: 80.2%-86.4%) and 81.9% (95%CI: 78.8%-84.9%), respectively. Score 1 and Score 2 had sensitivities of 41.1% (95%CI: 32.4%-49.5%) and 30.9% (95%CI: 22.6%-40.4%), and specificities of 77.3% (95%CI: 73.6%-80.8%) and 83.0% (95%CI: 79.5%-86.3%), respectively. Using WHO-criteria, 91/116 (78.4%) and 105/116 (90.5%) of children were at risk of undertreatment, compared to 68/116 (58.6%) and 80/116 (68.9%) when using developed scores. Conclusions: Developed scores demonstrated better sensitivity than WHO-criteria, however, performance remained suboptimal. Implementing shorter antituberculosis regimens without CXR remains challenging in children.

Research topics

  • Tuberculosis Research and Epidemiology
  • Pneumonia and Respiratory Infections
  • COVID-19 diagnosis using AI

Sustainable Development Goals

Read the original research

This page summarises published work. The authoritative version sits with the publisher.

DOI: 10.64898/2026.07.07.26357441

Is something wrong with this record? Report it or request removal.

Discussion

Discuss this research

Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.

No discussion yet. Open the first thread.