![[ Visit Client Website ]](images/banner.jpg)
BACKGROUND: Temperature and rainfall are among the factors that influence mosquito-borne disease incidence. El Niño Southern Oscillation (ENSO) is a global climate driver that affects regional weather in the U.S. This analysis characterizes the relationship between neuroinvasive West Nile Virus (WNV) disease cases, extreme heat and precipitation, and ENSO, from 2001-2013 by U.S. region.
METHODS: Monthly WNV disease case counts from ArboNET and annual state population estimates from the U.S. Census Bureau were aggregated into seven regions within the continental U.S. Monthly regional extreme weather indices were calculated from gridded temperature and precipitation data. Extreme heat (HI) and precipitation (PI) indices were categorized as non-extreme (<1.5 standard deviations (SD) above mean), extreme (1.5-2 SD above mean), or very extreme (≥2 SD above mean). Monthly global climate driver index values were obtained from the National Oceanic and Atmospheric Administration. Extreme weather and ENSO associations with WNV disease case counts were assessed with Poisson regression for the U.S. and regionally. Predictor variables included year, month (June-October), Niño 3.4 anomaly (1-month lag), and HI and PI (1- and 3-month lags). Rate ratios (RR) and p-values are reported.
RESULTS: Month (August) and year (national outbreaks) were the strongest predictors of WNV disease cases in the U.S. (August RR=22.72, p<0.0001; 2002 RR=3.62, p<0.0001) and regionally. However, monthly Niño 3.4 anomaly, HI, and PI were statistically significant, even when controlling for year and month. Positive ENSO decreased WNV risk in the U.S. (RR=0.54, p<0.0001), South East, East North Central, and Mountain regions. Negative ENSO decreased risk in West North Central and Mountain regions and increased risk in West South Central and Pacific regions. Both 1- and 3-month HI increased WNV risk nationally (RR ≥1.27, all p-values <0.0001). Positive associations were seen with 1-month HI in all regions except the South East and with 3-month HI in the western U.S. Both 1- and 3-month PI decreased WNV risk nationally (RR≤0.84, all p-values <0.0001). Negative associations with 1-month PI were seen in central and western regions and with 3-month PI in North East and Mountain regions.
CONCLUSIONS: The U.S. model was least suited for the South East (no extreme weather indices significant) and best suited for the Mountain and Pacific regions (all predictor variables significant). Results align with published literature but include more recent data and demonstrate that associations hold true at both regional and national levels. Next steps include reclassifying the PI to define optimal precipitation.