Background: Culture methods remain the main diagnostic methods in the definitive diagnosis of tuberculosis despite emerging new technologies. The aim of this study is to evaluate the results of MGIT 960, LJ medium, and the microscopic smear in clinical specimens from patients suspected to have tuberculosis.
Methods: We re-evaluated the results of the LJ solid medium, MGIT 960 liquid culture methods and microscopic smear results in the 17,183 specimens collected from 9,727 patients.
Results: The number of mycobacterial isolates was 1,419 (8.3%), 2,560 (14.9%), and 2,687 (15.6%) by microscopic smear, LJ, and MGIT 960, respectively. A total 2,805 samples from 820 patients were positive in at least one of the two culture methods. The contamination rate was 3.3% for the LJ medium and 1.44% for the MGIT 960. The average time-to-growth detection was 10.4 ± 5.3 days for the MGIT 960 and 24.5 ± 7.3 days for the LJ medium (p < 0.05). A good agreement was found between the LJ medium and the MGIT 960 (p < 0.001, kappa value = 0.87). The sensitivity of the LJ medium was 0.92, specificity was 0.98, and positive predictive value (PPV) was 0.92. The sensitivity, specificity, and PPV of the MGIT 960 were 0.96, 0.99, and 0.92, respectively. The area under the ROC curve for the MGIT 960, LJ medium, and microscopic smear was 0.968 (95% CI: 0.963 - 0.973, p < 0.001), 0.947 (95% CI: 0.941 - 0.953, p < 0.001), and 0.738 (95% CI: 0.726 - 0.751, p < 0.001), respectively.
Conclusions: The LJ medium and MGIT 960 system possess comparable performance in identifying tuberculosis. The short time of result delivery is an important advantage of the MGIT 960 system.