Abstract
Background
Microbiological diagnosis of childhood tuberculosis may be difficult. Oral swabs are a potential non-invasive alternative to sputum for diagnosis.
Methods
A prospective diagnostic accuracy study of oral swabs (buccal and tongue) for pulmonary tuberculosis (PTB) diagnosis in children (aged ≤15 years) in two South African hospital sites. Children with cough of any duration and either: a positive tuberculin skin test, TB contact, loss of weight or chest X-ray suggestive of PTB were enrolled. Two induced sputum specimens were tested with Xpert MTB/RIF (or Ultra) and liquid culture. Oral swabs were taken preceding sputum, frozen and later tested with Xpert MTB/RIF Ultra. Children were classified as microbiologically confirmed TB, Unconfirmed TB (received TB treatment) or unlikely TB according to NIH consensus definitions based on sputum microbiological results.
Results
Among 291 participants (median age 32 [IQR 14-73] months), 57 (20%) were living with HIV and 87 (30%) were malnourished. 90 (31%) had confirmed PTB (six (7%) with rifampicin-resistant TB), 157 (54%) unconfirmed PTB and 44 (15%) unlikely TB. One oral swab was obtained from 126 (43%) participants (96 tongue, 30 buccal) and two swabs from 165 (57%) (110 tongue, 55 buccal). Sensitivity was low; 22% (95% CI 15-32) for all swabs combined (confirmed PTB as reference), but specificity was high (100%, 95% CI 91-100). The highest sensitivity was 33% (15-58) among participants living with HIV. Overall yield was 6.9% with one oral swab, and 7.2% with two.
Conclusions
Ultra on oral swabs provides poor yield for microbiologic PTB confirmation in children.