Comparative Effectiveness Of Bedaquiline On One-Year Mortality In Rifampicin-Resistant Tuberculosis: A Target Trial Emulation

BackgroundThree phase II clinical trials generated the evidence for recommending bedaquiline for the treatment of rifampicin-resistant tuberculosis (RR-TB). These trials were not powered to assess the effect of bedaquiline on mortality. Observational studies reported lower mortality in patients treated with bedaquiline-containing regimens but did not fully account for differences between patients who did and did not receive bedaquiline in the real world. MethodsUsing data from two studies on 622 patients, of whom 195 initiated a bedaquiline-containing regimen, we applied rigorous causal inference by emulating a trial that would randomize patients diagnosed with RR-TB by the Xpert MTB/RIF assay to a bedaquiline-containing regimen or a non-bedaquiline-containing regimen. We used multiple imputation to address missing data, inverse probability of treatment weighting (IPTW) to emulate randomized assignment and estimated the odds of one-year mortality using a marginal structural logistic model. ResultsBy using IPTW, we achieved conditional exchangeability for observed differences in age, gender, HIV status, Mycobacterium tuberculosis resistance pattern, and history of tuberculosis treatment between patients who did or did not initiate a bedaquiline-containing regimen. By emulating the design of a randomized trial, we found that had all patients been treated with a bedaquiline-containing regimen, there would have been a 67% reduction in the odds of one-year mortality compared to when none of the patients initiated a bedaquiline-containing regimen (OR: 0{middle dot}33, 95%CI: 0{middle dot}19-0{middle dot}59) ConclusionBy emulating a randomized trial using real-world data, our results demonstrate that the initiation of a bedaquiline-containing regimen causes a 67% reduction in the odds of one-year mortality. Key messageWe assessed the causal effect of initiating a bedaquiline-containing regimen compared to a non-bedaquiline-containing regimen on one-year mortality. We found that a bedaquiline-containing regimen causes a 67% reduction in the odds of one-year mortality, underscoring the need for expanded access to such effective regimens.