Underreporting of tuberculosis cases from death surveillance

The objective was to analyze underreporting of tuberculosis cases from the capture of deaths registered at the Mortality Information System (SIM) from 2007 to 2011. We conducted a retrospective study in a city at the Northeast region of Brazil using a non-probabilistic association technique between SIM registries and the Information System on Diseases of Compulsory Declaration (Sinan), through the software Reclink III. We calculated odds ratios (OR) of individuals who died from tuberculosis not reported at Sinan, according sociodemographic characteristics. From 147 deaths registered at SIM, 72.8% were not notified at Sinan. Women were more likely to not be reported than men (OR = 2.60), people younger than 60 years (OR = 1.86), with less than eight years of education (OR = 4.42). Strategies should be re-signified in a way to capture more respiratory symptoms, to opportunely diagnose cases and to avoid flaws when completing registries.