Setting: The diagnosis of tuberculosis (TB) in children is seldom confirmed, and is based mainly on clinical signs, symptoms and special investigations. Various attempts in the form of diagnostic approaches have been made to rationalise this diagnostic process.
Aims: To review and describe published diagnostic approaches aimed at diagnosing mainly intrathoracic tuberculosis in children in developing countries; to compare diagnostic approaches with each other and with bacteriologically confirmed TB; and to describe modifications to the diagnosis of TB in HIV-infected or malnourished children.
Methods: Literature review classified into 1) diagnostic approaches, 2) characteristics used in diagnostic approaches, and 3) studies done to validate diagnostic approaches.
Results: Sixteen systems were analysed. Comparison of systems is difficult because characteristic definitions and the ranking of characteristics are not standardised, few studies have been performed to validate these diagnostic approaches, and the gold standard of diagnosis is not practicable in most settings. The minority of systems are adapted for HIV-infected and malnourished patients.
Recommendations: Characteristic definitions and ranking of characteristics should be standardised. Any new diagnostic approaches developed should be relevant to developing countries with limited resources, a high burden of tuberculosis, malnutrition and HIV/AIDS and a young population. Studies done to validate diagnostic approaches should be conducted scientifically.