Unsuccessful treatment in pulmonary tuberculosis: factors and a consequent predictive model.

Background Cure is particularly valuable in pulmonary cases (PTB), as unsuccessful treatment fuels incidence and resistance to antibiotics. This study aims to identify individual factors of PTB unsuccessful treatment in Portugal and to develop a consequent predictive model. Methods Using the Portuguese TB surveillance database (SVIG-TB), PTB cases older than 15 years notified from 2000 to 2012 in Continental Portugal were analyzed. Unsuccessful treatment included the WHO categories (failure, default, death and transferred out). Based on a literature review, predictors involved sociodemographic, behavioral, disease-related and treatment-related factors. Binary logistic regression was used to estimate unsuccessful treatment factors and to develop the predictive risk model. Results The unsuccessful outcome rate in PTB patients was of 11.9%. The predictive model included the following factors: TB/HIV co-infection (OR 4.93), age over 64 years (OR 4.37), IV drugs abuse (OR 2.29), other diseases (excluding HIV and Diabetes, OR 2.09) and retreatment (OR 1.44), displaying a rather good validity. Conclusion The overall treatment unsuccessful treatment rate in PTB patients complies with the 85% WHO success threshold. The predictive model of unsuccessful treatment proved well. Nomogram representation allows an early, intuitive identification of PTB patients at increased risk. The model is liable to widespread use as a prognostic tool.