ABSTRACT
BACKGROUND: Xpert MTB/RIF (Xpert) was piloted in Kazakhstan to detect tuberculosis (TB) and rifampicin resistance (RR-)TB among individuals at risk of multidrug-resistant (MDR-) TB. This study assessed the performance of Xpert compared to conventional diagnostic methods, RR-TB case detection among various risk groups, treatment initiation and time to diagnosis and treatment. METHODS: Eligible individuals were tested with Xpert, smear microscopy, culture and drug-susceptibility testing (DST) at the national TB reference laboratory and three provincial laboratories. Data was collected prospectively from August 2012 to May 2013 from routine laboratory and treatment registers. RESULTS: A total of 5,611 Xpert tests were performed mostly targeting contacts of MDR-TB patients, 'other' presumptive MDR-TB patients, and retreatment cases (26%, 24% and 22%, respectively). Compared to phenotypic DST, the positive predictive value of Xpert to detect RR-TB was 93.1% and 96.4% and the negative predictive value was 94.6% and 92.7% using solid and liquid culture media, respectively. RR-TB detection was highest among (former) prisoners, retreatment cases, people living with HIV/AIDS (PLWHA), and TB patients with positive smears after intensive phase of treatment (59%, 58%, 54% and 53% among TB positives, respectively). 88.9% of RR-TB patients were registered to have started second-line TB treatment. Median time to diagnosis with Xpert was 0.0 days (IQR 0.0-1.0), time from diagnosis to start of first-line treatment 3.0 days (IQR 1.0-7.0), and to start of second-line treatment 7.0 days (IQR 4.0-16). CONCLUSIONS: Compared to conventional culture and DST, Xpert had a shorter result turn-around-time and excellent concordance to detect RR-TB. Time from sputum collection to start of second-line treatment was reduced to one week. The yield of Xpert could be maximized by increasing referrals from penitentiary and HIV centers to TB centers.
