GH015-0009
Modeling Coccidioidomycosis (Valley Fever) in Relation to Climate in Arizona and California

Tuesday, 15 December 2020
Poster
Qianqian Li, University of Nebraska Medical Center, Environmental, Agricultural and Occupational Health, Omaha, NE, United States, Azar Mohammad Abadi, University of Nebraska Medical Center, Environmental, Agricultural & Occupational Health, Omaha, NE, United States, Hongying Dai, University of Nebraska Medical Center, Department of Biostatistics, Omaha, NE, United States and Jesse E Bell, University of Nebraska Medical Center, Environmental, Agricultural, and Occupational Health, Omaha, NE, United States
Abstract:
Coccidioidomycosis (Valley fever) is an infection caused by inhaling soilborne Coccidioides fungus. These fungal spores can become aerosolized and inhaled by people and animals when the soil is disturbed. The disease is mainly endemic to the southwestern United States. Most cases (>95%) are reported in Arizona and California. The variation of Valley fever (VF) incidence in space and time may partially relate to the impacts of changing climate on growth and dissemination of the fungus Coccidioides. This study aims to model and assess spatiotemporal variations of VF incidence and their associations with climate factors in Arizona and California during 2000-2014. We used monthly VF datasets by county for Arizona and California. We calculated the VF incidence in 100,000 based on the population of the county. To evaluate the influence of climate on VF outbreaks, we developed a comprehensive climate dataset from multiple sources that includes monthly values of temperature, precipitation, soil moisture, wind speed, dust storm activity, and Evaporative Demand Drought Index (EDDI). Our correlation matrices showed no strong collinearity among climate variables. These matrices also showed that all the climate variables were significantly associated with VF incidences in AZ, while in CA, only EDDI, wind speed, and soil moisture were significant. We performed multiple linear regression, Bayesian linear regression, and Poisson regression (negative binomial for overdispersion). Negative binomial models in both AZ and CA give the smallest AIC, and hence statistically provide the best fit among the three regression approaches. Soil moistures contribute most to the variation of VF incidence in both AZ and CA. In AZ, VF incidence tended to be greater in areas with drier soils, warmer temperatures, and more frequent dust storms. In CA, EDDI seems to be periodically associated with the fluctuation of VF incidence. The results of this study will help to understand the relationship between climate and Valley fever incidences, as well as their spatiotemporal trends.
<< Previous Abstract | Next Abstract >>