Quantifying gaps in the tuberculosis care cascade in Brazil: A mathematical model study using national program data

In Brazil, many individuals with tuberculosis (TB) do not receive appropriate care due to delayed or missed diagnosis, ineffective treatment regimens, or loss-to-follow-up. This study aimed to estimate the health losses and TB program costs attributable to each gap in the care cascade for TB disease in Brazil. We constructed a Markov model simulating the TB care cascade and lifetime health outcomes (e.g., death, cure, postinfectious sequelae) for individuals developing TB disease in Brazil. We stratified the model by age, human immunodeficiency virus (HIV) status, drug resistance, state of residence, and disease severity, and developed a parallel model for individuals without TB that receive a false-positive TB diagnosis. Models were fit to data (adult and pediatric) from Brazil's Notifiable Diseases Information System (SINAN) and Mortality Information System (SIM) for 2018. Using these models, we assessed current program performance and simulated hypothetical scenarios that eliminated specific gaps in the care cascade, in order to quantify incremental health losses and TB diagnosis and treatment costs along the care cascade. TB-attributable disability-adjusted life years (DALYs) were calculated by comparing changes in survival and nonfatal disability to a no-TB counterfactual scenario. We estimated that 90.0% (95% uncertainty interval [UI]: 85.2 to 93.4) of individuals with TB disease initiated treatment and 10.0% (95% UI: 7.6 to 12.5) died with TB. The average number of TB-attributable DALYs per incident TB case varied across Brazil, ranging from 2.9 (95% UI: 2.3 to 3.6) DALYs in Acre to 4.0 (95% UI: 3.3 to 4.7) DALYs in Rio Grande do Sul (national average 3.5 [95% UI: 2.8 to 4.1]). Delayed diagnosis contributed the largest health losses along the care cascade, followed by post-TB sequelae and loss to follow up from TB treatment, with TB DALYs reduced by 71% (95% UI: 65 to 76), 41% (95% UI: 36 to 49), and 10% (95% UI: 7 to 16), respectively, when these factors were eliminated. Total health system costs were largely unaffected by improvements in the care cascade, with elimination of treatment failure reducing attributable costs by 3.1% (95% UI: 1.5 to 5.4). TB diagnosis and treatment of false-positive individuals accounted for 10.2% (95% UI: 3.9 to 21.7) of total programmatic costs but contributed minimally to health losses. Several assumptions were required to interpret programmatic data for the analysis, and we were unable to estimate the contribution o