Predictors of mortality and treatment success during treatment for rifampicin-resistant tuberculosis within the South African National TB Programme, 2009 to 2011: a cohort analysis of the national case register

Background: The South African Electronic Drug-Resistant Tuberculosis Register (EDRweb) is the national database of registered drug-resistant tuberculosis (DR-TB) cases. Methods: This study was a retrospective, de-identified secondary analysis of EDRweb patients initiating treatment for rifampicin-resistant TB (January 2009 to September 2011). The relative risks of death and treatment success were estimated using modified Poisson regression with robust error estimation. Results: Seventeen thousand six hundred and ninety-seven cases of DR-TB were registered and met the inclusion criteria; 52.0% (n = 9207) were male and the median age was 35 years (interquartile range 27–43 years). Of the 9419 cases with HIV infection (53.2%), 7157 (76.0%) were on antiretroviral therapy. Most had undergone previous TB treatment (76.5%, n = 13 531). Multidrug-resistant TB was the most common diagnosis, at 80.6% (n = 14 272). No treatment outcome was available for 6934 patients (39.2%). For patients with outcomes, 4227 (39.4%) were successfully treated, 2987 (27.8%) died, 2533 (23.7%) were lost to follow-up, and 996 (9.3%) failed. Second-line drug resistance was the strongest predictor of death during DR-TB treatment; extensively drug-resistant TB patients were more likely to have died during treatment (adjusted relative risk 2.63, 95% confidence interval 2.45–2.84). Conclusions: Testing for second-line drug resistance at initiation of DR-TB treatment can identify patients most at risk of treatment failure and death and most in need of individualized treatment