Systematic screening for drug-resistant tuberculosis with Xpert(®) MTB/RIF in a referral hospital in Cambodia.

Limited access to drug susceptibility testing (DST) in referral hospitals contributes to delayed detection of multidrug-resistant tuberculosis (MDR-TB).To document the impact of identifying rifampicin (RMP) resistance using Xpert(®) MTB/RIF on time to diagnosis and time to treatment, and evaluate its performance under programmatic conditions.Using a prospective observational study, we screened presumptive MDR-TB cases with Xpert and solid culture/conventional DST. RMP resistance was confirmed using a line-probe assay (LPA). We recorded diagnostic and treatment delays. We performed rpoB gene sequencing post hoc to resolve discordant RMP susceptibilities.We screened 299 of 345 presumptive MDR-TB individuals, and identified 44 Xpert RMP-resistant cases: 16/165 (10%) were new and 28/136 (20%) retreated. The median time to diagnosis was 2 days (Xpert) vs. an additional 6 with LPA; the median time to treatment was 14 days. Confirmatory LPA on 39/44 revealed 27 concordant, 6 discordant and 6 invalid results. Xpert RMP resistance was confirmed in respectively 24/30 (80%) and 21/23 (91%) by phenotypic DST and rpoB sequencing.Screening presumptive MDR-TB patients with Xpert enabled rapid diagnosis and treatment of MDR-TB. Xpert performed well, provided appropriate risk assessment was done. Rapid confirmatory testing added little to clinical decision making. Our findings support the latest World Health Organization guidelines to abandon confirmatory LPA in favour of repeat Xpert when in clinical doubt, pending phenotypic DST.