Chagas disease mortality in Brazil: A Bayesian analysis of age-period-cohort effects and forecasts for two decades

Abstract

Background: Chagas disease (CD) is a neglected chronic parasitic infection and a public health problem that is preventable, and has serious complications. In this study, the effects of age, period and birth cohort (APC Effects) on the evolution of the mortality of that disease in Brazil, from 1980-2014, according to sex and geographic region of the country, were analyzed. Mortality forecasts from the years 2015 to 2034 were estimated. Methods: This is an ecological cross-sectional study in which death records and population data were extracted from the DATASUS (Department of Information Technology of the National Health System) website, in age groups from 20-24 years of age to 80 years and over, from 1980 to 2014. The rates were standardized according to age and sex distributions using the direct method. The APC models were estimated using the Bayesian approach, and the INLA (Integrated Nested Laplace Approximations) method was used for parameter inference. Super dispersion of the data was considered, and we included unstructured random terms in the models. Results: During the analyzed period, there were 178,823 deaths in Brazil (3.85 annual deaths per 100,000 inhabitants). It was found that temporal effects on CD mortality varied by sex and region. In general, there was an increase in mortality rates up to 30 years of age, and the mortality rates were higher between 50 and 64 years of age. On average, men died five years younger than women. Mortality rates were highest in the Central West and Southeast regions. The Central West, Southeast and Southern regions had a reduction over time in the rate of CD deaths between 2000 and 2014. The mortality rate in the Northeast was not statistically different in any period analyzed, while the North had tendency to increase; however, a significant risk increase was only observed between 1995 and 1999. The rate of mortality was high in older birth cohorts. The overall prediction for the next two decades showed a progressive decline in CD mortality, which will be highest among the young. The expected average reduction was 76.1% compared to the last observed period (2010-2014) and the last predicted (2030-2034) period. The average reduction ranged from 86% in the 20-24 age group to 50% in the 80 and over age group. Conclusions: In the present study, a higher death rate was observed for ages above 30 years, especially 50 to 64 years, and in the older birth cohorts. We believe these results can be related to period effects of vector control actions and preventive and care measures by the health system of Brazil, in addition to demographic changes in the period. The differences among the regions reflect socioeconomic inequities and access to the healthcare systems in the Brazilian population.