Doença de Chagas

Abstract

Chagas Disease (CD), caused by the protozoan Trypanosoma cruzi (T. cruzi), is considered one of the thirteen main neglected tropical diseases that still afflicts many Latin American countries, including Brazil. This thesis presents five different studies about CD. 1) The first study aimed to develop models and evaluate different machine learning (ML) techniques to predict death within two years of patients with CD. The data came from a multicenter cohort study called SaMi-Trop (Research Center for Tropical Medicine São Paulo-Minas Gerais). About 2000 individuals with CD from 21 endemic municipalities in the North of Minas and Vale do Jequitinhonha, have been monitored since 2014 by SaMi-Trop. The predictor variables came from the baseline and the outcome, which was death, came from the first follow-up. The outcome prediction was performed with three different ML configurations: only with interview variables, only with complementary exam variables and, finally, with interview variables and complementary exam variables. All models were evaluated in terms of Sensitivity, Specificity and G-Mean. Among the 1959 individuals with CD considered, 134 (6.8%) died within two years of follow-up. The different ML techniques achieved an accuracy in predicting death from 62% to 78%. 2) The second study aimed to investigate the influence of contextual and individual factors on self-perceived health in CD patients. This is a cross-sectional multilevel study. Individual data came from the 1st follow-up of the SaMi-Trop cohort study. Contextual data relating to the 21 municipalities included in the SaMi-Trop were also collected from public databases of the Brazilian government. The dependent variable was self-perceived health (positive vs negative). Hierarchical multilevel logistic regression was conducted. It was observed that 22% of patients with CD had negative self-perception. There was a lower chance of negative self-perception among those who lived in cities with a smaller population (OR=0.600), and a greater chance among those who lived in cities with fewer physicians/inhabitants (OR=1.512) and who lived in cities with a higher rate of illiteracy (OR=1.558). Among the individual variables, there was a greater chance of negative self-perception among those with functional limitations (OR=2,000), who had an altered NT-pro BNP level (OR=1.911), with arterial hypertension (OR=1,500), with lower income (OR=1.523), who lived more than 100 km from the Health Unit (OR=2.529), who reported irregular monitoring of the Family Health Strategy (FHS) (OR=1.752), who did not practice physical activity (OR=1.853) and who smoked (OR=2.261). 3) The third study aimed to investigate the contribution of the social context in the occurrence of cardiovascular events in CD, also using multilevel modeling. Individual data came from the SaMi-Trop cohort and contextual data from government public access platforms. The outcome was the occurrence of cardiovascular events (death, atrial fibrillation and/or pacemaker implantation) considering two years of follow-up. Hierarchical multilevel logistic regression was conducted. It was observed that 12.5% of patients with CD manifested new cardiovascular events within a 2-year follow-up period. There was protection for cardiovascular events among those who lived in municipalities with a larger rural population (OR=0.509); and greater chance among those residing in municipalities with fewer physicians/inhabitants (OR=1.698) and residing in municipalities with lower FHS coverage (OR=1.468). A greater chance of cardiovascular events was also observed among the elderly (OR=1.474), without stable union (OR=1.420), who had not previously used Benznidazole (BZN) (OR=1.599), with worse functional class (OR= 2,007), who had a longer QRS complex duration (OR=1.583) and had an altered NT-proBNP level (OR=6.424). 4) The fourth study of this thesis aimed to understand the medical practice and management of patients with CD in the scope of Primary Health Care (PHC) in endemic regions. Descriptive cross-sectional study conducted with 104 physicians from 39 different municipalities in the North and Vale do Jequitinhonha macro-regions in Minas Gerais. Physicians had a mean age of 33(±9.88) years, 4(±7.26) years of experience in the PHC, 49% reported that their graduation did not offer sufficient training in CD. Insecurity, lack of knowledge and lack of training on CD were observed. 5) This last study addressed adverse effects related to the treatment of CD with BZN. A systematic review of randomized controlled clinical trials was conducted. Five studies were included and there were reports of adverse reactions to BZN in all. Adverse reactions vary according to the patient's age and therapeutic regimen adopted (dose and duration). The adverse reaction rate reached 38% and the most common outcome was skin rash. Conclusion: This thesis showed that it is possible to achieve models with a high prediction of death in patients with CD, through LM techniques. It was also found that individual conditions are not isolated protagonists of conditions related to the health of patients with CD, and the context in which individuals live is also a determining factor. He also observed that there is a failure in medical training regarding CD and that the prescription of BZN was infrequent and permeated with insecurity. And finally, it revealed that adverse reactions to the use of BZN vary according to the patient's age and therapeutic dose regimen and adopted time.