Research and Products:
Predicting Valley Fever Incidence

Status of Project

Two goals of this research are to determine basic relationships between climate and valley fever incidence and to develop predictive models based on climate variables.  Exploratory analysis provided insight into the conditions up to four years prior to a month with high or low incidence. This process also aided in the selection of candidate variables for the multivariate models. Predictive models were developed using backward stepwise regression, and incorporated temperature and precipitation variables at varying periods prior to the month being predicted. The resulting models included variables that were mainly from periods of more than one year prior to the month being predicted. Also, winter climate conditions appear to be important incidence predictors, as winter temperature and precipitation variables frequently appear in the models. Months with the highest percentage of total annual incidence have the best-performing models. Therefore, valley fever incidences in the months that experience the greatest number of cases are most predictable.

Model results indicate strong bimodal seasonality of the disease in Pima County, Arizona. Peaks in exposure to fungal spores occur in June–July and in October–November, generally drier and dustier months (Figure 6). Fewer exposures occur during the wetter and less dusty months of February–March and August–September. Additionally, one of the most important factors in valley fever incidence is precipitation in the arid foresummer 1.5–2.0 years before exposure. Above-average precipitation during this period generally increases chances of valley fever incidence.

Figure 6. Mean monthly coccidioidomycosis incidence. [more details]

 

 

 

 

 

 

 

< Research Methods

References >