Department of Health and Human Services
Centers for Disease = Control and=20 Prevention

Emerging Infectious Diseases

ISSN: 1080-6059

Email this = page=20

Printer-friendly=20 version

Download = this=20 article as PDF

Article Contents

Methods=20
Results=20
Discussion= =20
References= =20
Figure 1=20
Figure 2=20
Figure 3=20
Table 1=20
Table 2=20
Suggested=20 Citation

EID Journal=20 Home > Volume = 14,=20 Number 10=96October 2008

Volume 14, Number 10=96October 2008

Research

Ecological Factors Associated with West Nile Virus Transmission,=20 Northeastern United States

Heidi E. Brown,¹ = James E.=20 Childs, Maria A. Diuk-Wasser, and Durland Fish 3D"Comments
Author affiliation: Yale University, New = Haven,=20 Connecticut, USA

Suggested=20 citation for this article

Abstract
Since 1999, West Nile virus (WNV) = disease has=20 affected the northeastern United States. To describe the spatial = epidemiology=20 and identify risk factors for disease incidence, we analyzed 8 years = (1999=962006)=20 of county-based human WNV disease surveillance data. Among the 56.6 = million=20 residents in 8 northeastern states sharing primary enzootic vectors, we = found=20 977 cases. We controlled for population density and potential bias from=20 surveillance and spatial proximity. Analyses demonstrated significant = spatial=20 spreading from 1999 through 2004 (p<0.01, r² =3D 0.16). A=20 significant trend was apparent among increasingly urban counties; county = quartiles with the least (<38%) forest cover had 4.4-fold greater = odds (95%=20 confidence interval [CI] 1.4=9613.2, p =3D 0.01) of having above-median = disease=20 incidence (>0.75 cases/100,000 residents) than counties with the most = (>70%) forest cover. These results quantify urbanization as a risk = factor for=20 WNV disease incidence and are consistent with knowledge of vector = species in=20 this area.

West Nile virus (WNV) disease arrived in the United States in 1999 in = New=20 York City, yet how the disease became established and details concerning = the=20 nature of the transmission cycle in the United States remain unclear. = Experience=20 in the northeastern United States suggests an urban concentration of = human WNV=20 disease cases (1,2); = however,=20 environmental factors, such as urbanization, that underlie the patterns = of=20 transmission to humans have not been explicitly evaluated. We used human = surveillance data to describe and quantify the spread of WNV cases in = the=20 northeastern United States and empirically tested the hypothesis that = human WNV=20 disease is linked to the urban environment independent of human = population=20 density.

In the northeastern United States, a mainly urban cycle of WNV = transmission=20 is supported by the role of bird and mosquito species. This enzootic = cycle=20 occurs in urban bird species; human cases occur in late summer (2=967).= Culex=20 pipiens Linnaeus is the most commonly implicated mosquito vector in = the=20 maintenance of WNV in birds (1,2,8,9= ). In the=20 northeastern United States, this species feeds on birds found in urban = areas,=20 such as the American robin (Turdus migratorius), house sparrow=20 (Passer domesticus), and European starling (Sturnus = vulgaris)=20 (2,10). = The role of Cx. pipiens mosquitoes as primary WNV = vector is=20 supported by consistent isolations of WNV from mosquitoes captured in=20 surveillance traps (8,11=9614) and by=20 associations between virus-infected mosquitoes and dead-bird reports = (15).

A more contentious issue is the role of different mosquito species in = transmitting, or bridging, WNV between birds and other vertebrates, = including=20 humans. Cx. pipiens mosquitoes are known to breed in the = organically=20 rich water of artificial containers frequently found in urban areas = (16=9618). Habitat=20 modeling of potential WNV vectors in the northeastern United States = indicates an=20 urban focus for Cx. pipiens mosquitoes (19). = However,=20 its tendencies to mostly feed on birds make it an unlikely bridge = vector,=20 although other researchers have suggested that this species exhibits = late season=20 host switching to humans (5). = Aedes=20 vexans and Cx. salinarius mosquitoes have been = implicated=20 as bridge vectors in this region (1=963) = because of=20 their abundance and more nonspecific feeding patterns (20). = Although=20 both are present in urban areas, other land uses have been found to be = more=20 predictive of their distribution (19). = These=20 other studies do not indicate whether human incidence would be linked to = the=20 same ecological factors driving enzootic transmission.

In this study, we explicitly tested whether both enzootic and bridge=20 transmission occur in urban areas by evaluating human WNV disease and = degree of=20 urbanization within counties. We estimated the initial spatial spread in = time to=20 first case in Queens, New York, the site of first WNV detection = (21),=20 from 1999 through 2006. We also examined the trend for increasing = incidence with=20 decreasing forest cover while attempting to control for surveillance = efforts and=20 removing the effect of spatial proximity. The methods provide an example = of how=20 surveillance data with low spatial resolution can be used to quantify = risk.

Methods

The study was focused in 8 northeastern states (Connecticut, = Delaware,=20 Massachusetts, Maryland, New Jersey, New York, Pennsylvania, and Rhode = Island)=20 where the same mosquito species are likely to act as primary vectors. = States to=20 the north of the study area have had limited numbers of cases and may = involve=20 different mosquito species. States farther south and west are likely to = involve=20 different species of mosquitoes; hybridization between Cx.=20 pipiens and Cx. quinquefaciatus is more = common in=20 southern latitudes (16).
Human Incidence Data

We used annual numbers of human WNV cases reported to the Centers for = Disease=20 Control and Prevention (CDC) from 1999 through 2006. Human case data = were=20 acquired through multiple sources but met the CDC case definition, which = includes clinical disease with laboratory confirmation. Data for 1999 = were=20 extracted from the Morbidity and Mortality Weekly Report (22), = and data=20 for 2000 were downloaded from the National Atlas website (http://nationalatlas.gov/;=20 23).=20 Human case data for 2001 through 2006 were downloaded from the US = Geological=20 Survey maps page (http://nationalatlas.gov/printable/wnv.html; 24). = To protect=20 anonymity, human data from these sources are compiled at the county = level. All=20 other data were aggregated by county to match this resolution.

Geographic Data

County boundaries for the United States and 2000 census data were = downloaded=20 from the National Atlas website (http://nationalatlas.gov/boundaries and http://nationalatlas.gov/people), and county = centroids were=20 identified to facilitate the calculation of distances between counties. = Land-use=20 data were downloaded by state from the US Geological Survey National = Land Cover=20 Institute (http://landcover.usgs.gov/natllandcover.php; 24). = Percentage=20 of land cover class by county was extracted by using Fragstats Software = (25). = Land uses=20 classified as low-intensity residential, high-intensity residential,=20 commercial/industrial/transportation, and urban/recreational grasses = were=20 grouped into a class called urban. Land uses classified as = deciduous,=20 evergreen, and mixed forest were grouped into a class called = forest.=20 These 2 land use types were considered biologically relevant to the = study=20 question.

Statistical Analyses

To document evidence for the temporal and spatial spread of WNV = disease, we=20 generated cumulative incidence curves by state and by year and examined = the=20 distance between counties with cases. Time-to-first-case detection (in = years)=20 was used as the outcome predicted by distance to the origin, which was = Queens,=20 New York. For distance calculations, we ignored counties reporting no = WNV=20 disease cases because the first case is theoretically still to be = determined. To=20 visualize WNV disease spread, we plotted the mean incidence by year, = using the=20 spatial statistics tools of ArcGIS (26).
Distance measures were then used to adjust for the effect of spatial=20 proximity in the regression analyses (27).=20 Incorporating measures of spatial proximity in a regression model = removes the=20 effect of spatial structure that might otherwise result in = overestimation of the=20 strength of the association between the outcome, WNV incidence, and the=20 explanatory environmental variables (28,29)= .

Logistic regression modeling was initially used to identify the = relevant=20 predictors and to quantify their relative effects by calculation of odds = ratios=20 (ORs). Number of cases per county was standardized by using the 1990 US = Census=20 population density. Cumulative WNV disease incidence data from 1999 = through 2006=20 were dichotomized at their median to provide 2 categories of high and = low risk.=20 Predictor variables, percent urban, percent forested, county area, and = per=20 capita county income were stratified by quartiles. Logistic models were = tested=20 by using the Hosmer-Lemeshow goodness-of-fit test. The best model was = selected=20 based on the Akaike information criterion (AIC), which is a measure of = fit that=20 accounts for the number of parameters in the model. Models within 2 AIC = units=20 are considered comparable; models within 7 AIC units have less support = but are=20 still comparable; and models with differences >10 AIC units are not=20 comparable (30). = The=20 relationship between increasing cases and decreasing percentage of = forested land=20 was tested by using generalized least-square regression in STATA (31).
A risk model of total incidence was developed by using log (count +1) = transformed incidence as the response variable and the variables = identified as=20 important in the logistic regression analyses as predictors. To obtain a = better=20 fit, predictor variables were entered as continuous values for this = regression.=20 The κ statistic was used to assess agreement greater than chance = between the=20 median dichotomized original incidence and the predicted incidence, for = which=20 <0.21 is considered slight to poor and >0.61 is considered = substantial to=20 almost perfect (32).
All models were initially run using only the land-use predictors; and = the=20 Moran I test was used to assess whether closer observations = were more=20 similar than those farther apart. This finding of an association based = on=20 spatial location could indicate that proximity, rather than = environmental=20 factors, explains the distribution of disease incidence. Distance = variables=20 control for this potential spatial proximity effect and reflect the = presumed=20 biological relationships within the data.

The models were also adjusted for surveillance effort. Human disease=20 surveillance data must be interpreted with knowledge of the biases = inherent to=20 its collection (33). = County per=20 capita income was used as a measure of potential investment in = surveillance and=20 laboratory testing, as has been used in prior studies of surveillance = for animal=20 rabies (34).
Results

The Epidemic

Figure=20 1

Figure 1. Box plot of = total=20 incidence of West Nile virus disease in humans, by county, for the = 8=20 northeastern states in the study area...

Figure=20 2

Figure 2. A) Epidemic = curve of=20 mean incidence (log+1 transformed) of West Nile virus disease in = humans,=20 by state, 1999=962006. The 4 states depicted are representative of = the=20 variation among the 8 states in the study area.

Figure=20 3

Figure 3. Incidence of = human=20 West Nile virus disease cases in 8 northeastern states, = 1999=962006.=20 Deviation ellipses indicate 1 SD of the geographic mean yearly = incidence=20 calculated as the incidence weighted average in space for each=20 county...

From 1999 through 2006, the 204 counties in the 8 states reported 977 = WNV=20 disease cases (county mean 4.8, SD 8.7, median 1, range 0=9649) (Table 1). The median county incidence over = the 8-year=20 interval was 0.75 cases/100,000 residents (mean 1.77, SD 3.0, range=20 0=9620.2/100,000). The median incidence, excluding counties with no = reported=20 cases, was 1.70/100,000 residents (mean 2.94, SD 3.45, range = 0.22=9620.2/100,000)=20 (Figure = 1).=20 The highest incidence occurred in Forest County, (20.2/100,000), = followed by=20 Cameron County (16.8/100,000) and Adams County (15.3/100,000), all rural = counties in central Pennsylvania with very few cases (Forest County n = =3D 1,=20 Cameron County n =3D 1, and Adams County n =3D 14 [13 in 2003, 1 in = 2004]), and=20 small populations, probably representing data outliers.

Associations Based on Spatial Proximity

A cursory examination of the epidemic curve of WNV cases reported = from each=20 state during the 8-year study indicated that peak incidence was broadly=20 overlapping in all states (Figure = 2, panel=20 A). However, cumulative distribution functions of total WNV cases (Figure = 2, panel=20 B) by year indicated that New York experienced its median case earlier = in the=20 regional epidemic than did other states (Massachusetts, New Jersey, and=20 Connecticut), which suggests a spatiotemporal spread of WNV. Because a = spatial=20 component to spread was evident, we evaluated distance between counties = to=20 assess the spatial relationship between counties and to control for the = effect=20 of spatial proximity. The spatial component alone explained 15% of the = variance=20 in time to first case when Queens, New York, was used as the origin (n = =3D 123=20 counties with cases reported, p =3D 0.001). After 2004, no new counties = reported=20 WNV cases, and the incidence centroids of cases in 2005 and 2006 were = close to=20 one another and had shifted back toward the origin, which suggests that = the=20 disease may have reached endemicity in the region (Figure = 3).

Environmental Risk Factors

Risk (high or low) for WNV cases was significantly associated (by = county=20 quartile) with measures of urbanization and with percentage of forested = or urban=20 land. Because these 2 measures were highly correlated, we used only a = single=20 measure in the final analysis (Table 2). Total county area and other = demographic=20 indices (age) were not significant predictors and are not shown.

A logistic regression of the median split for total incidence with=20 categorical predictor variables of percentage forested area and = county-based per=20 capita income showed that percentage of forested land = (χ² =3D 26.13, df=20 =3D 6, p<0.001) and percentage of urban land (χ² =3D = 5.62, df =3D 6, p =3D=20 0.02) were both significant predictors of incidence (Table 2). Both models provided a good fit = (forested:=20 Pearson χ² =3D 7.82, df =3D 9, p =3D 0.55; urban: = Pearson χ² =3D=20 3.26, df =3D 8, p =3D 0.92). No effect of spatial proximity was found = among the=20 residuals for either model (forested: Moran I =3D =960.008, Z = =3D =960.49, p =3D=20 0.31; urban: Moran I =3D =960.002, Z =3D 0.40, p =3D 0.34).

To adjust for surveillance bias and the spatial relationship among = proximal=20 counties, we included the variables of county-based per capita income = and=20 distance from Queens, New York, respectively (Table 2). Both forested (χ² = =3D 36.67, df =3D=20 11, p<0.001) and urban (χ² =3D 33.55, df =3D 11, = p<0.001)=20 predictors were significantly associated with WNV incidence and provided = a good=20 fit (forested: Pearson χ² =3D 209.27, df =3D 192, p =3D = 0.19; urban:=20 Pearson χ² =3D 202.78, df =3D 192, p =3D 0.28). As = before, no effect of=20 spatial proximity was found in the residuals (forested: Moran I =3D = =960.007, Z =3D=20 =960.38, p =3D 0.35; urban: Moran I =3D 0.001, Z =3D 0.93, p = =3D 0.18). Although=20 all models were significant and fit the data, the latter model was = preferred on=20 the basis of AIC (not controlling for spatial proximity = AIC_forested =3D=20 270.7, AIC_urban =3D 281.2; controlling for spatial proximity=20 AIC_forested =3D 264.1, AIC_urban =3D 267.3) and = included=20 biologically relevant controls for the effect that spatial proximity = might have=20 in estimating the association between the outcome, disease incidence, = and=20 environmental variables of interest. A general, dose-dependent trend = indicated=20 increasing incidence as measures of urbanization increased (higher = incidence=20 with decreasing percentage of pixels classified as forest in each = county: χ=20 ² =3D 9.47, df =3D 1, p<0.01; goodness of fit χ = ² =3D 3.50,=20 df =3D 2, p =3D 0.17; higher incidence with increasing percentage urban = land: χ=20 ² =3D 7.13, df =3D 1, p<0.01; goodness of fit χ = ² =3D 1.98,=20 df =3D 2, p =3D 0.37).

The logistic regression model of dichotomized total incidence for the = 8 years=20 of the study, controlling for income (categorical variable by quartile) = and for=20 the effect of spatial proximity (distance variables), also showed a = distinct=20 trend of increasing incidence with percentage of forest cover; counties = with=20 <38% forest cover were 4.4=D7 more likely (95% confidence interval = 1.4=9613.2, p=20 =3D 0.01) to have high WNV incidence than were counties with >70% = forest cover=20 (Table 2).

Predictive Model

We used the predictors identified in the logistic regression analysis = to=20 develop a linear regression model to predict total incidence (log count = + 1=20 transformed for a normal distribution), using the quartile percent = forested land=20 by county. Per capita income (as a continuous variable) was used to = control for=20 surveillance effort. This model explains 9.7% of the variance in the = total=20 incidence (log count + 1) (p<0.001); however, the residuals indicated = an=20 effect due to spatial proximity (Moran I =3D 0.0349, Z =3D = 5.925,=20 p<0.001). Controlling for this spatial effect and surveillance effort = resulted in a better model (r² =3D 0.20, p<0.001; = Moran=20 I =3D =960.003, Z =3D 0.26, p =3D 0.40). The κ statistic = indicated good=20 agreement (κ =3D 0.343, SE =3D 0.066, Z =3D 5.22, p<0.001, = agreement =3D 67.16%)=20 between the predicted and the observed outcomes when the binomial = categorization=20 of incidence was used and resulted in 51 county incidence entries being=20 correctly identified as being below the median and 86 being correctly = identified=20 as being above the median. Errors were primarily in the direction of = predicting=20 the incidence above the median. When surveillance and spatial proximity = were=20 controlled for, the risk for WNV disease increased by 0.25% for every 1% = decrease in forest cover. For more direct comparison with the logistic=20 regression outcome, moving from the highest category of forest cover=20 (>69.59%) to the lowest (<38.29%), resulted in a 6.16% increased = risk for=20 WNV disease.

Discussion

This study documents the concentration of WNV cases within urban = areas of the=20 northeastern United States and provides a quantitative estimate of the = effect of=20 varying degrees of urbanization on the risk for WNV infection at the = county=20 level. Land-use data were used to ascribe degree of urbanization as a = predictor=20 for WNV disease risk; incidence models were generated, controlling for = human=20 population density, environment-based spatial associations in the = predictors,=20 and potential biases in WNV incidence reporting resulting from the = unequal=20 resource bases among counties.

Beginning in 1999, human WNV cases were reported in counties distant = from=20 Queens, New York, the presumed origin of infection. Although the = epidemic=20 initially appeared to spread in a west/southwesterly direction in the = 8-state=20 region examined, by 2005 the initial epidemic appeared to wane, and = reports of=20 disease among newly affected counties dropped to zero. The resulting = incidence=20 maps suggest a WNV disease=96endemic situation in the northeastern = United States.=20 The initial spread was not continuous along neighboring counties; = rather,=20 greater incidence was seen in urban counties after controlling for human = population density, surveillance bias, and the effect of spatial = proximity. The=20 best model indicates 4=D7 the risk for disease in the counties that fall = in the=20 lowest incidence quartile of forested land compared with the highest. = The=20 predictive nature of the data is also explored with the caveat that = additional=20 predictor variables are needed; nonetheless, it indicates increasing = risk for=20 WNV disease with decreasing forested lands.

The association between urban land use and human cases indicates that = urban/suburban land use enhances environmental conditions for both = enzootic and=20 bridge transmission, at least at the county level. The spatial = resolution of=20 human surveillance data did not allow for finer evaluation of = within-urban=20 associations. Brownstein et al. linked human WNV cases to greenness = indices in=20 urban areas and found an optimal vegetation index associated with higher = human=20 cases (35). = Brown et=20 al. found an environmental separation of bridge and enzootic vectors; = bridge=20 vectors occurred in areas with vegetation that might be associated with=20 residential areas within a city (36). = Finer=20 spatial resolution human data would allow for within-county analyses = that might=20 provide better estimations of where the cases (urban, periurban) are = occurring.=20 This would improve the predictive power of land use in the models, and = the=20 better association between land use and cases might help further = elucidate which=20 mosquito species are involved as bridge vectors.

Because of the type and resolution of the data, a sample predictive = model,=20 and not a predictive map, is provided. Nonetheless, the data and = analysis=20 provided are insightful as potentially predictive models. Additional = data, such=20 as bird abundance and perhaps also mammal abundance, are needed (37). = Because of=20 the often strict host and habitat preferences of mosquito species, = mosquito=20 surveillance data could also improve the predictive power and validity = of the=20 model. Our best predictive model explains only 20% of the variance; = additional=20 variables such as these might improve the model because the abundance of = hosts=20 and mosquito species will have a considerable effect on WNV = transmission.

Despite the reluctance to use human surveillance data for models of = disease=20 transmission (33), = such data=20 can provide information about spatial associations in vector-borne = disease as=20 shown here and by others (34,38,39). This=20 type of human surveillance modeling provides some useful insight into = the=20 distribution of human WNV cases and supports the current understanding = of the=20 transmission cycle.

To predict WNV disease requires understanding of the factors driving = both=20 enzootic transmission and bridging to humans. Different data = availability and=20 scales are involved in studying these 2 processes. We took advantage of = the=20 national coverage of the human incidence dataset to examine the = spatiotemporal=20 spread of WNV in this region and to generate a risk model based on land = use,=20 adjusted for the effect from spatial proximity. We show that human = surveillance=20 data at the county level are consistent with the urban nature of this = disease=20 system, as has been found in studies of enzootic transmission, = indicating that=20 the 2 processes occur in or near urban areas.

This work was supported by the US Department of Agriculture, = Agricultural=20 Research Service (1USDA-58-0790-5-068). H.E.B. was supported through = the CDC=20 Fellowship Training Program in Vector-Borne Disease at Yale = University.

Dr Brown graduated from the Vector Ecology Laboratory at Yale = University=20 and is a researcher with the Spatial Ecology and Epidemiology Group,=20 University of Oxford, Oxford, UK. In her research she seeks a better=20 understanding of vector-borne disease transmission through the = investigation=20 of the relationship between vectors and the environment in which they=20 persist.

References

Petersen LR, Roehrig JT. West Nile virus: a reemerging global pathogen. = Emerg Infect=20 Dis. 2001;7:611=964.=20
Molaei G, Andreadis TG, Armstrong PM, Anderson JF, Vossbrinck CR. = Host feeding patterns of Culex mosquitoes and = West Nile=20 virus transmission, northeastern United States. Emerg Infect Dis.=20 2006;12:468=9674.=20
Marra PP, Griffing S, Caffrey C, Kilpatrick AM, McLean R, Brand C, = et al.=20 West Nile virus and wildlife. Bioscience. 2004;54:393=96402. DOI=20
Roberts RS, Foppa IM. Prediction of equine risk of West Nile Virus = infection based on dead bird surveillance. Vector Borne Zoonotic Dis.=20 2006;6:1=966. PubMed DOI=20
Kilpatrick AM, Kramer LD, Jones MJ, Marra PP, Daszak P. West Nile = virus=20 epidemics in North America are driven by shifts in mosquito feeding = behavior.=20 PLoS Biol. 2006;4:606=9610. PubMed DOI=20
Gibbs SE, Allison AB, Yabsley MJ, Mead DG, Wilcox BR, Stallknecht = DE. West=20 Nile virus antibodies in avian species of Georgia, USA: 2000=962004. = Vector=20 Borne Zoonotic Dis. 2006;6:57=9672. PubMed DOI=20
Komar N, Panella NA, Burns JE, Dusza SW, Mascarenhas TM, Talbot = TO. Serologic evidence for West Nile virus infection in = birds in the=20 New York City vicinity during an outbreak in 1999. Emerg Infect = Dis.=20 2001;7:621=965.=20
Andreadis TG, Anderson JF, Vossbrinck CR. Mosquito surveillance for West Nile Virus in = Connecticut, 2000:=20 isolation from Culex pipiens, Cx. restuans, Cx. salinarius, = Culiseta=20 melanura. Emerg Infect Dis. 2001;7:670=964.=20
Turell MJ, O'Guinn M, Oliver J. Potential for New York mosquitoes to transmit West = Nile=20 Virus. Am J Trop Med Hyg. 2000;62:413=964.=20
Turell MJ, Sardelis MR, Dohm DJ, O'Guinn ML. Potential North American vectors of West Nile = virus. Ann N Y=20 Acad Sci. 2001;951:317=9624. PubMed=20
Anderson JF, Andreadis TG, Vossbrinck CR, = Tirrell S,=20 Wakem EM, French RA, et al. Isolation of West Nile virus from = mosquitoes,=20 crows, and a Cooper's hawk in Connecticut. Science. = 1999;286:2331=963. PubMed DOI=20
Andreadis TG, Anderson JF, Vossbrinck CR, Main AJ. Epidemiology of = West=20 Nile virus in Connecticut: a five-year analysis of mosquito data = 1999=962003.=20 Vector Borne Zoonotic Dis. 2004;4:360=9678. PubMed DOI=20
Anderson JF, Andreadis TG, Main AJ, Kline DL. Prevalence of West Nile virus in tree = canopy=96inhabiting=20 Culex pipiens and associated mosquitoes. Am J Trop Med = Hyg.=20 2004;71:112=969.=20
Nasci RS, Savage HM, White DJ, Miller JR, Cropp BC, Godsey MS, et = al. West Nile virus in overwintering Culex = mosquitoes, New=20 York City, 2000. Emerg Infect Dis. 2001;7:742=964.=20
Nasci RS, Komar N, Marfin AA, Ludwig GV, Kramer LD, Daniels TJ, et = al. Detection of West Nile virus=96infected mosquitoes and = seropositive juvenile birds in the vicinity of virus-positive dead = birds.=20 Am J Trop Med Hyg. 2002;67:492=966.=20
Carpenter SJ, La Casse WJ. Mosquitoes of North America (north of = Mexico).=20 Berkeley (CA): University of California Press; 1955.=20
Means RG. Mosquitoes of New York. Albany (NY): University of the = State of=20 New York, State Education Department, State Science Service, New York = State=20 Museum; 1979.=20
Andreadis TG, Thomas MC, Shepard JJ. Identification guide to the=20 mosquitoes of Connecticut. New Haven (CT): The Connecticut = Agricultural=20 Experiment Station; 2005.=20
Diuk-Wasser MA, Brown HE, Andreadis TG, Fish D. Modeling the = spatial=20 distribution of mosquito vectors for West Nile virus in Connecticut, = USA.=20 Vector Borne Zoonotic Dis. 2006;6:283=9695. PubMed DOI=20
Apperson CS, Harrison BA, Unnasch TR, Hassan HK, Irby WS, Savage = HM, et=20 al. Host-feeding habits of Culex and other = mosquitoes=20 (Diptera : Culicidae) in the Borough of Queens in New York City, with=20 characters and techniques for identification of Culex = mosquitoes.=20 J Med Entomol. 2002;39:777=9685.=20
Hayes EB, Komar N, Nasci RS, Montgomery = SP, O'Leary=20 DR, Campbell GL. Epidemiology and transmission dynamics of West Nile = Virus=20 disease. Emerg Infect Dis. 2005;11:1167=9673.=20
Centers for Disease Control and Prevention. Update: West = Nile=96like viral=20 encephalitis=97New York, 1999 [reprinted from MMWR; 1999:48:890=962]. = JAMA.=20 1999;282:1714. PubMed DOI=20
US Geological Survey. National atlas of the United States. 2006 = May 22=20 [cited 2007 Jan 1]. Available from http://nationalatlas.gov/=20
US Geological Survey. West Nile virus: human [cited 2008 Aug 7]. = Available=20 from http://diseasemaps.usgs.gov/wnv_us_human.html=20
McGarigal K. Cushman SA, Neel MC, Ene E. FRAGSTATS. In: Spatial = pattern=20 analysis program for categorical maps. Amherst (MA): University of=20 Massachusetts; 2002.=20
Environmental Systems Research Institute. I., ArcGIS. Redlands = (CA):=20 Environmental Systems Research Institute, Inc.; 2004.=20
Waller LA, Gotway CA. Linking spatial exposure data and health = events. In:=20 Applied spatial statistics for public health data. New York: John = Wiley and=20 Sons; 2004.=20
Legendre P. Spatial autocorrelation=97trouble or new paradigm. = Ecology.=20 1993;74:1659=9673. DOI=20
Lichstein JW, Simons TR, Shriner SA, Franzreb KE. Spatial = autocorrelation=20 and autoregressive models in ecology. Ecol Monogr. 2002;72:445=9663.=20
Burnham KP, Anderson DR. Multimodel inference=97understanding AIC = and BIC in=20 model selection. Sociol Methods Res. 2004;33:261=96304. DOI=20
StataCorp LP. Intercooled Stata 9.0. = College Station=20 (TX): StataCorp LP; 2005.=20
Landis JR, Koch GG. Measurement of observer agreement for = categorical=20 data. Biometrics. 1977;33:159=9674. PubMed DOI=20
Brookmeyer R, Stroup DF. Monitoring the health of populations: = statistical=20 principles and methods for public health surveillance. Oxford (NY): = Oxford=20 University Press; 2004.=20
Childs JE, Krebs JW, Real LA, Gordon ER. Animal-based national=20 surveillance for zoonotic disease: quality, limitations, and = implications of a=20 model system for monitoring rabies. Prev Vet Med. 2007;78:246=9661. PubMed DOI=20
Brownstein JS, Rosen H, Purdy D, Miller JR, Merlino M, Mostashari = F, et=20 al. Spatial analysis of West Nile virus: rapid risk assessment of an=20 introduced vector-borne zoonosis. Vector Borne Zoonotic Dis. = 2002;2:157=9664. PubMed DOI=20
Brown H, Duik-Wasser M, Andreadis T, Fish D. Remotely sensed = vegetation=20 indices identify mosquito clusters of West Nile virus vectors in an = urban=20 landscape in the northeastern US. Vector Borne Zoonotic Dis. = 2008;8:197=96206. PubMed DOI=20
Kramer LD, Bernard KA. West Nile virus infection in birds and = mammals. Ann=20 N Y Acad Sci. 2001;951:84=9693. PubMed=20
Lucey BT, Russell CA, Smith D, Wilson ML, Long A, Waller LA, et = al.=20 Spatiotemporal analysis of epizootic raccoon rabies propagation in=20 Connecticut, 1991=961995. Vector Borne Zoonotic Dis. 2002;2:77=9686. PubMed DOI=20
Waller LA, Goodwin BJ, Wilson ML, Ostfeld RS, Marshall SL, Hayes = EB.=20 Spatio-temporal patterns in county-level incidence and reporting of = Lyme=20 disease in the northeastern United States, 1990=962000. Environ Ecol = Stat.=20 2007;14:83=96100. DOI

Figures

Figure 1. Box plot of total = incidence of=20 West Nile virus disease in humans, by county, for the 8 northeastern = states in=20 the study area...
Figure 2. A) Epidemic curve of = mean=20 incidence (log+1 transformed) of West Nile virus disease in humans, by = state,=20 1999=962006. The 4 states depicted are representative of the variation = among the 8=20 states in the study area...
Figure 3. Incidence of human = West Nile=20 virus disease cases in 8 northeastern states, 1999=962006. Deviation = ellipses=20 indicate 1 SD of the geographic mean yearly incidence calculated as the=20 incidence weighted average in space for each county...

Tables

Table 1. Incidence (per 100,000 = persons)=20 of West Nile virus disease in humans, northeastern United States,=20 1999=962006
Table 2. Odds ratios for median = split=20 incidence of West Nile virus diseases in humans, for significant=20 variables

Suggested Citation for this Article

Brown HE, Childs JE, Diuk-Wasser MA, Fish D. Ecological factors = associated=20 with West Nile virus transmission, northeastern USA. Emerg Infect Dis = [serial on=20 the Internet]. 2008 Oct [date cited]. Available from=20 http://www.cdc.gov/EID/content/14/10/1539.htm

DOI: 10.3201/eid1410.071396

¹Current affiliation: University = of Oxford,=20 Oxford, UK.

Comments to the Authors

Please use the form below to submit correspondence to the authors or = contact=20 them at the following address:

Durland Fish, Yale University, LEPH 600, 60 College St, New Haven, CT = 06520,=20 USA; email: durland.fish@yale.edu

=20
Return email address:

Please note: To prevent email errors, please use no = web=20 addresses, email addresses, HTML code, or the characters <, >, and = @ in=20 the body of your message.

= =

Comments to the EID Editors

Please contact the EID Editors at eideditor@cdc.gov

The opinions expressed by authors contributing to this = journal=20 do not necessarily reflect the opinions of the U.S. Department of Health = and=20 Human Services, the Public Health Service, the Centers for Disease = Control and=20 Prevention, or the authors' affiliated institutions. Use of trade names = is for=20 identification only and does not imply endorsement by any of the groups = named=20 above.

This page posted September 25, 2008

Journal Contents

Home=20
Expedited=20
Current=20 Issue=20
Ahead=20 of Print=20
Past=20 Issues=20
Announcements =

General Information

About Us=20
Instructions to = Authors=20
Submit Manuscript=20
Subscribe=20
Contact=20 Us

Home=20
Policies and=20 Regulations=20
Disclaimer=20
e-Government=20
FOIA=20
Contact Us =

USAGov -- The = U.S.=20 Government's Official Web Portal

Department of Health and Human=20 Services

Safer Healthier People
Centers for Disease Control and Prevention, 1600 Clifton Rd, Atlanta, = GA=20 30333, U.S.A
Tel: (404) 639-3311 / Public Inquiries: (404) 639-3534 / = (800)=20 311-3435