To investigate the sensitivity of NNDT method, we repeated the calculations with human cases reported at several random locations within a particular zip code.
Hence we can conclude that the NNDT methodology is sensitive to the number and density of infected cases.
We also analyzed the results of NNDT when maximum and minimum distances were used between the centroid of zip codes and the location of infected dead birds and positive mosquito pools.
The NNDT identified zero transmission cycles when the distance between the location of infected dead birds, positive mosquito pools, and human cases at the centroid of the reported zip codes was minimum.
Using the maximum distance as a threshold resulted in more centroids of zip codes fulfilling the spatial rule of NNDT by connecting to both the locations of infected dead birds and positive mosquito pools in 2004, 2005, and 2006.
Using maximum distance did result in more spider webs with both dead birds and mosquito pools but it did not satisfy the temporal criteria of the NNDT model.
We demonstrated the use of NNDT as a new approach to delineate WNV exposure areas.
The results obtained with NNDT provide evidence that dead bird reports are an essential part of the delineation of WNV cycles on a localized scale.
The NNDT methodology, for example, found that the WNV transmission cycle in North Saint Paul in 2004 remained active for 12-16 days prior to the date of human illness and then subsided with a substantial decrease in dead bird reporting.