Advancing early identification of children at risk of developing life-threatening infection in low-resource settings

Constantinos Koshiaris, Assistant Professor of Medical Statistics at the Medical School, has contributed to an international study recently published in the medical journal Nature Medicine, aimed at improving the early identification of children at high risk of severe infection in low-resource settings.

Led by researchers from Médecins Sans Frontières and the Mahidol Oxford Tropical Medicine Research Unit, the study introduces a novel approach to rapidly assessing and prioritising children when presenting for care. This approach aims to ensure the early identification of life-threatening infections while reducing unnecessary hospital referrals in resource-constrained settings.

Titled ‘Predicting referral need for febrile children in low-resource community settings in South and Southeast Asia’, the research developed clinical prediction models to identify children presenting with fever at highest risk of severe outcomes. In this context, Dr Koshiaris led the statistical design and analysis as well as the development of the prognostic models underpinning the work.

The study, which followed 3,500 children across seven countries in South and Southeast Asia, found that current World Health Organization (WHO) danger signs would miss nearly half (44.5%) of children whose health would later deteriorate. In contrast, the new models identified approximately 89% of high-risk children, while reducing referral rates from 17% to less than 5%.

Commenting on the publication, Dr Koshiaris said: ‘Developing prediction tools for these settings requires models that are both robust and simple enough to be used in routine care. Our approach shows that it is possible to reliably identify children at highest risk while keeping the tools practical for frontline use in low-resource settings’.