Effect of introducing ethambutol and integrating drugs into fixed-dose tablets on mortality in patients with tuberculosis

Abstract

At the end of 2009, due to the increase in primary resistance to isoniazid, the Brazilian Ministry of Health established changes in the treatment regimen for tuberculosis. The changes included the addition of ethambutol as the fourth drug in the intensive treatment phase and the integration of the four drugs into fixed-dose combination tablets. The introduction of fixed combination doses also led to changes in the dosage of isoniazid and pyrazinamide in the intensive and maintenance phases. Objective: To estimate the effect of changing the tuberculosis treatment regimen on all-cause mortality and, secondly, outcomes such as mortality due to tuberculosis, cure, and loss to follow-up. Methods: We compared the cohorts of people diagnosed with tuberculosis from 2008 to 2013, aged ten years or older, who started treatment for tuberculosis in Brazil before and after the change in the regimen adopted in 2009 (n: 145528 vs. 161264). Data were extracted from the Notifiable Diseases (Sinan) and Mortality (SIM) information systems. The missing data were imputed, and the effects were estimated using multilevel logistic models, with the state as the aggregation cluster. A directed acyclic graph guided the selection of covariates. Results: The current (modified) regimen was not associated with significant changes in all-cause mortality (Relative Risk [RR]: 1.01; 95% confidence interval [95%CI]: 0.98 - 1.04), or tuberculosis mortality (RR: 0.98; 95%CI: 0.95 - 1.02). For cure, when transfers and missing outcome data (MOD) were considered an absence of outcome, there were no differences between treatments. When they were assumed as cured or imputed, the cure was less frequent in the current treatment. When transfers and MOD were imputed or considered an absence of loss to follow-up, the latter was more frequent in the current treatment. There were no differences in loss to follow-up between treatments when transfers and MOD were interpreted as that outcome. Conclusion: The implementation of the modified treatment regimen was not associated with increased mortality in tuberculosis patients. Although there was a lower record of cures and a higher frequency of loss to follow-up during the second period, the sensitivity analysis indicated that a reduction in transfers and unknown outcomes could explain these associations.