Predictive model to the tuberculosis treatment abandonment

This study aimed to build a predictive model taking as explanatory variables the characteristics of individuals with tuberculosis so to define the most appropriate treatment. A descriptive observational research of quantitative approach was applied. The target population (n=1925) was collected from the Reportable Diseases Information System of João Pessoa-PB, Brazil, between 2001 and 2008. In the final model odds ratio (OR) the following quitting treatment risk variables were considered: education in high school (OR=1.56) and in college (OR=42.99), race (OR=2.29) and re-entry after quitting (OR=2.71). The decision-making process should be directed to the best fit model in order to identify likely-to-quit-treatment individuals.