The use of diagnostic systems for tuberculosis in children.

Effective management of tuberculosis (TB) in children and important data of disease burden continue to rely on a clinical approach to diagnosis, as diagnosis of childhood TB is not confirmed in the majority. Many diagnostic scoring systems have been developed to aid with diagnosis. This article reviews the use and evaluation of these approaches. The diagnostic systems are often closely related and all rely on the well-known clinical features associated with TB disease in children. The scoring systems are not well validated and validation is limited by the lack of a gold standard for comparison. When they have been validated, some systems perform reasonably well but may bias to identify the most obvious clinical cases. They perform less well in important sub-groups that pose the greatest diagnostic challenge and are at greatest risk for poor outcome, such as the young, malnourished or HIV-infected. There is marked variation in performance between these diagnostic approaches. The better validated systems may have a role as a screening tool in some settings, but this would need careful consideration as to the most useful and safest approach. More attention is being given to improving diagnosis and management of child TB, including within National TB Programmes. Research with new diagnostics should include children so that there is less reliance on clinical features alone. However, the clinical approach will continue to be relevant and so it is important to strive to improve the diagnostic approach to TB in children, and to validate the approach in different settings.