Impact of Xpert MTB/RIF and decentralized care on linkage to care and drug-resistant tuberculosis treatment outcomes in Johannesburg, South Africa

BACKGROUND: In 2011, South Africa improved its ability to test for rifampicin-resistant TB (RR-TB) by introducing GeneXpert MTB/RIF. At the same time, the South African National TB program adopted a policy decentralized, outpatient treatment for drug resistant (DR-) TB. We aim to analyze the impact of these changes on linkage to care and DR-TB treatment outcomes. METHODS: We retrospectively matched adult patients diagnosed with laboratory-confirmed RR-TB in Johannesburg from 07/2011–06/2012 (early cohort) and 07/2013–06/2014 (late cohort) with records of patients initiating DR-TB treatment at one of the city’s four public sector treatment sites. We determine the proportion of persons diagnosed with RR-TB who initiated DR-TB treatment and report time to treatment initiation (TTI) before and after the implementation of Xpert MTB/RIF roll-out in Johannesburg, South Africa. We conducted a sub-analysis among those who initiated DR-TB treatment at the decentralized outpatient DR-TB centers to determine if delays in treatment initiation have a subsequent impact on treatment outcomes. RESULTS: Five hundred ninety four patients were enrolled in the early cohort versus 713 in the late cohort. 53.8 and 36.8% of patients were diagnosed with multi-drug resistant TB in the early and late cohorts, respectively. The proportion of RR-TB confirmed cases diagnosed by Xpert MTB/RIF increased from 43.4 to 60.5% between the early and late cohorts, respectively. The proportion who initiated treatment increased from 43.1% (n = 256) to 60.3% (n = 430) in the late cohort. Pre-treatment mortality during the early and the late cohort reduced significantly from 17.5 to 5.8% while lost to follow-up remained high. Although TTI reduced by a median of 19 days, from 33 days (IQR 12–52) in the early cohort to 14 days (IQR 7–31) in the late cohort, this did not translate to improved treatment outcomes and we found no difference in terms of treatment success or on-treatment mortality for those that initiated without delay vs. those that deferred initiation. CONCLUSION: Pre-treatment mortality reduced significantly during late Xpert MTB/RIF coverage but there was no significant difference after treatment was initiated. Despite improvements there is still a significant diagnosis and treatment gap for patients diagnosed with RR-TB and improving treatment outcomes remains critical. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1186/s12913-018-3762-x) contains supplementary material, which is available to authorized users.