Predictive model of unfavorable outcomes for multidrug-resistant tuberculosis

• Autor(a) principal: Arroyo,Luiz Henrique
• Data de Publicação: 2019
• Outros Autores: Ramos,Antônio Carlos Vieira, Yamamura,Mellina, Berra,Thais Zamboni, Alves,Luana Seles, Belchior,Aylana de Souza, Santos,Danielle Talita, Alves,Josilene Dália, Campoy,Laura Terenciani, Arcoverde,Marcos Augusto Moraes, Bollela,Valdes Roberto, Bombarda,Sidney, Nunes,Carla, Arcêncio,Ricardo Alexandre
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Revista de Saúde Pública
• Texto Completo: http://old.scielo.br/scielo.php?script=sci_arttext&pid=S0034-89102019000100269
• topic: Tuberculosis, Multidrug-Resistant, complications; Tuberculosis, Multidrug-Resistant, mortality; Risk Factors; Treatment Adherence and Compliance

ABSTRACT OBJECTIVE to analyze the temporal trend, identify the factors related and elaborate a predictive model for unfavorable treatment outcomes for multidrug-resistant tuberculosis (MDR-TB). METHODS Retrospective cohort study with all cases diagnosed with MDR-TB between the years 2006 and 2015 in the state of São Paulo. The data were collected from the state system of TB cases notifications (TB-WEB). The temporal trend analyzes of treatment outcomes was performed through the Prais-Winsten analysis. In order to verify the factors related to the unfavorable outcomes, abandonment, death with basic cause TB and treatment failure, the binary logistic regression was used. Pictorial representations of the factors related to treatment outcome and their prognostic capacity through the nomogram were elaborated. RESULTS Both abandonment and death have a constant temporal tendency, whereas the failure showed it as decreasing. Regarding the risk factors for such outcomes, using illicit drugs doubled the odds for abandonment and death. Besides that, being diagnosed in emergency units or during hospitalizations was a risk factor for death. On the contrary, having previous multidrug-resistant treatments reduced the odds for the analyzed outcomes by 33%. The nomogram presented a predictive model with 65% accuracy for dropouts, 70% for deaths and 80% for failure. CONCLUSIONS The modification of the current model of care is an essential factor for the prevention of unfavorable outcomes. Through predictive models, as presented in this study, it is possible to develop patient-centered actions, considering their risk factors and increasing the chances for cure.