Diagnostic strategies for childhood tuberculosis in the context of primary care in a high burden setting: the value of alternative sampling methods.

Background: Hospital studies have demonstrated the usefulness of alternative sampling strategies to expectorated sputum and new diagnostics for the diagnosis of childhood tuberculosis (TB) but there is limited evidence of how these approaches work in the primary-care setting. Aim: To assess the feasibility and yield of a variety of sample types and diagnostic tests for childhood TB at a primary-care clinic. Methods: A prospective cohort of children (<10 years) with signs and symptoms of TB was enrolled at a primary-care clinic in Johannesburg, South Africa. Tuberculin skin testing (TST) and chest X-ray (CXR) were performed in all. In those unable to expectorate, one induced sputum (IS), one ambulatory gastric aspirate (GA) and two nasopharyngeal aspirates (NPA) were collected. Stool was collected from all. Samples were processed for smear microscopy, liquid culture and Xpert MTB/RIF. The Determine TB LAM Ag (LAM) test was used for HIV-positive children. Results: From July 2013-December 2014, 119 children were enrolled, 21 (18%) of whom were HIV-positive. TST was positive in 25/105 (24%) and 70/116 (70%) had a positive CXR. Four (3%) had confirmed TB, 101 (85%) unconfirmed TB and 15 (13%) unlikely TB. Of the 469 samples collected, smear microscopy was positive in none, Xpert was positive in four (<1%) and culture was positive in two (<1%). Three of 11 (27%) HIV-positive patients were positive by LAM. Treatment was commenced in 48/119 (40%). Conclusions: At primary-care, alternative sampling strategies proved feasible but resulted in a low diagnostic yield. Extensive efforts to bacteriologically diagnose children did not contribute to clinical management.