Geographical Clusters and Predictors of Rabies in Three Southeastern States

Abstract

The rabies virus causes progressive encephalomyelitis that is fatal in nearly 100% of untreated cases. In the United States, wildlife act as the primary reservoir for rabies; prevention, surveillance, and control costs remain high. The purpose of this study is to understand the current distribution of wildlife rabies in three southeastern states, with particular focus on raccoons as the primary eastern reservoir, as well as identify demographic and geographic factors which may affect the risk of human exposure. This ecologic study obtained county-level rabies surveillance data from state health departments and the United States Department of Agriculture Wildlife services for North Carolina, Virginia, and West Virginia from 2010 to 2013. A spatial statistical analysis was performed to identify county clusters with high or low rates of raccoon rabies in the three states. Potential demographic and geographic factors associated with these varying rates of rabies were assessed using a multivariable negative binomial regression model. In North Carolina, raccoons constituted 50% of positive tests, in Virginia, 49%, and in West Virginia, 50%. Compared to persons residing in West Virginia counties, persons in North Carolina counties had 1.67 times the risk of exposure (p < 0.0001) to a rabid raccoon and those in Virginia counties had 1.82 times the risk of exposure (p < 0.0001) to a rabid raccoon. Compared to those counties where farmland makes up less than 17% of the total area, persons residing in counties with 17–28% farmland had a 32% increased risk of exposure to a rabid raccoon. In counties with 28–39% farmland, there was an 84% increased risk of exposure. State, rurality, and percent of area designated as farmland were the best predictors of risk of raccoon rabies exposure. Further research is needed to better understand the effect of the oral rabies vaccine program in controlling the risk of human exposure to raccoon rabies.

Introduction

Rabies as a disease is an acute viral infection, causing progressive encephalomyelitis that is fatal in nearly 100% of untreated cases (Heymann 2008). All mammals are susceptible to rabies, and the virus is typically transmitted in the saliva of rabid mammals through a bite, although there exists rare documentation of transmission through contamination of mucous membranes, aerosol transmission, and corneal and organ transplantation (CDC 2016, Rupprecht 2010). All Lyssaviruses are antigenically related, but the use of monoclonal antibodies and nucleotide sequencing demonstrates that there are different variants of the virus dependent on species or geographical region (Heymann 2008). Worldwide, the canine rabies variant is the most prevalent and of greatest concern, causing ∼90% of human cases and 99% of human deaths (Knobel et al. 2005, World Health Organization 2016). In the United States and Canada, where canine rabies variant has been eradicated, wildlife act as the primary reservoirs for rabies (Kuzmina et al. 2013). In the United States the four primary reservoirs are raccoons, bats, skunks, and foxes (Monroe et al. 2016).

Although the number of human deaths due to rabies in the United States has fallen to less than four per year, rabies remains a significant public health concern because of its high case fatality rate and continued presence in the wildlife population (CDC 2016). Rabies is 100% preventable through prompt medical care. This results in a large public health impact related to resources required for managing exposures. The CDC estimates that 40,000 postexposure prophylaxes are given each year in the United States at an average cost of $1000 per course (CDC 2016). The best form of human prevention is to vaccinate pets and avoid contact with wildlife, but as the population expands to overlap with wildlife habitats, the latter has become much more difficult.

The high costs associated with surveillance, diagnostic testing, and postexposure treatment of humans potentially exposed to rabies have resulted in coordinated efforts to control the expansion of rabies, particularly in raccoon populations (Russell et al. 2005). In 1990 the United States Department of Agriculture (USDA) began using oral rabies vaccines (ORV) to reduce the prevalence of rabies in specific wildlife species in targeted states (USDA 2016). Beginning in 2005, the ORV program established a barrier along the Appalachian Mountains to prevent the westward expansion of raccoon variant rabies. Surveillance for rabies, performed by multiple agencies, is an important program component for exposures, risk tracking, and evaluation of ORV program success. Surveillance is performed conditionally at state health departments (SHDs) when animals are submitted for testing due to a potential human exposure and randomly by the USDA Wildlife Services to track prevalence around the ORV zones.

There were two primary objectives of this study. First, the USDA Wildlife Services enhanced surveillance data and SHD data (North Carolina, Virginia, and West Virginia) were used to focus on the current distribution of raccoon rabies in Central Appalachia and identify clusters of unusually high or low rates of rabies at the county level. Then demographic and geographic characteristics of these counties which may act as indicators of increased exposure risk to rabies were investigated. It was hypothesized that those areas with greater farmland coverage and greater population size would have a higher proportion of positive raccoons.

Materials and Methods

An ecologic study of wildlife rabies in the counties and independent cities of North Carolina, Virginia, and West Virginia from 2010 to 2013 was undertaken. The primary data involving rabies rates were collected from two sources as follows: SHD and USDA Wildlife Services. SHD data were collected from North Carolina, Virginia, and West Virginia. The CDC has listed rabies as a notifiable disease in both humans and animals, although surveillance is passive. As a result, any potentially rabid animal that has been in contact with humans or domestic animals and thus poses a threat for human rabies exposure can be submitted to the state for rabies testing. Depending on laboratory protocol, states may also test animals with abnormal behavior or assist with environmental spot checks at the discretion of the state public health veterinarian. Virginia's laboratory protocol indicates that testing is permitted for significant human or domestic animal exposure from an animal exhibiting behavior of a rabid animal or at the recommendation of a public health official using appropriate guidance (DCLS 2014). North Carolina's protocol states that only animals that have potentially exposed a person, household pet, or livestock to rabies should be submitted (NCPH 2016). West Virginia's protocol allows for testing of animals involved in potential exposure to humans, domestic animals, or livestock, as well as those exhibiting rabid behavior with no exposure and for the purposes of environmental spot checking (WVDHHR 2014). States archive these tests each year to monitor rates of rabies within their borders. Each state provided data on rabies testing from 2010 to 2013, which included county of submission, total number of animals submitted, total number of raccoons submitted, total animals testing positive, and total raccoons testing positive. The USDA Wildlife Services branch provided similar rabies surveillance data for North Carolina, Virginia, and West Virginia.

As part of the ORV program, the USDA regularly traps and tests animals around the designated ORV zone to determine the efficacy of the vaccine and the success of the program. The Rabies Management Program of the USDA provided data on rabies testing from 2010 to 2012, which included the county where the animal was trapped, the total number of animals tested, the total number of raccoons tested, the total animals testing positive, and the total raccoons testing positive. County-level demographic variables (number of counties and independent cities, population, average median household income, rural urban continuum code [RUCC], and educational attainment) were obtained from the 2010 US Census (Census 2010). County-level total land area and farmland area were collected from the USDA 2012 Agriculture Census (USDA 2012). County-level forest land area was collected from the USFS 2012 Census (USFS 2012). This research was ruled exempt by the University of Kentucky Institutional Review Board, as well as by the institutional review boards of the three participating states, as it did not meet the federal definition of human subjects (45 CFR 46.102(f)).

Spatial analysis

County-level rates of positive tests for rabies in raccoons were mapped using ArcGIS v10.1 (ESRI 2011) where the denominator was total number of raccoons submitted for testing from 2010 to 2013. County and state borders were downloaded from the US Census, TIGER/Line website (United States Census Bureau 2013). SHD rates were designated with graduated colors, while USDA rates were designated with graduated symbols. Areas of low or high rates of raccoon and overall rabies at the county level were identified using Kulldorff's spatial scan statistic implemented in SaTScan (Kulldorff 2011). (SaTScan™ is a trademark of Martin Kulldorff. The SaTScan software was developed under the joint auspices of Martin Kulldorff, the National Cancer Institute and Farzad Mostashari at the New York City Department of Health and Mental Hygiene.) Because our data were collapsed across 4 years, a purely spatial cluster analysis using the discrete Poisson probability model was used to scan for nonoverlapping counties with either significantly higher or lower rates of raccoon rabies compared to the rest of the study area.

SaTScan calculates expected rates of positive tests based on the rate across the entire study area and then calculates rate ratios for positive results in each significant cluster by dividing the rate of observed among expected for each cluster by the same rate in the remaining portion of the study area. For this analysis, a maximum spatial cluster size of 25% of the total population at risk was used. As will be shown, the spatial distribution of high rates of positive tests does not necessarily require statistical verification, but this analysis was nonetheless conducted to quantify and visualize differences between the highest and lowest risk regions within the study area.

Regression analysis

Given the high percentage of overall rabies submissions that were raccoons and the historical focus on raccoon rabies in the eastern United States, the primary outcome of interest in this ecologic analysis was rates of raccoon rabies by county. Distributions of the percent farmland and percent forest land were examined to determine cut points for categorical interpretation and to explore dose response. Educational attainment for each county was described by the category of educational level achieved by the greatest percentage of people greater than 25 years of age. The original Census data contained seven categories which were collapsed into three categories in this study for increased sample size per category and to simplify interpretation. These categories were high school diploma or less, some college or an associate's degree, and bachelor's degree or greater. Similarly, the original RUCC defined by the USDA Economic Research Service contains nine codes depending on metropolitan or nonmetropolitan status and population size. The nine categories were collapsed into three ignoring population size. These categories were metropolitan area, nonmetropolitan area adjacent to a metropolitan area, labeled “nonmetro,” and nonmetropolitan not adjacent to a metropolitan area, labeled “rural.”

Regression analyses were performed using PROC GENMOD in SAS v9.3 (SAS Institute 2011). Due to the high number of counties with zero submissions or zero positive tests, the negative binomial regression provided a better fit for the data than a Poisson regression. The first model included all seven demographic and geographic variables, and in each subsequent model the variable with the highest nonsignificant p-value (where alpha = 0.05) was removed. This was continued until the final model contained only significant variables. One variable, median household income, was removed from the final model in place of RUCC despite it being statistically significant when RUCC was not; this decision was made because income produced a negligible log effect on the rabies rate. We assessed goodness of fit using Deviance, AIC, and BIC. All counties with submissions to the SHD were included in both the cluster analysis and the regression analysis.

Results

Counts of the total numbers of submitted animals and those testing positive for rabies from the USDA and the SHD data are given in Table 1. In North Carolina, 12,516 animals were submitted to the SHD for testing, of which 1504 were raccoons. Approximately 41.2% of raccoons submitted tested positive for rabies, while only 5.6% of all other species submitted tested positive. In Virginia, 15,556 animals were submitted to the SHD, of which 2481 were raccoons with 44.3% of raccoons submitted testing positive for rabies, while only 8.7% of all other species submitted tested positive. In West Virginia, 2642 animals were submitted to the SHD, of which 565 were raccoons. Of the raccoons, 28.5% tested positive for rabies and 7.6% of all other species submitted tested positive. In the USDA animal samples from North Carolina, rabies was present in 9.2% of raccoons and 4.8% of other species; from Virginia, rabies was present in 7.1% of raccoons and 6.5% all other species; and from West Virginia, rabies was present in 2.1% of raccoons and 2.5% of all other species.

Table 1.

Animals Submitted to the State Health Departments and to the United States Department of Agriculture for Rabies Testing in North Carolina, Virginia, and West Virginia (2010 to 2013)

	North Carolina			Virginia			West Virginia
	Total animals	Rabies positive	Prevalence (95% CI)	Total animals	Rabies positive	Prevalence (95% CI)	Total animals	Rabies positive	Prevalence (95% CI)
Raccoon
SHD^a	1504	619	41.2 (38.7–43.7)	2481	1099	44.3 (42.4–46.3)	565	161	28.5 (24.9–32.3)
USDA^b	141	13	9.2 (5.2–14.9)	636	45	7.1 (5.3–9.3)	2604	54	2.1 (1.6–2.7)
Other species
SHD	11,012	622	5.6 (5.2–6.1)	13,075	1140	8.7 (8.2–9.2)	2.077	159	7.7 (6.6–8.9)
USDA	42	2	4.8 (0.8–14.9)	586	38	6.5 (4.7–8.7)	442	11	2.5 (1.3–4.3)
Total species
SHD^a	12,516	1241	9.9 (9.4–10.5)	15,556	2239	14.4 (13.9–15.0)	2642	320	12.1 (10.9–13.4)
USDA^b	183	15	8.2 (4.8–12.9)	1222	83	6.8 (5.5–8.3)	3046	65	2.1 (1.7–2.7)

SHD data submissions by state: North Carolina counties (total N = 100) with raccoon submissions (n = 90) with any animal submissions (n = 100); Virginia counties (N = 134) with raccoon submissions (n = 117) with any animal submissions (n = 118); and West Virginia counties (N = 55) with raccoon submissions (n = 52) with any animal submissions (n = 55).

USDA submissions by state: North Carolina counties (total N = 100) with submissions (n = 11); Virginia counties (N = 134) with submissions (n = 28); and West Virginia counties (N = 55) with submissions (n = 45).

CI, confidence interval; SHD, state health department; USDA, United States Department of Agriculture.

The geographic distribution of rabies rates over the 3-year time period in raccoons by county, according to both the USDA and each of the SHDs, is provided in Figure 1. Higher rates of raccoon rabies according to the USDA are indicated by larger circles, while higher rates according to the SHD are indicated by darker shades of blue. Both symbologies show a clear change in rabies rates that coincides with the location of the ORV zone. Counties in the zone or west of it generally have very low rates of rabies, while those to the east of the ORV zone have much higher rates.

FIG. 1.

Racoon rabies rates in Central Applachia (2010–2013). SHD, state health department; USDA, United States Department of Agriculture.

The spatial cluster analysis of SHD raccoon rabies rates is given in Figure 2. Two significant high rate and two significant low rate clusters were identified. The first high rate cluster consisted of 23 counties and independent cities in far eastern West Virginia and northern Virginia. There were 493 confirmed positive raccoons, while only 356 were expected. Residents of counties within this cluster were at 1.52 times the risk of being exposed to a raccoon positive for rabies relative to residents of counties outside the cluster. A second high rate cluster was found among 67 counties and independent cities throughout south-central Virginia and central North Carolina. This cluster had 577 confirmed positive raccoons, with 458 expected positives, a rate 37% higher than outside the cluster (rate ratio [RR] = 1.37, p < 0.0001). Meanwhile, two low-risk clusters were identified along the ORV zone. The first cluster, containing 25 counties and independent cities along the West Virginia-Virginia border, had 0 confirmed positives with 68 expected. The second cluster, containing 22 counties and independent cities in southwest Virginia and western North Carolina, had 52 confirmed positives with 146 expected and a rate 66% lower than outside the cluster (RR = 0.34, p < 0.0001).

FIG. 2.

Significant spatial clusters of raccoon rabies rates (2010–2013).

The negative binomial regression analysis to examine the predictors of raccoon rabies rates is given in Table 2. Compared to those living in West Virginia counties, all other factors being held constant, residents of North Carolina counties were 1.67 times more likely to encounter a rabid raccoon (p < 0.0001), while residents of Virginia counties and independent cities were 1.82 times more likely to encounter a rabid raccoon (p < 0.0001). Residents of counties with 17–28% farmland were 1.32 times more likely to encounter a rabid raccoon (p = 0.013), while residents with 28–39% farmland were 1.84 times more likely (<0.0001) and residents with 39–100% farmland were 1.64 times more likely (p ≤ 0.0001) to encounter a rabid raccoon, in comparison to those living in counties where farmland makes up less than 17% of total area. Compared to rural counties, all other factors held constant, residents of nonmetropolitan counties were 1.56 times more likely to have been exposed to a rabid raccoon (p = 0.005), while those in metropolitan areas were 1.41 times more likely (p = 0.024).

Table 2.

Negative Binomial Model for Positive Raccoon Rabies and Other Species Rabies-Relative Risk and 95% Confidence Intervals, Ecologic Analysis by County

	Raccoon rabies			Other species
Variable	RR	95% CI	p	RR	95% CI	p
State
North Carolina	1.67	1.31–2.12	<0.0001	1.13	0.82–1.55	0.4516
Virginia	1.82	1.43–2.31	<0.0001	2.05	1.51–2.79	<0.0001
West Virginia	1.00	—	—	—	—	—
Percentage farmland
39–100	1.64	1.34–2.02	<0.0001	1.58	1.18–2.12	0.0021
28–38.9	1.84	1.49–2.27	<0.0001	1.97	1.47–2.64	<0.0001
17–27.9	1.32	1.06–1.65	0.0131	1.55	1.14–2.09	0.0045
<17	1.00	—	—	—	—	—
Rural urban continuum code
Metro	1.41	1.05–1.89	0.0240	0.66	0.48–0.92	0.0137
Nonmetro	1.56	1.15–2.13	0.0045	0.99	0.69–1.36	0.8600
Rural	1.00	—	—	—	—	—
Goodness of fit	Deviance = 1.18			Deviance = 1.16
	AIC = 1080.0 BIC = 1111.3			AIC = 1246.5 BIC = 1278.2

RR, rate ratio.

Discussion

The purpose of this study was to identify the current distribution of wildlife rabies in central Appalachia, as well as identify potential demographic factors which might be associated with variation in this distribution. As expected from the initial spatial overview of the rates of raccoon rabies in central Appalachian counties, the spatial scan statistic identified several county clusters with high or low rates of rabies compared to the area as a whole. Counties in clusters with low rates were generally located near the USDA's ORV zone or west of the zone, while those clusters with high rates were generally located east of the zone. It was also evident (Fig. 1) that there were several counties where USDA random surveillance suggested low rates of rabies, while the SHDs reported upwards of 50% positive test results. The negative binomial regression analysis found that the variation in the distribution of raccoon rabies rates by county could be best explained by the percent of area designated as farmland, the state in which the county was located, and the rurality of the county as classified by the USDA Economic Research Service's Rural-Urban Continuum Codes.

This is the first known ecological analysis of wildlife rabies rates in the United States which investigates potential factors explaining the variation in endemic rabies rates at the county level. The study contributes to understanding of the current distribution of raccoon rabies in a highly endemic region and highlights potential factors associated with varying rates of rabies by county. This is especially important in a time when wildlife rabies, particularly in raccoons, has reached historically high prevalence rates (Rupprecht and Smith 1994, Rupprecht et al. 1995). The results of this study suggest that there are strong variations in the rates of raccoon rabies in central Appalachian counties, which might be explained by certain human demographic characteristics. Higher rates of rabies in more metropolitan counties could be explained by the greater number of people available to contact a wild animal and submit it for testing, which has been supported in similar studies focused on raccoon rabies epizootics (Anthony 1990, Jones et al. 2003). However, the regression analysis for county population alone resulted in a relationship that was not statistically significant (p = 0.78). The finding that some measure of population density and percent farmland were both related to risks of raccoon rabies rates is supported by the natural tendency of raccoons to live in close proximity to humans (Riley et al. 1998), as well as by findings of another study focusing on potential rabies epizootics in the tidewater region (Jones et al. 2003).

There was a statistically significant increased risk of rabies exposure by state. Both North Carolina and Virginia counties had higher relative risks of rabies exposure than West Virginia counties. Given the scattered distribution of raccoon rabies rates across this region, as well as the widely varied demographics of each county, it was expected that county-level demographics would have an impact on the rates of raccoon rabies; however, only the rurality of each county and the proportion of farmland in the county showed significant associations. The USDA ORV zone runs along the spine of the Appalachian Mountains, almost cutting West Virginia in half, as well as covering a large portion of its surface area. The placement of this zone was intentional, utilizing the natural geographic barrier of the Appalachian Mountains to isolate raccoon variant rabies to the east coast, as studies in New York and Europe have shown that increased elevation is directly associated with lower rates of endemic rabies in terrestrial reservoir species (Pastoret et al. 2004, Recuenco 2007, Recuenco 2008). Elevation was not factored into this study primarily because analysis was done at the county level, and elevation across counties may change dramatically, particularly in this region of the United States. The impact of elevation on rabies rates and raccoon densities shown in other studies, as well as the impact of state unexplained by other demographics in this study, suggests that the ORV zone was not only placed strategically to utilize the elevation change but also continues to be an effective barrier against the westward expansion of raccoon rabies over time.

In future studies, a more critical assessment of rabies rates with regard to proximity to the ORV zone should be undertaken. This would allow estimation of the true impact this ORV program has on the reduction of raccoon rabies and thus the decreased risk of exposure to humans. It would also be beneficial to conduct a comparative analysis of USDA surveillance rates and SHD surveillance rates where the USDA was able to trap over all counties, not just those in proximity to the ORV zone. A more comprehensive study such as this would allow us to compare models with raccoons trapped randomly to those voluntarily submitted to the SHD. Previous studies have demonstrated an association between natural geographic factors such as elevation and water bodies on raccoon rabies epizootics (Smith 2002, Recuenco 2007). Given that raccoon rabies is endemic to central Appalachia, more studies should utilize natural geographic barriers and examine the prevalence of raccoon rabies within and between them rather than utilizing politically defined barriers. Finally, it may be beneficial to expand this study to states on both sides of the ORV zone to not only explore how the ORV zone affects raccoon rabies but also to provide a better model for other reservoir species, such as the southeastern skunk.

Regarding study limitations, there were several counties with no submissions to the SHD, which does not indicate that rabies is not present or that humans are not at risk of being exposed. It could be that there were exposures but the animal was not available for testing, or the animal was available but was delivered in such a condition that it was inadequate for testing. This study also relies on the accuracy of the reporting individuals. When animals are submitted to the SHD for rabies testing, the CDC and most state protocols dictate that it should coincide with direct or indirect risk of human exposure. The exact type of exposure is not always clear in the rabies report when an animal is submitted, and these particular data do not include any information about the potentially exposed individual. Thus, there may be some tested animals included in our SHD data with no direct risk to human health. However, the close proximity of wildlife to humans due to overlapping habitat use provides a constant risk for spillover, and these test results are still valuable in terms of the potential risk to humans and their pets.

Conclusion

This study contributes to knowledge about the current distribution of wildlife rabies in central Appalachia,with particular focus on the relationship between rabies rates and human demographic factors. Expansion of this study across states on both sides of the ORV zone would provide greater insight into the effect that the ORV program has on reducing this risk of human exposure to rabies in the United States. Future studies may also consider using natural geographic borders rather than political borders to identify high risk clusters, as wildlife are more likely to adhere to natural barriers. From this study, it appears that the best predictors of raccoon rabies exposure risk at the county level are percent farmland, state, and rurality. The results of this and similar studies demonstrate a need for continued research concerning wildlife rabies in the United States. Through this understanding we will be able to focus our efforts on rabies prevention and education and, thus, reduce the overall cost of rabies prevention and control.

Footnotes

Acknowledgments

The authors acknowledge the following officials who provided SHD and USDA WS data for this research: Dr. Julia Murphy, DVM, MS, DACVPM; Mr. Waqas Humayon; Ms. Jordona Kirby; Dr. Barry J. Meade, MS, DVM, PhD; Ms. Peggy Brantley; Dr. Miguella Mark-Carew, AB, PhD; and Ms. Susan Stowers.

Author Disclosure Statement

No competing financial interests exist.

References

Anthony

, Childs

, Glass

, Korch

, et al. Land use associations and changes in population indices of urban raccoons during a rabies epizootic. J Wildl Dis, 1990; 26:170–179.

Census of the United States.. American Fact Finder. 2010. Available at http://factfinder.census.gov/faces/nav/jsf/pages/index.xhtml

Centers for Disease Control and Prevention.. Rabies. 2016. Available at http://www.cdc.gov/rabies/

DCLS (The Virginia Department of General Services, Division of Consolidated Laboratory Services). Rabies Specimen Submission Protocol. 2014. Available at http://vdhweb/epi/Rabies/D2_DCLS_Lab_Protocol_and_Coop_Program_Aug_2014.pdf?CondID=123

ESRI.. ArcInfo GIS version 10. Redlands, CA: California Environmental Systems Research Institute Inc., 2011.

Heymann

. Control of Communicable Disease Manual. Washington, DC: American Public Health Association, 2008.

Jones

, Curns

, Krebs

, Childs

. Environmental and human demographic features associated with epizootic raccoon rabies in Maryland, Pennsylvania, and Virginia. J Wildl Dis, 2003; 39:869–874.

Knobel

, Cleaveland

, Coleman

, Fevre

, et al. Re-evaluating the burden of rabies in Africa and Asia. Bull World Health Organ, 2005; 83:360–368.

Kulldorff

. SatScan v. 9.1.1 Software for the spatial and space-time scan statistics. Silver Springs, MD: Information Management Services Inc, 2011.

10.

Kuzmina

, Lemey , Kuzmin

, Mayes

, et al. The phylogeography and spatiotemporal spread of south-central skunk rabies virus. PLoS One, 2013; 8:e82348.

11.

Monroe

, Yager

, Blanton

, Birhane

, et al. Rabies surveillance in the United States during 2014. J Am Vet Med Assoc, 2016; 248:777–788.

12.

NCPH (North Carolina Public Health) State Laboratory of Public Health.. Virology/Serology: Rabies Virus. 2016. Available at http://slph.ncpublichealth.com/virology-serology/rabies.asp

13.

Pastoret

, Kappeler

, Aubert

. European rabies control and its history. In: King

, Fooks

, Aubert

, Wandeler

, eds. Historical Perspectives of Rabies in Europe and the Mediterranean Basin. OIE, Paris, pp. 337–350.

14.

Recuenco

, Eidson

, Cherry

, Kulldorff

, et al. Factors associated with endemic raccoon (Procyon lotor) rabies in terrestrial mammals in New York State, USA. Prev Vet Med, 2008; 86:30–42.

15.

Recuenco

, Eidson

, Kulldorff

, Johnson

, et al. Spatial and temporal patterns of enzootic raccoon rabies adjusted for multiple covariates. Int J Health Geogr, 2007; 6:1.

16.

Riley

, Hadidian

, Manski

. Population desnity, survival, and rabies in racoons in an urban national park. Can J Zool, 1998; 76:1153–1164.

17.

Rupprecht , Charles

, et al. Use of a reduced (4-dose) vaccine schedule for postexposure prophylaxis to prevent human rabies: recommendations of the advisory committee on immunization practices. MMWR Recomm Rep, 2010; 59:1–9.

18.

Rupprecht

, Smith

. Racoon rabies: The re-emergence of an epizootic in a densely populated area. Semin Virol, 1994; 5:155–164.

19.

Rupprecht

, Smith

, Fekadu

, Childs

. The ascension of wildlife rabies: A cause for public health concern or intervention?. Emerg Infect Dis, 1995; 1:107–114.

20.

Russell

, Smith

, Childs

, Real

. Predictive spatial dynamics and strategic planning for raccoon rabies emergence in Ohio. PLoS Biol, 2005; 3:e88.

21.

SAS Institute. SAS 9.3. Cary, NC: SAS Institute Inc. 2011.

22.

Smith

, Lucey

, Waller

, Childs

, et al. Predicting the spatial dynamics of rabies epidemics on heterogeneous landscapes. Proc Natl Acad Sci U S A, 2002; 99:3668–3672.

23.

United States Census Bureau.. Tiger/Line^® Shapefiles. 2013. Available at https://www.census.gov/geo/maps-data/data/tiger-line.html

24.

USDA (United States Department of Agriculture).. APHIS. National Rabies Management Program. 2016. Available at https://www.aphis.usda.gov/aphis/ourfocus/wildlifedamage/programs/nrmp/!ut/p/z1/04_iUlDg4tKPAFJABpSA0fpReYllmemJJZn5eYk5-hH6kVFm8X6Gzu4GFiaGPu6uLoYGjh6Wnt4e5mYGZo5G-l76UfgVFGQHKgIABPI_xQ!!/

25.

USDA.. USDA Census of Agriculture. State and county level data. 2012. Available at https://www.agcensus.usda.gov/Publications/2012/

26.

USFS.. “USFS Forest Inventory and Analysis National Program.” FIA Datamart County Level Forest Coverage. 2012. Available at http://apps.fs.fed.us/fiadb-downloads/datamart.html

27.

World Health Organization.. Rabies. 2016. Available at www.who.int/rabies/en

28.

WVDHHR (West Virginia Department of Health and Human Resources) Office of Laboratory Services.. Rabies. 2014. Available at www.wvdhhr.org/labservices/labs/rabies/index.cfm