BACKGROUND: Temperature is one of the most important factors in determining West Nile virus (WNV) transmission risk to humans and is readily accessible. Therefore using temperature data would be useful to determine potential high risk WNV transmission periods particularly in areas where mosquito or bird surveillance data is lacking. This study compared two simple temperature degree day (DD) threshold models that were created to determine high risk periods of human WNV transmission at the county level in 12 Nebraska counties.
METHODS: Human case data for 2011-2105 was obtained from ArboNet. The ten counties with the most human WNV cases over the five years and at least one county from each of Nebraska’s NOAA climate divisions were used in the study. For each county an accumulated 109 DD and 90 DD threshold model was created and compared. Both models used a minimum base temperature for viral replication of 14.3° C. For the 109 DD model, a mosquito extrinsic incubation period (EIP) threshold of 109 accumulated DDs for median virus transmission by Culex tarsalis mosquitoes (Nebraska’s primary WNV vector) was selected based on Reisen et al. (2006). The 109 DD model was then recalibrated using visual inspection to fit the human case data, lowering the DDs required to reach the mosquito EIP threshold to 90 DDs. A human case was considered correctly predicted if it occurred 2-14 days before the onset date and at least one day during this time accumulated enough DDs to reach the mosquito EIP threshold.
RESULTS: The 109 DD model correctly predicted 79.1% of the human cases occurring during a high risk period from the 12 counties. Individual county results ranged from 38.5%—100.0% with a median of 78.5% of human cases being correctly predicted. Use of the 90 DD model resulted in an increase of accuracy of 92.1% of human cases correctly predicted during high risk periods. Improved accuracy was seen in 10 of the 12 counties with individual county results ranging from 69.2%—100.0% with a median of 93.1%.
CONCLUSIONS: The 109 DD model underestimated the number of WNV high risk transmission days resulting in lower accuracy at predicting human cases. The recalibrated 90 DD model improved overall case accuracy to greater than 90% and improved accuracy in most of the counties observed. This study demonstrates that a simple temperature degree day model can be used to help determine potential high risk human WNV transmission periods.