Bedaquiline versus placebo for management of multidrug-resistant tuberculosis: A systematic review

Sir, Charan et al. compared time to conversion of positive sputum to negative and mortality between bedaquiline and placebo. In all the forest plots as heterogenicity, I2 = 0% or nonsignificant, they should have pooled data under fixed‐effect model rather than using random effect model. We have certain observations that need clarifications. Authors’ conclusion with regard to increased mortality associated with bedaquiline cannot be ascertained as the authors of the bedaquiline trial[1] stated that no causal pattern was evident to bedaquiline use. Moreover, mortality in a meta‐analysis[2] involving 9153 patients with multidrug‐resistant tuberculosis (MDR‐TB) was 15% and in another open‐label, phase II trial[3] of bedaquiline in 233 patients with newly diagnosed or previously treated MDR‐TB was 7%. In a study by Diacon et al., there were significantly more number of HIV‐positive patients in placebo group (P = 0.04). We expect that antiretroviral therapy would have resulted in reduced mortality in placebo group as compared to bedaquiline.[4,5] Not only there was increased background resistance to second‐line drugs in bedaquiline group, but also significantly more number of patients in the placebo group had at least one new antitubercular drug added to their background regimen (58% vs. 47%). Finally, comparing mortality in phase II trial of bedaquiline[1] would have resulted in an alpha error of 0.45, i.e., 45% as mortality was the ninth comparison (two analysis for sputum conversion, five subgroups, and one‐drug resistance analysis). Thus, concluding falsely decreased Mortality in placebo and increased mortality in bedaquiline group in this meta‐analysis by Charan et al. without considering the above‐stated factors is not warranted. Considering 15% mortality in MDR‐TB[2] from such a large patient pool it becomes difficult to explain 2% mortality in placebo groups of the studied selected for this meta‐analysis. Thus, we assume that the authors’ conclusion of bedaquiline associated increase in mortality was based on wrong interpretation of published literature.