Setting: Screening and isoniazid preventive therapy (IPT) of child contacts of tuberculosis (TB) patients is poorly implemented in resource-limited countries, in part due to difficulties in TB diagnosis in children.
Objective: To assess the feasibility and yield of hospital-based screening and IPT in Uganda, and to evaluate the utility of symptom-based screening.
Design: Household child (age <5 years) contacts of adults with pulmonary TB were assessed for TB or latent tuberculous infection (LTBI). Children classified as 'LTBI' or 'uninfected' were prescribed IPT and followed for 9 months. Screening algorithms based on combinations of symptoms associated with TB were constructed post hoc, and their performance evaluated against a radiological-based reference standard.
Results: Of 281 contacts (median age 33 months), 44 (15.7%) started anti-tuberculosis treatment and 234 (83.3%) received IPT, 80.3% of whom completed a 6-month course. After adjustment for age and human immunodeficiency virus status, cough (aOR 4.10, 95%CI 1.39-12.11) and reduced playfulness (aOR 7.79, 95%CI 2.12-25.18) were associated with radiological TB. Screening based on cough or reduced playfulness had a sensitivity of 81.8% and a negative predictive value of 97.6%.
Conclusion: Hospital-based screening appears to be feasible, and confirms the potential utility of symptom-based screening to select children for IPT and those for further investigations.