An automated smear microscopy system to diagnose tuberculosis in a high-burden setting

Abstract

Objectives: TB-EASM (Howsome, Shanghai, China), an automated system combining smear preparation, staining and microscopy in a single platform, was evaluated for tuberculosis (TB) diagnosis in a high disease-burden setting.

Methods: Sputum samples from individuals with pulmonary TB were processed in parallel using conventional manual smear microscopy (MS), TB-EASM, liquid culture and GeneXpert. Method sensitivity and specificity were compared with Mycobacterium tuberculosis detection by mycobacteria growth indicator tube (MGIT) and/or GeneXpert MTB/RIF.

Results: Of 524 samples, 496 met evaluation criteria for study inclusion. The proportion of M. tuberculosis detected by TB-EASM was 28.2% (150/496), significantly higher than for MS (111/496, 21.2%, p 0.01) and comparable to the rate for MGIT (163/496, 32.9%, p > 0.05). For 190 M. tuberculosis-positive cases identified using MGIT and/or GeneXpert MTB/RIF, the reference standard detection methods, TB-EASM detected 140 positive cases, for an overall sensitivity rate of 73.7% (140/190, 95% CI 67.4-79.9), which was significantly higher than for MS (105/190, 55.3%, 95% CI 48.2-62.3, p < 0.01). Specificities were 96.7% (296/306, 95% CI 94.7-98.7) for TB-EASM and 98.0% (300/306, 95% CI 96.5-99.6) for MS.

Conclusion: TB-EASM outperformed conventional MS for M. tuberculosis detection in sputum specimens.

Keywords: Automated; Diagnosis; Microscopy; TB-EASM; Tuberculosis.