RESUMO
INTRODUCTION: Childhood pulmonary tuberculosis (TB) remains a diagnostic challenge. This study aimed to evaluate the performance of Xpert Ultra for the diagnosis of pulmonary TB in children in a low TB prevalence setting. METHODS: Prospective, multicentre, diagnostic accuracy study. Children with clinical or radiological suspicion of pulmonary TB were recruited at 11 paediatric units in Spain. Up to three gastric or sputum specimens were taken on 3 consecutive days, and analysed by Xpert MTB/RIF, Xpert Ultra and culture in parallel. RESULTS: 86 children were included (median age 4.9 years, IQR 2.0-10.0; 51.2% male). The final diagnosis was pulmonary TB in 75 patients (87.2%); 33 (44.0%) were microbiologically confirmed. A total of 219 specimens, comprising gastric aspirates (n=194; 88.6%) and sputum specimens (n=25; 11.4%), were analysed. Using culture as reference standard and comparing individual specimens, the sensitivity was 37.8% (14/37) for Xpert MTB/RIF and 81.1% (30/37) for Xpert Ultra (p<0.001); specificity was 98.4% (179/182) and 93.4% (170/182), respectively (p=0.02). In the per-patient analysis, considering positive results on any specimen, the sensitivity was 42.9% (9/21) for Xpert MTB/RIF and 81.0% for Xpert Ultra (17/21, p=0.01); specificity was 96.9% (63/65) and 87.7% (57/65, p=0.07), respectively. CONCLUSIONS: In children with pulmonary TB in a low burden setting, Xpert Ultra has significantly higher sensitivity than the previous generation of Xpert assay and only marginally lower specificity. Therefore, in children undergoing evaluation for suspected pulmonary TB, Xpert Ultra should be used in preference to Xpert MTB/RIF whenever possible.