Diagnostic performance of the equine IgM capture ELISA for serodiagnosis of West Nile virus infection.

The objectives of these studies were to assess the diagnostic performance (sensitivity and specificity) of the IgM capture enzyme-linked immunosorbent assay (ELISA; MAC) for diagnosis of West Nile (WN) virus in horses and to examine the performance of this test by using different criteria for seropositivity. A total of 36 horses classified as WN virus infected (group 1) and 383 horses from 4 subpopulations of hoses classified as noninfected (groups 2, 3, 4, and 5) were used in the study. The sensitivity (proportion of infected horses that tested positive for WN virus IgM antibodies) and specificity (proportion of noninfected horses that tested negative) were calculated at different cutoff points by using receiver operating curve (ROC) analysis. Using a selected cutoff point = 2.0, the sensitivity and specificity of the MAC were 91.7 and 99.2%, respectively. The area under the ROC curve = 0.95 (95% confidence interval [CI], 0.89 to 1.0), suggesting that the MAC is a useful tool for diagnosis of recent WN virus exposure in horses. In fulfillment of the 2nd objective, 2 other indices were developed and these indices approached 1.0 for the AUC with smaller 95% CIs. These indices were then used to test 602 additional diagnostic samples submitted from suspect horses between 2002 and 2004. Using the standard cutoff, 194 (32%) of the horses were interpreted as positive. Utilizing newly predicted cutoff criteria from each index, additional horses were identified as positive. In conclusion, the MAC as used for identification of WN virus-diseased horses undergoing recent exposure performs reliably at the standard cutoff for seropositivity. A negative test might not completely rule out WN virus disease, but horses that test negative were most likely not exposed to WNV. Performance of the test can be further improved by investigation of other indexes of seropositivity.

Surveillance results from the first West Nile virus transmission season in Florida, 2001.

After West Nile virus (WNV) was first detected in Florida in July 2001, intensive surveillance efforts over the following five months uncovered virus activity in 65 of the state's 67 counties with 1,106 wild birds, 492 horses, 194 sentinel chickens, and 12 people found infected with the virus. Thirteen of 28 mosquito isolations came from Culex mosquitoes. As seen in the northeastern United States, wild bird mortality was the most sensitive surveillance method. However, unlike the predominantly urban 1999 and 2000 epizootics, the Florida transmission foci were rural with most activity detected in the northern part of the state. All human cases were preceded by the detection of WNV in animals; however, only eight of the twelve cases were preceded by reports of WNV activity in the county of residence. West Nile virus-positive animals detected by multiple surveillance systems preceded seven of these cases by two weeks or more.