Exploratory Longitudinal, Spatial and Spatiotemporal Evaluation of the Raccoon Variant Rabies Virus in West Virginia, 2000- 2015

Background and Objectives: Rabies is an almost invariably fatal viral disease of mammals.In West Virginia, there are two variants of the rabies virus, bat and raccoon.Raccoon rabies (RRV) was introduced subsequent to a release by hunters, and has since spread throughout the Northeastern United States and into Canada.There has been a notable lack of westward movement, however.The objectives of these studies are to 1) examine the behavior of RRV temporally during the study period 2) determine whether there are spatial determinants that may be associated with RRV incidence and 3) examine whether the change of spatial determinants with time over the study period can be correlated with changes in RRV incidence.Secondary objectives included evaluation of the oral rabies vaccine program, determination of whether clustering of disease occurred in the state, and examination of whether there were changes in RRV behavior in varying animal hosts.Methods: County-level surveillance data were obtained from the West Virginia state health department and combined with geographic data from a number of publicly available databases.In the first study, RRV case numbers were examined to determine any trends over the study period, 2000 -2015, and trends were compared for counties where oral rabies vaccination had occurred and those without vaccination program implementation, using negative binomial regression techniques and z-score comparisons.The second study evaluated several geospatial characteristics of the counties for association with disease incidence and clustering of disease, using Poisson spatial conditional autocorrelation regression within a Bayesian environment and local indicators of spatial autocorrelation.In the third study, integrated nested Laplace approximation (INLA) and negative binomial regression were utilized to allow simultaneous adjustment for the random effects of both time and space.In each study, analyses were performed for the data as a whole, in addition to analyses of various host subpopulations.Results: Study 1. Analyses found statistically significant (p < 0.05) reductions in case numbers over the study period and showed that incidence rates were declining at significantly higher rates in counties where oral rabies vaccination had been deployed.Incidence showed significant declines for all animals in the study and the raccoon only subpopulation, but similar declines could not be shown in non-raccoon wildlife or domestic animals.Study 2. Statistically significant clustering of RRV cases was demonstrated for all host types examined, with clustering tending to occur in the eastern portion of the state and overlapping for the various host types.Several geospatial variables were shown to be significantly associated (credible intervals did not include 0) with RRV incidence, with several of the variables recurring among the host types.Study 3. Regression analyses that compensated for spatial and temporal autocorrelation showed that many of the variables found to be associated in study 2 did not retain those associations in univariate analyses, while multivariate analyses indicated that none of the variables that had changed over the study period were associated with RRV incidence.Conclusions: Across the three studies, differences were noted between the host types with regard to RRV behavior.It appears that reductions in RRV incidence for raccoons do not extend to the other host types.This is of particular concern in the case of domestic animals, as they represent the most likely source of human exposure.The oral rabies vaccine project does appear to be having an impact on RRV incidence.Spatial clustering of disease is occurring in all host types.Although there was some variation in significant determinants between the different hosts, as well as between the analytic approaches, geographic location and proximity to the Virginia border were most consistently present.In the multivariate INLA analyses, those variables that cannot be changed over time were the only ones of significance.The INLA approach was a substantial improvement over the more traditional methods and tended to corroborate the previous findings.The findings of these studies highlight the importance of using statistical methodologies that can accommodate the spatial and temporal structures embedded within public health surveillance data.Epidemiology for the opportunity and support necessary for this project.Specifically, I would also like to thank Drs. Sarah Knox, Sijin Wen, Michael McCawley, Thomas Hulsey, Brad Hillgartner and Sheldon Owen for the time and effort they put forth to see the project to completion.Additionally, I would like to extend my appreciation to the West Virginia Department of Health and Human Resources, Bureau of Public Health for providing the data crucial to the

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.