Modelling Dengue Transmission Risk in Central and South America using Climate Data

Thesis (Master's)--University of Washington, 2020Objective: The principal objective of this project was an assessment of the seasonal change in the geographical risk and human population at risk of dengue virus (DENV) infection, which causes dengue fever (DF). This was accomplished through the development of a DENV transmission suitability model, using the daily survival probability of the mosquito vector and the virus extrinsic incubation period as a function of temperature and humidity, and dengue case data in Brazil, Colombia, and Costa Rica. Methods: This descriptive study design was a secondary longitudinal data analysis of data for confirmed DENV infections and meteorological conditions in Brazil, Colombia, and Costa Rica from 2012 until 2018. The presence of dengue transmission was made binary in the study based on the human incidence rate within the countries. A logistical regression was utilized to quantify the relationship between dengue incidence and the modeled probability that the DENV carrying mosquito, Aedes aegypti, will survive the extrinsic incubation period. This probability will be referred to as the risk index (RI) and was calculated using temperature and relative humidity. A logistic model was created by using the weekly RI and the dengue incidence rate at each respective Administrative Level 1 unit to calculate the probability that the conditions would support DENV transmission. Results: The output of the individual country models, as well as, the combined model provided several statistically significant findings. All the models had at least a 70% accuracy, except for Costa Rica. Each of the models quantified statistical significance among the dependent (human DENV cases) and independent variables (RI). For every 0.01 increase in RI (probability the mosquito will survive past the incubation period), the odds of dengue increased by 12.6% (Brazil), 23.6% (Columbia), 69.5% (Costa Rica), and 23% (Combined). Interpretation of the combined model was that for every 0.01 increase in RI, the odds of the presence of DENV increased by 23%. Conclusions: The model results suggest a positive association between favorable environmental conditions and DENV incidence, which is of public health importance for the general population and for the most vulnerable populations. The ability to forecast favorable environmental conditions may aid in appropriate future DENV resource allocation and planning not only in these countries but throughout the region. Applicable mitigation and adaptation strategies must be developed and implemented to address the increasing incidence of DENV