Predictive Mapping of Human Risk for West Nile Virus (WNV) Based on Environmental and Socioeconomic Factors

PLOS ONE, August 2011

21853103

Ilia Rochlin, David Turbow, Frank Gomez, Dominick V. Ninivaggi, Scott R. Campbell

A West Nile virus (WNV) human risk map was developed for Suffolk County, New York utilizing a case-control approach to explore the association between the risk of vector-borne WNV and habitat, landscape, virus activity, and socioeconomic variables derived from publically available datasets. Results of logistic regression modeling for the time period between 2000 and 2004 revealed that higher proportion of population with college education, increased habitat fragmentation, and proximity to WNV positive mosquito pools were strongly associated with WNV human risk. Similar to previous investigations from north-central US, this study identified middle class suburban neighborhoods as the areas with the highest WNV human risk. These results contrast with similar studies from the southern and western US, where the highest WNV risk was associated with low income areas. This discrepancy may be due to regional differences in vector ecology, urban environment, or human behavior. Geographic Information Systems (GIS) analytical tools were used to integrate the risk factors in the 2000-2004 logistic regression model generating WNV human risk map. In 2005-2010, 41 out of 46 (89%) of WNV human cases occurred either inside of (30 cases) or in close proximity (11 cases) to the WNV high risk areas predicted by the 2000-2004 model. The novel approach employed by this study may be implemented by other municipal, local, or state public health agencies to improve geographic risk estimates for vector-borne diseases based on a small number of acute human cases.