Diagnosing childhood pulmonary tuberculosis using a single sputum specimen on Xpert MTB/RIF at point of care

  • N Gous
  • LE Scott
  • S Khan
  • G Reubenson
  • A Coovadia
  • W Stevens


Background. The GeneXpert MTB/RIF (Cepheid, USA) (Xpert) has proved successful for pulmonary  tuberculosis (TB) diagnosis on decontaminated/concentrated induced sputum specimens from children.  Capacity to perform induction in many settings is limited.
Objective. To assess: (i) volumes of ‘routinely obtained’ sputum in a district-level academic hospital; (ii) whether sputum specimens not meeting Xpert-required testing volumes could still be tested; and (iii) performance of Xpert on a single paediatric sputum specimen at point of care (POC).
Methods. Two sputa were collected from paediatric TB suspects (≤14 years) at Rahima Moosa Mother and Child Hospital, Johannesburg, South Africa. One specimen was weighed at POC; if the volume was  ≥0.1 mL but <0.5 mL, it was increased to 0.5 mL using saline. On-site Xpert testing (G3 cartridge) was performed by a dedicated laboratory technician. The second specimen was referred for TB smear microscopy and culture as per standard of care (SOC).
Results. A total of 484 patients presumed to have TB (median age 24 months) were eligible for this  study, performed between June 2011  and May 2012. Xpert could not be used on 4.1% of specimens because of volumes <0.1 mL, and 62.8% required addition of saline prior to Xpert testing. Xpert  generated a 2.2% error and 3.7% invalid rate, compared with the SOC that rejected 2.3% because of insufficient volume and 2.3% that were contaminated. The diagnostic performance compared with culture was 62.5% (95% confidence interval (CI) 24.7 - 91) and 99.1% (95% CI 97.4 - 99.8) sensitivity and specificity, respectively, for Xpert (n=345) and 33.3% (7.9 - 69.9) and 99.5% (98.1 - 99.9) sensitivity and specificity, respectively, for smear microscopy (n=374).
Conclusions. Up to 67% of ‘routinely obtained’ sputum specimens from children (≤14 years) are below the required volume for Xpert testing but can be ‘topped up’ with saline. Xpert MTB/RIF performed better than microscopy and generated clinically relevant, timeous results, but sensitivity did not reach the  same levels as culture in children.

Journal Identifiers

eISSN: 2078-5135
print ISSN: 0256-9574