Diagnostic performance of a centrifugation-free stool processing kit in conjunction with Xpert®MTB/RIF Ultra for paediatric tuberculosis: a multicentre prospective study in five low and middle-income countries

Abstract

Diagnosing paediatric tuberculosis (TB) remains challenging. Currently available diagnostic tests have limited accuracy, and most sampling methods are invasive. In 2021, the World Health Organisation (WHO) endorsed the use of molecular WHO-recommended rapid diagnostic tests (mWRDs) with stool as an initial diagnostic tool for paediatric TB. In this study, we evaluated the diagnostic test accuracy (DTA) of stool samples processed with the Stool Processing Kit (SPK), one of three centrifugation-free methods available at the time. We conducted a multi-centre prospective diagnostic validation study in South Africa, Tanzania, Malawi, Mozambique, and India. Children (< 15 years) under investigation for TB provided at least two (induced) sputum samples at enrolment, one stool sample, and one nasopharyngeal aspirate (NPA) if aged < 5 years and a chest x-ray. TB testing included Xpert ® MTB/RIF Ultra (Ultra) and culture. Diagnostic classification followed NIH consensus statements. Stool samples were processed using the SPK and tested with Ultra. We calculated DTA in the overall cohort and predefined subgroups of interest. Out of 5313 children screened, 975 were enrolled, and data from 817 children recruited between January 2019 and June 2021 were analysed. The overall microbiological confirmation rate was 203/817 (24.9%), and an additional 358/817 (43.8%) were classified as unconfirmed TB and 256/817 (31.3%) were classified as unlikely TB. The median age was 5.3 years (IQR: 2.0–9.2), and 390/817 (47.7%) of participants were under the age of five. 134/817 (16.4%) of children were living with HIV, and 88/817 (10.8%) had severe acute malnutrition (SAM). The overall sensitivity and specificity of SPK/Ultra against a Strict Reference Standard (SRS) were 33.8% (95% CI: 27.6–40.6) and 86.8% (95% CI: 80.5–91.3), respectively. Among children under one year of age and those with SAM, sensitivity was 43.4% (95% CI: 27.4–60.8) and 41.4% (95% CI: 25.5–59.3), respectively, with specificity reaching 100% in young children and 85% (95% CI: 64.0-94.8) in children with SAM. Classic TB symptoms and chest X-ray abnormalities at baseline were associated with increased likelihood of SPK/Ultra positivity. This is one of the largest studies to evaluate stool as an alternative specimen for diagnosing paediatric TB. Although the WHO’s target product profile was not fully met, stool testing with SPK/Ultra represents a viable alternative to invasive sampling methods, especially in subgroups of interest. ClinicalTrials.gov, NCT03734172 , June 22, 2018.