Background: The diagnosis of tuberculosis (TB) in young children can be challenging, especially in severely malnourished children. There is a critical need for improved diagnostics for children. Thus, we sought to evaluate the performance of a technique that measures antibodies in lymphocyte supernatant (ALS) for the diagnosis of TB in severely malnourished children presenting with suspected pneumonia.
Methods: Children less than 5 years with severe acute malnutrition and radiological features of pneumonia admitted to the Dhaka Hospital of International Centre for Diarrhoeal Disease Research, Bangladesh, were enrolled consecutively following informed written consent. In addition to clinical and radiological assessment, samples taken for TB diagnosis included gastric lavage fluid and induced sputum for microbiological confirmation. ALS was measured from venous blood, and results were evaluated in children classified as "confirmed", "non-confirmed TB" or "not TB".
Results: Among 224 children who had ALS analysis, 12 (5.4%) children had microbiologically "confirmed TB", a further 41 (18%) had clinically diagnosed "non-confirmed TB" and the remaining 168 (75%) were considered not to have TB. ALS was positive in 89 (40%) and negative in 85 (39%) of children, with a large number (47 or 21%) reported as "borderline". These proportions were similar between the three diagnostic groups. The sensitivity and specificity of ALS when comparing "Confirmed TB" to "Not TB" was only 67% (95% CI: 31-91%) and 51% (95% CI: 42-60%), respectively.
Conclusions and significance: Our data suggest that ALS is not sufficiently accurate to improve the diagnosis of TB in children with severe malnutrition.