Setting: National reference laboratory in Zambia, a high-incidence setting with a high prevalence of HIV infection.
Objective: To compare the performance of a commercial bacteriophage kit with a nucleic acid amplification kit and an 'in-house' bacteriophage method for rapid diagnosis of pulmonary tuberculosis (TB).
Methods: Sputum specimens from suspected pulmonary TB cases were examined by direct fluorescence microscopy and culture on Löwenstein Jensen (LJ). In a blinded study, remaining samples were tested by AMTD and FASTPlaqueTB or an in-house bacteriophage assay. Two specimen decontamination protocols were investigated.
Results: Microbial contamination of 40.4% was observed when using the FASTPlaqueTB kit specimen preparation protocol. When compared to culture on LJ, the sensitivity of the FASTPlaqueTB test was 20.7%. Implementation of a modified Petroff's decontamination protocol reduced contamination to 5.8% and the FASTPlaqueTB test detected 8/25 (32%) of culture-positive specimens. The sensitivity of AMTD and smear microscopy for these specimens were 64% and 48%, respectively. In a separate experiment the sensitivity of an in-house bacteriophage assay was 45.3% compared to 64.2% for AMTD and 45.3% for direct smear microscopy.
Conclusions: Additional analysis of sputum specimens by bacteriophage assay provided no advantage in this setting. For the rapid diagnosis of TB, AMTD offered improved sensitivity over direct smear microscopy.