The burden of Yellow Fever in Brazil: Quantifying disease mortality and producing short-term forecasts

Abstract

Yellow Fever is a mosquito-borne viral disease impacting much of South America and sub-Saharan Africa. It is endemic in several nations, causing hundreds of thousands of annual infections and tens of thousands of deaths. In Brazil, cases in recent decades have been seen in western areas of the nation, typically in regions adjacent to the Amazon Rainforest. A major outbreak beginning in December 2016, however, saw a major increase in cases, particularly in southeastern states. As a result, there is interest in finding ways to predict when and where future Yellow Fever cases are expected. Also of interest is the ability to anticipate future fatalities by finding the proportion of cases that are fatal. Several mechanisms influence risk of Yellow Fever cases, including human activities and environmental conditions. The latter is comprised of many characteristics outside of human control, representing a component of risk that cannot be targeted for direct intervention, but only used for preparations. Using environmental conditions to predict future Yellow Fever burden typically employs the use of either mathematical or statistical models to quantify relationships between environmental predictors and disease burden. In developing such models, several assumptions are made out of necessity. One particular assumption, relating to the time units used in developing a model, is not commonly investigated for its potential impact on describing disease dynamics. In order to investigate these various topics, four studies were conducted in order to predict Yellow Fever burden using various environmental conditions, estimate fatality risk among severe Yellow Fever cases, and examine assumptions of time unit sizes when describing disease incidence probabilistically. The findings of the various studies show sensitivity when changing time unit sizes, offer an update of the estimated fatality risk among severe Yellow Fever cases, and estimate potential for Yellow Fever case burden throughout Brazil both using annual environmental trends and weekly weather patterns. Methodological contributions are discussed.