RESUMO
SETTING: Despite the widespread introduction of Xpert(®) MTB/RIF in developing countries, reports of its use and value in routine conditions remain limited. OBJECTIVE: To describe Xpert results in relation to microscopy, treatment initiation, cost and workload under routine conditions at four sites in Cambodia, Georgia, Kenya and Swaziland. DESIGN: Laboratory and clinical information on presumed TB patients were obtained from routine registers over a period of at least 6 months between March and November 2012. RESULTS: Among the 6086 presumed TB patients included in the analysis, Xpert testing increased the number of biologically confirmed cases by 15% to 67% compared to microscopy. Up to 12% of the initial Xpert results were inconclusive. Between 56% and 83% of patients were started on treatment based on microscopy and/or Xpert results, with median delays of 1-16 days. Rifampicin resistance was detected in 3-19% of Xpert-positive patients. CONCLUSION: Despite the additional numbers of cases detected by Xpert compared to microscopy, large proportions of patients are still started on treatment empirically in routine practice. Patient and specimen flow should be optimised to reduce delays in treatment initiation. Simple, non-sputum-based point-of-care tests with high sensitivity are needed to improve TB diagnosis and management.