Liquid vs. solid culture for tuberculosis: performance and cost in a resource-constrained setting.

SETTING: National Health Laboratory Services tuberculosis (TB) laboratory, South Africa. OBJECTIVES: To compare Mycobacterium Growth Indicator Tube (MGIT) with Löwenstein-Jensen (LJ) medium with regard to Mycobacterium tuberculosis yield, time to positive culture and contamination, and to assess MGIT cost-effectiveness. DESIGN: Sputum from gold miners was cultured on MGIT and LJ. We estimated cost per culture, and, for smear-negative samples, incremental cost per additional M. tuberculosis gained with MGIT using a decision-tree model. RESULTS: Among 1267 specimens, MGIT vs. LJ gave a higher yield of mycobacteria (29.7% vs. 22.8%), higher contamination (16.7% vs. 9.3%) and shorter time to positive culture (median 14 vs. 25 days for smear-negative specimens). Among smear-negative samples that were culture-positive on MGIT but negative/contaminated on LJ, 77.3% were non-tuberculous mycobacteria (NTM). Cost per culture on LJ, MGIT and MGIT+LJ was respectively US$12.35, US$16.62 and US$19.29. The incremental cost per additional M. tuberculosis identified by standard biochemical tests and microscopic cording was respectively US$504.08 and US$328.10 using MGIT vs. LJ, or US$160.80 and US$$109.07 using MGIT+LJ vs. LJ alone. CONCLUSION: MGIT gives higher yield and faster results at relatively high cost. The high proportion of NTM underscores the need for rapid speciation tests. Minimising contaminated cultures is key to cost-effectiveness.