BackgroundDelays in tuberculosis (TB) diagnosis, particularly in prisons, is associated with detrimental outcomes. The new GeneXpert MTB/RIF assay (Xpert) offers accurate and rapid diagnosis of active TB, but its performance in improving case detection in high-transmission congregate settings has yet to be evaluated. We assessed the diagnostic accuracy of a single Xpert assay in an intensified case finding survey among HIV-infected prisoners in Malaysia.MethodsHIV-infected prisoners at a single site provided two early-morning sputum specimens to be examined using fluorescence smear microscopy, BACTEC MGIT 960 liquid culture and a single Xpert. The sensitivity, specificity, negative and positive predictive values of Xpert were calculated relative to gold-standard results using MGIT 960 liquid culture. Relevant clinical and demographic data were used to examine correlates of active TB disease.ResultsThe majority of enrolled subjects with complete data (N=125) were men (90.4%), age <40 years (61.6%) and had injected drugs (75.2%). Median CD4 lymphocyte count was 337 cells/µL (IQR 149-492); only 19 (15.2%) were receiving antiretroviral therapy. Of 15 culture-positive TB cases, single Xpert assay accurately detected only eight previously undiagnosed TB cases, resulting in a sensitivity, specificity, positive predictive value and negative predictive value of 53.3% (95% CI 30.12-75.2%), 100% (95% CI 96.6-100%), 100% (95% CI 67.56-100%) and 94.0% (95% CI 88.2-97.1%), respectively. Only 1 of 15 (6.7%) active TB cases was smear-positive. The prevalence (12%) of undiagnosed active pulmonary TB (15 of 125 prisoners) was high and associated with longer duration of drug use (AOR 1.14, 95% CI 1.03-1.26, for each year of drug use).ConclusionsSingle Xpert assay improved TB case detection and outperformed AFB smear microscopy, but yielded low screening sensitivity. Further examination of the impact of HIV infection on the diagnostic performance of the new assay alongside other screening methods in correctional settings is warranted.