Introduction The importance of Xpert^® MTB/RIF (GeneXpert) for the rapid confirmation of tuberculosis (TB) and indicator for drug resistance is now well established for high burden, resource poor countries, with a high prevalence of smear positive pulmonary disease. The phenotype of disease differs significantly in high income countries like the UK with fewer smear positive cases and more extra-pulmonary disease. The role of GX in this setting is unclear.

Objectives To evaluate how GeneXpert was being used in local practice and determine factors that influenced test performance.

Methods We performed a retrospective analysis of all GeneXpert tested patient samples between October 2011 and April 2014. Out of area sample requests and unprocessed specimens were excluded. Positive GeneXpert results were categorised as very low, low, medium and high. Sensitivity and specificity analyses excluded clinically diagnosed TB without supporting evidence and stratified samples by type and smear status. Statistical analyses were computed on SPSS (v.16).

Results 217 assays from 193 patients were included for analysis. 101 patients (52%) were treated for TB (74 pulmonary). A clinical diagnosis of TB was made in In 14 patients. 145 samples (68%) were from the respiratory tract. The remainder were categorised as: fluid (44); tissue (15) and pus (9). Overall 68 (32%) samples were AFB smear positive and 111 (52%) samples were mycobacteria culture positive (80 M.tuberculosis). There were 78 (36%) GeneXpert positive results. The assay had superior performance for diagnosis of TB and predicting M.tuberculosis culture positive outcomes in AFB smear positive compared with smear negative samples (table). For smear negative, culture positive samples, false negative GeneXpert results were associated with a significantly longer time to culture (mean difference 10.4 days, p = 0.006). For smear negative GeneXpert positive samples, the mean time to positive diagnosis was reduced by 13.3 days but this did not alter the time to starting treatment.

Conclusion In our practice, GeneXpert has high specificity to reliably inform a positive TB diagnosis but lacks sensitivity in smear negative disease to reliably exclude the diagnosis. The decision to start treatment continues to be governed by clinical suspicion in this group.