[Gaps in drug-resistant tuberculosis surveillance: probabilistic database linkage in Brazil].

The study aimed to assess the coverage and reliability of drug-resistant tuberculosis (DR-TB) case closure in the Information System on Special Treatments for Tuberculosis (SITE-TB in Portuguese) in Brazil from 2013 to 2016, based on probabilistic linkage with the Information System on Diseases of Notification (SINAN), Laboratory Environment Manager (GAL), and Mortality Information System (SIM). The study population consisted of DR-TB cases that initiated treatment from 2013 to 2016 in Brazil. Linkage with SINAN assessed the coverage and estimated underreporting of DR-TB cases. The capture-recapture method was applied, using the Chapman estimator. Linkage with GAL identified cases diagnosed by the laboratory that had not been reported to SITE-TB. Linkage with SIM assessed the reliability of case closure by death in SITE-TB, using the kappa coefficient. We estimated a population of 2,945 (95%CI: 2,365-3,602) new cases of DR-TB with the Chapman estimator. We located 1,626 individuals in the GAL database that had not been reported to SITE-TB, even with laboratory confirmation of drug resistance. PABAK (prevalance and bias adjusted kappa) of 0.86 (95%CI: 0.85-0.87) was classified as excellent for the agreement in death as the outcome between the SITE-TB and SIM databases. The results pointed to persistent gaps related to diagnosis and treatment of DR-TB in Brazil. Underreporting of DR-TB cases in the SITE-TB database poses a challenge for TB control. Identification of these individuals and early start of treatment should be prioritized in health services.