Stool-based xpert ultra testing for pulmonary tuberculosis in children in Kampala, Uganda

Abstract

Background.. Globally, the diagnosis of tuberculosis (TB) in children is challenging principally because of the paucibacillary nature of TB in children and the difficulty in obtaining good quality sputum samples. More than 200,000 children die from pulmonary TB each year , due in large part to diagnostic failures. Children (< 5 years) are often unable to expectorate sputum, and sputum induction is not readily available in resource-limited settings, and it also poses practical limitations. Hence, there is a critical need for tests using , non-sputum samples that can rapidly and accurately detect TB in children (1). Methods. In a cross-sectional diagnostic study in Mulago, Uganda, we used stool samples obtained from consecutively enrolled children under 15 years of age being assessed for pulmonary TB from November 2018-September 2020. Clinical investigations included tuberculin skin testing, chest x-ray and HIV testing. Respiratory specimens were obtained for Xpert MTB/RIF Ultra (Ultra), smear microscopy, and liquid and solid culture. We collected stool before treatment initiation and used the FIND Stool Processing Kit (SPK) followed by Ultra testing (SPK-Ultra). We compared the sensitivity and specificity of SPK-Ultra to sputum Xpert Ultra and smear microscopy using confirmed and unlikely TB as the reference, and evaluated the sensitivity among culture-positive cases. We calculated any sensitivity improvement with a second SPK-Ultra and the yield in children with unconfirmed TB. Results: We enrolled 156 children, including 40 (26%) with confirmed TB. The median age was 4 years (IQR 1.8-7), 51% were male, 12% were HIV-positive (median CD4 cell count 730), 60.4% were underweight, and 40% had diarrhea by the Bristol stool scale. Fourteen (9.2%) stool Ultra tests were invalid but reduced to three (2%) on repeat testing. SPK-Ultra was positive in 14 (15%) children, with sensitivity 44.8% (95% CI 26.4-64.3) and specificity 98.1% (95% CI 89.9-100). Sensitivity was similar to smear (smear sensitivity 50% vs. 58.3% from SPK-Ultra, -8.3% difference, 95% CI -28.5 to +11.8). Among culture-positive cases, sensitivity was lower than sputum Ultra (70.8% vs. 58.3%, difference +12.5%, 95% CI -4.9 to +29.9), although not statistically significant.

Conclusions: Stool Xpert Ultra detected only about a third of children with confirmed TB.