Abstract
OBJECTIVE
West Nile virus (WNV) became notifiable in horses in 2003 in Canada and has been reported every year since. The objective of this study was to describe the spatiotemporal distribution of WNV in horses between 2003 and 2020 in Canada.
ANIMALS
The 848 symptomatic and laboratory-confirmed WNV cases in horses reported to the Canadian Food Inspection Agency between 2003 and 2020.
METHODS
Canada was divided into eastern and western regions for analysis. For each case, location and date of notification were captured. Triennial maps were made to describe the spatiotemporal distribution and expansion of reported cases. The association between year and latitude of cases was investigated with simple linear regressions, and space-time clusters were detected with a permutation scan test.
RESULTS
Most of the western region showed an extended distribution of WNV cases from 2003 to 2005 and a high recurrence of cases at the census division level. In the eastern region, the expansion of cases was gradual, with new infected census divisions mostly contiguous to previous ones. There was no association between year and latitude of cases. Six spatiotemporal clusters were detected.
CLINICAL RELEVANCE
This study confirmed the endemicity of WNV in parts of both regions with local peaks in risk varying in time. Prevention and control efforts should focus on previously infected areas based on the spatiotemporal regional distribution patterns. Incursions of WNV to new areas should also be anticipated. These findings could also contribute to enhancing monitoring and prevention of WNV infections in an integrated surveillance system.
Introduction
West Nile virus (WNV) is a mosquito-borne zoonotic pathogen and more than 20 years after its introduction in North America is still a major cause of disease in horses.1,2 It caused 842 confirmed clinical cases in horses between 2003 and 2019 in Canada.1 The number of WNV cases varied greatly from year to year during this period (from 1 to 304 cases/y), with yearly peaks showing differences in both timing and magnitude between eastern and western Canada.1 In Canada, WNV is statutorily regulated as an immediately notifiable disease, meaning diagnostic laboratories are required to report suspected or confirmed cases to the Canadian Food Inspection Agency (CFIA).3 These passive surveillance data are reported and displayed on the World Animal Health Information System,2 in weekly surveillance reports by the Public Health Agency of Canada,4 and on the Canadian Animal Health Surveillance System Equine Diseases Dashboard.5
Horses are a good surrogate for human risk of infection, and since they mostly live in areas with low human density, they allow a more complete representation of WNV distribution in an integrated surveillance system.6,7 However, usefulness of horse data may be limited; if WNV vaccine coverage in the horse population is not spatially homogenous, this might lead to a spatial risk distribution that is not representative of the risk of WNV exposure. Despite this drawback, information on the geographic distribution of WNV cases over time may still help horse owners and veterinarians to target more effectively preventive measures such as vaccination8 and mosquito control.9
Cases of WNV in horses are known to form clusters.10–12 A cluster is referred to as an area or period with an excess in risk for an event, unlikely to have happened by chance.13 Clusters are often used to drive public health interventions14 or identify areas with potential risk factors.15 Studies investigating clusters of WNV in horses only included a few years of surveillance data, and none have been conducted on a territory as large as Canada. In humans, a large-scale study15 found repeated yearly clusters of WNV in the Northern Great Plains of the US and another study14 from Ontario, Canada, found that some spatial clusters could be predicted by previous data, both suggesting that cluster analysis for WNV may help with targeting high-risk areas for future interventions.
Effects of climate change on the distribution and incidence of vector-borne diseases are a growing concern, especially for the northern latitudes of North America.16–18 Mosquito-borne diseases, such as WNV, may respond more quickly to climate change than other diseases because of the vectors’ short life cycle.17 It has also been suggested that warming temperatures and climate variability will affect vector distribution19 and therefore lead to reemergence and more frequent WNV outbreaks in the future.17 For example, the main vectors of WNV in Canada, Culex tarsalis16 and Culex pipiens,19 have been predicted to expand their distribution in future years. However, such predictions have yet to be documented under field conditions and, to date, no study has investigated the hypothesis regarding the potential northward expansion of WNV Canadian cases over the years. In Europe, the expansion of WNV is documented in countries where ecological conditions are suitable for WNV emergence and establishment, likely related to environmental changes affecting the life cycle of mosquitoes and their ability to replicate the virus.20
The goal of this study was to describe the spatiotemporal distribution of WNV reported cases in horses using 18 years of Canadian WNV surveillance data, whereby findings would aid in preventing future WNV infections. The specific objectives were the following:
- (1)To describe the geographic distribution and spatial patterns of WNV cases over time.
- (2)To investigate the association between year and latitude of WNV cases.
- (3)To investigate the presence of spatiotemporal clusters of WNV cases.
Methods
Study design, area, and period
A nationwide retrospective study was conducted on reported WNV cases in horses in Canada from 2003 to 2020 inclusively. Canada is made up of 10 provinces and 3 territories. Provinces were grouped into 2 regions for analysis: provinces of the Atlantic (New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador), Quebec, and Ontario formed the eastern region, and Manitoba, Saskatchewan, Alberta, and British Columbia formed the western region. Territories (Yukon, Northwest Territories, and Nunavut) were excluded, as there have been no locally acquired reported cases of WNV in mammals, birds, and mosquitoes, as previously described.1
Data extraction and case definition
Information on WNV cases in horses reported to the CFIA between 2003 and 2020 inclusively was extracted from the immediately notifiable disease database. Cases were defined as horses with compatible neurological signs and laboratory confirmation, as previously described.1,21 Data captured were the date of notification and location of the case. The owner/horse location and veterinary clinic location were recorded as city, postal code, and legal land description (when applicable) and were typically captured from the laboratory report used for notification to the CFIA. When both the horse and owner locations were available from this report or any follow-up communication, the information on the horse location was preferentially captured. Cases since 2015 were investigated to ascertain whether the horse had traveled 21 days prior to the positive laboratory result. In the event of a confirmed travel history, the previous location was used and considered as the likely site of infection. If the province was the only location information available, the cases were excluded from the study.
Geocoding
Cases were geocoded to the corresponding census division (CD) and the centroid of the ecumene of the corresponding census subdivision (CSD) on the basis of owner/horse location when available; otherwise, the veterinary clinic location was used as a surrogate. Census subdivision generally represents the boundaries of a municipality. For the purpose of this study, the ecumene was defined as the territory covered by either the national agricultural ecumene, which represents areas with significant agricultural activities, or the national population ecumene, which represents areas with a minimal population density.22 It was assumed that merging the 2 ecumenes would represent the most likely area where horses live, as they can be located on either farms or nonagricultural properties. The centroid of the ecumene of each CSD was computed and forced to be inside the ecumene (ArcGIS version 10.7.1; Esri).
Census divisions were used for descriptive choropleth maps since they have the most stable geographic boundaries over time after the provinces and represent a good compromise in the study area between visualization capacity and geographic resolution. Census subdivisions were used for cluster and regression analyses. The 2016 Census boundary files of Statistics Canada were used for geocoding and subsequent mapping.23
Descriptive mapping
To describe the distribution of WNV cases over time, a cumulative choropleth map was generated using 3-year time intervals for each region (ArcGIS version 10.7.1; Esri). This time interval was chosen arbitrarily to facilitate the illustration of 18 years of passive surveillance data without generating an excessive number of maps while ensuring that each map represented the same number of years. A CD was considered positive if at least 1 case occurred in this CD in a given year (positive CD). The number of years each CD was positive was summed over each 3-year time interval and cumulatively over all time intervals. Census divisions that reported cases during each 3-year time interval were also identified on the maps.
Statistical analysis
A map representing the distribution of WNV cases at the CSD level, along with a graph of the case distribution by year, was used to present data in each region. For each region, simple linear regressions between latitude (CSD centroid) of cases as outcome and year of case occurrence as explanatory variable were used to investigate directional trends in the distribution of WNV over time (SAS version 9.4; SAS Institute Inc). The regression coefficient estimates, r2 estimates, and P values were reported. Assumptions (linearity, homoscedasticity and normality of residuals, and absence of outliers) were visually assessed.
A spatiotemporal cluster analysis was conducted by region to detect high-risk clusters of WNV cases in horses using a space-time permutation model, as previously described,24,25 with 999 Monte Carlo replications (SaTScan 10.0.2; SaTScan). The year was used as the time unit to avoid detection of seasonal trends. The maximum scanning window (circular) was set at 50% of the population at risk (equivalent to 50% of the cases in a permutation model24) and a time period of a maximum 50% of the study period (9 years). Overlaps were only allowed if the centroid of the detected cluster was not located in another more likely cluster and did not contain the center of a more likely cluster. Coordinates of center in decimal degrees, radius in kilometers, period, observed and expected cases in the cluster, and P values were reported for each cluster. For the regression and cluster analyses, results were considered significant at P < .05.
Results
Between 2003 and 2020, a total of 848 WNV cases in horses were reported to the CFIA, of which 846 could be geocoded at the CD and CSD levels. For the 2 remaining cases, horses were diagnosed just a few days after traveling from another province and no information on the previous location was available.
Information on the owner/horse location was available for 753 of 846 (89%) cases, and 471 of 753 (63%) also had the veterinary clinic location disclosed. The veterinary clinic location was available for all cases without a owner/horse location (93/846 cases), and it was used as a surrogate location for these cases. The owner/horse and veterinary clinic locations were in the same CSD in 141 of 471 (30%) cases and in the same CD in 300 of 471 (64%) cases. These 93 cases occurred between 2003 and 2016. Applying the proportion above to validate the use of the veterinary clinic as a surrogate location, 60 of 93 (65%) of the cases without an owner/horse location were likely geocoded to the correct CD, leaving an estimated 33 cases (4% of total reported cases) not geocoded in their CD of occurrence.
Descriptive mapping
In the eastern region (Figure 1), the 6 triennial maps showed different distributions of CDs with cases and an increase in the number of positive CDs over the study period in Ontario and Quebec with new positive CDs detected on each map. No cases (hence, no positive CDs) were reported in the Atlantic provinces. The maps showed a higher occurrence of positive CDs in southern Ontario, including one that was positive for 7 years, the highest in the region. Over the study period in Ontario, there was a northwest trend of WNV-positive CDs along the border with the US. In Quebec, positive CDs were initially reported on the south shore of the St. Lawrence River and then reported on both shores after 2009. In Quebec, the maximum number of years for which the same CD was positive was 5 years. After 2005, 45 new positive CDs were observed and their number increased by 3, 10, 18, 11, and 3 chronologically between each period. New positive CDs were mostly (33/45) observed as contiguous (first order) to previously identified positive CDs, but a few isolated positive CDs were also observed farther north in Quebec. At the end of the study period, in Quebec and Ontario, 63 of 147 (43%) CDs were positive and, of those, 46% (29/63) had been positive for only 1 year.
Distribution of census divisions (CDs) according to the cumulative number of years with West Nile virus (WNV) cases in horses reported between 2003 and 2020 in the eastern region of Canada: Ontario (ON), Quebec (QC), and the Atlantic provinces (New Brunswick [NB], Nova Scotia [NS], Prince Edward Island [PEI], and Newfoundland and Labrador [NL]).
Citation: Journal of the American Veterinary Medical Association 261, 10; 10.2460/javma.23.05.0259
In the western region (Figure 2), spatiotemporal distribution of WNV-positive CDs in the Canadian Prairies (Alberta, Saskatchewan, and Manitoba) differed from the one in British Columbia. Positive CDs were reported across the Prairies with an increase in occurrence in southern areas. Also in the Prairies, there was a high recurrence, meaning the same CDs were repeatedly positive, except for the 2009 to 2011 period in which only 4 positive CDs were identified in Alberta and Saskatchewan and none in Manitoba. No new positive CDs were identified after 2005 in the Prairies. The maximum number of years for which the same CD was positive in Canada was 11 years, and it was located in southern Alberta. In Saskatchewan and Manitoba, the maximum number of years for which the same CD was positive was 10 and 5 years, respectively. In the Prairies, 48 of 60 (80%) CDs were positive during the study period. Of these, 4 of 48 (8%) were positive for only 1 year and all were located in Manitoba. All positive CDs in Alberta and Saskatchewan were positive for at least 3 years over the study period. The northernmost positive CD in Manitoba was only positive in 2003 (single case). In British Columbia, new positive CDs were gradually identified in the southeastern part of the province with little recurrence over the study period. Most new positive CDs in British Columbia were contiguous to previously identified positive CDs, and the maximum number of years for which the same CD was positive was 3 years.
Distribution of CDs according to the cumulative number of years with WNV cases in horses reported between 2003 and 2020 in the western region of Canada: British Columbia (BC), Alberta (AB), Saskatchewan (SK), and Manitoba (MB).
Citation: Journal of the American Veterinary Medical Association 261, 10; 10.2460/javma.23.05.0259
Regression analysis
The association between latitude and year of cases was not significant in either the eastern (1.3 km/y; P = .71) or western region (–1.1 km/y; P = .33). The r2 value was < 0.01 in both cases. No departure from model assumptions was noted.
Cluster analysis
The 846 geocoded cases were distributed across 441 different CSDs, and cases were reported every year in either or both regions (Figure 3). Six significant high-risk clusters were identified in 5 provinces (Figure 4; Table 1). Clusters all occurred in the second half of the study period and were mostly located along the US border. The cluster of 5 cases in Manitoba included 3 cases that were geocoded only at the veterinary clinic level and all to the same clinic.
Distribution of WNV cases in horses between 2003 and 2020 by census subdivisions (CSDs) and by year. Western and eastern regions are represented separately.
Citation: Journal of the American Veterinary Medical Association 261, 10; 10.2460/javma.23.05.0259
Space-time clusters of WNV cases in horses in Canada between 2003 and 2020 using a space-time permutation model. Western and eastern regions are represented separately.
Citation: Journal of the American Veterinary Medical Association 261, 10; 10.2460/javma.23.05.0259
Characteristics of reported high-risk space-time clusters of West Nile virus cases in horses in Canada between 2003 and 2020 using a space-time permutation model.
| Characteristics | Cluster ID | |||||
|---|---|---|---|---|---|---|
| BC | AB 1 | AB 2 | SK | MB | ON | |
| Coordinates of center (o)a | 49.095, –116.514 | 49.415, –112.869 | 51.797, –114.641 | 49.206, –108.386 | 49.526, –96.685 | 42.985, –81.049 |
| Radius (km) | 99 | 222 | 280 | 224 | 10 | 50 |
| Period | 2016–2017 | 2017–2018 | 2017–2018 | 2012–2013 | 2012 | 2018 |
| Observed cases | 14 | 69 | 66 | 21 | 5 | 7 |
| Expected cases | 1.5 | 32.3 | 37.4 | 7.1 | 0.3 | 1.1 |
| P value | < .001 | < .001 | .021 | .034 | .013 | .015 |
AB = Alberta. BC = British Columbia. MB = Manitoba. ON = Ontario. SK = Saskatchewan.
aCoordinates in NAD83 decimal degrees.
Discussion
This study demonstrated regional and provincial differences in the spatiotemporal patterns of WNV cases in horses over time, with an extended distribution of WNV cases across the Canadian Prairies following its introduction in Canada and a gradual expansion of cases in British Columbia, Ontario, and Quebec.
New positive CDs in British Columbia, Ontario, and Quebec were frequently contiguous to previously positive CDs, which could be explained by increased awareness, overwintering, and/or seasonal reintroduction of WNV. Increased awareness in areas where WNV was previously reported may have led to better monitoring for signs of disease and increased diagnostic efforts by owners and veterinarians. The contiguity observed in these provinces, especially in southeastern British Columbia, southern Ontario, and southern Quebec, could suggest the presence of endemic foci of WNV, which may persist locally through overwintering in mosquitoes and birds.26 This pattern is characterized by a recurrence in specific CDs and slow expansion to their surrounding CDs. However, seasonal reintroduction26 of the virus by infected mosquitoes and birds is also likely, given the lower recurrence in some positive CDs. For example, some positive CDs were noncontiguous and farther away, such as those in western Ontario and northern Quebec. This pattern supports incursions from long distances to new areas by either mosquitoes or birds,26 or the absence of horse population between these areas. Studies on seasonal reintroduction and overwintering of WNV, through genomic and sequencing analyses for example, have not yet been conducted in Canada.
The high recurrence of cases in the same CDs in the Prairies in the last half of the study, mainly in the CDs that had previously been positive in Alberta and Saskatchewan between 2003 and 2008, supports the possible overwintering and endemicity of WNV in the western region. The persistence of WNV could be driven by fine-scale factors favoring viral transmission and having a certain stability over time (eg, landscape and weather).27 For WNV to be detected in the same CDs over the years by passive surveillance, susceptible horses (eg, not previously infected and unvaccinated) are required. A study9 in Saskatchewan highlighted that regional seroprevalence of WNV in horses varied from 20% to 76%, supporting a high variability in the proportion of susceptible horses in endemic areas. In local horse populations, population dynamics (eg, migration and births) could result in partial renewal of the pool of susceptible animals every year. While previous studies in Canada have not examined the local recurrence of WNV cases in horses, a study14 of WNV cases in humans in Ontario found that a spatial cluster in 2012 could be predicted by data from a 2005 outbreak, supporting such local recurrence. On the other hand, 1 study28 in Quebec investigated the ability to predict new occurrence of cases in humans on the basis of cases reported in the previous year within small geographic units and found it to be low. The geographic and time unit of analysis may play a role in such discrepancies. In fact, the high recurrence of cases in this study could also be due in part to the aggregation of cases at an intermediate-sized geographic unit (CD) over 3-year intervals, masking some local and temporal variability.
The high recurrence of cases in infected areas raises concern on the use of preventive measures, such as horse vaccination, especially following the declaration of cases in an area. It is possible that the impact of previous cases on the implementation of preventive measures was very local or low or could not be detected at the geographic level used (CD). The presence of many sectors of activity in the horse industry (eg, breeding, sport, riding, and dressage)29 may also contribute to different practices regarding vaccination and prevention strategies for arboviruses.
Even though an expansion of the geographic distribution of C pipiens has been predicted in Ontario and Quebec, few cases (Figure 3) were outside of the current reported distribution of C pipiens.19 The descriptive maps (Figure 1) showed an increase over the course of the study in the number of positive CDs for Ontario and Quebec, but no apparent direction was observable, as supported by the absence of relationship between year and latitude in the linear regressions. In the Prairies, the persistent widespread distribution of WNV, especially along the US border, was also consistent with the absence of northern expansion noted in this region. While warmer temperatures are expected at higher latitudes in future years, it is worth considering that the northern regions of the Prairies may not provide a suitable habitat for C tarsalis. A previous study9 found that the seroprevalence of WNV in horses was higher in the southern ecoregions of Saskatchewan, likely attributable to climate and land coverage.
The 6 identified clusters of WNV corroborated previous research findings that WNV cases in horses tend to form clusters.10–12 Between 2016 and 2018, 3 space-time clusters were identified in British Columbia and Alberta, indicating a higher risk of WNV in these regions during this period than expected. These findings are consistent with a previously reported increase in incidence over the last decade in the western provinces, with a peak in 2018.1 Overall, these results suggest that WNV poses an ongoing threat to horses in Canada and high-risk areas are hard to predict. Since the model used in the analysis cannot identify purely spatial clusters, prevention should not only focus on the identified clusters, but on all positive CDs located in the ecumene of southeastern British Columbia and the Canadian Prairies and within the bounded ecumene shown on the eastern region map for Ontario and Quebec. It is noteworthy that incursions to new areas are also possible.
The absence of complete horse population was a limitation of this study, as some CDs in the Prairies could have a high recurrence of cases due in part to their very large horse populations or, inversely, the low recurrence of cases in other CDs could be driven by small horse populations.29 In Canada, only limited data are available for horse populations. Every 5 years, Statistics Canada conducts the Census of Agriculture, which only includes horses living on farms.30 A survey was also conducted by the horse industry in 2010 to estimate the total horse population in Canada,29 but it has not been updated since then. Given this limitation, as previously suggested,25 the space-time permutation model was used instead of a population-based model (Poisson) to investigate space-time clusters. However, the space-time permutation model assumes homogenous variations in the at-risk population through time and space and can therefore identify clusters in which there was an important shift over time in horse population.24,25 Such information on the variations of the horse population is only partially available through the Census of Agriculture. In addition, unvaccinated horses appear to be more at risk for WNV since they represent 96% of reported cases in Canada.1 High-risk clusters could also be due to a shift in vaccination, which would have influenced the at-risk population in an area for a period of time. Data on WNV vaccination in horses are only available at the provincial level,31 and without access to fine-scale vaccination and horse population data, the impact of vaccination on the cluster analysis remains unknown. Vaccination could also have impacted the presented distribution of cases. For example, repeated high recurrence of positive CDs in the Prairies could represent areas where vaccination is lower.
Finally, the absence of precise information on the location of some cases is noteworthy. For example, the small cluster in Manitoba could represent a locally enhanced awareness at the veterinary clinic level where WNV diagnoses were more sought than elsewhere, but an information bias due to the imprecision in the geocoding of cases cannot be ruled out. In addition to the estimated 4% of cases that are not geocoded at their correct horse/owner location, some horses can also live in a different CD than their owner or have undisclosed participation in shows or competitions outside their CD. Thus, the findings may not accurately reflect the areas at risk for WNV acquisition in horses.
This study highlights the spatial and temporal patterns of WNV cases in horses across Canada, suggesting that the geographic distribution of the disease over time was different between the eastern and western regions. Further investigations into the different vectors, local weather conditions, and land cover in each region could provide insights into the underlying ecological factors contributing to the observed difference. These findings also underscore the importance of region-specific approaches to prevention and control, such as widespread vaccination in western provinces and more targeted vaccination efforts to positive CDs in eastern provinces. This study also details the important contribution of horses in an integrated surveillance system to better inform on the distribution of the virus in Canada. To better understand the local recurrence of WNV, studies exploring the drivers and extent of vaccination and prevention measures and the impact of climate change on vector distribution and potential for overwintering are needed in Canada. This study further details the need for complete information on animal populations to better understand and effectively respond to infectious zoonotic diseases.
Acknowledgments
The authors acknowledge the support of the Fonds de Recherche du Québec–Nature et Technologies and of the National Sciences and Engineering Research Council of Canada with student grants attributed to Antoine Levasseur. The authors also acknowledge the contribution of the Canadian Food Inspection Agency staff to data management, geocoding, and reviewing the manuscript.
Disclosures
The authors have nothing to disclose. No AI-assisted technologies were used in the generation of this manuscript.
Funding
The authors have nothing to disclose.
References
- 1.↑
Levasseur A, Arsenault J, Paré J. Surveillance of West Nile virus in horses in Canada: a retrospective study of cases reported to the Canadian Food Inspection Agency from 2003 to 2019. Can Vet J. 2021;62(5):469-476.
- 2.↑
World Animal Health Information System. World Organisation for Animal Health. Accessed November 30, 2022. https://wahis.woah.org/
- 3.↑
Health of Animals Regulations. Part X General Provisions, Section 91.2 and Schedule VII Immediately Notifiable Diseases. Government of Canada, Department of Justice. Accessed June 10, 2020. https://laws-lois.justice.gc.ca/
- 4.↑
Mosquito-borne diseases national surveillance report. Public Health Agency of Canada. Updated February 25, 2022. Accessed October 18, 2022. https://www.canada.ca/en/public-health/services/diseases/west-nile-virus/west-nile-virus-other-mosquito-borne-disease.html
- 5.↑
Diseases E. Canadian Animal Health Surveillance System. Updated September 9, 2022. Accessed November 18, 2022. https://cahss.ca/cahss-networks/equine
- 6.↑
Epp T, Waldner C, Corrigan R, Curry P. Public health use of surveillance for West Nile virus in horses: Saskatchewan, 2003-2005. Transbound Emerg Dis. 2008;55(9-10):411-416. doi:10.1111/j.1865-1682.2008.01051.x
- 7.↑
Todoric D, Vrbova L, Mitri ME, et al. An overview of the National West Nile Virus Surveillance System in Canada: a One Health approach. Can Commun Dis Rep. 2022;48(5):181-187. doi:10.1056/NEJM200106143442401
- 8.↑
Epp T, Waldner C, Townsend HGG. A case-control study of factors associated with development of clinical disease due to West Nile virus, Saskatchewan 2003. Equine Vet J. 2007;39(6):498-503. doi:10.2746/042516407X248476
- 9.↑
Epp T, Waldner C, Leighton FA, Berke O, Townsend HG. Seroprevalence and risk factors for infection with West Nile virus in Saskatchewan horses, 2003. Can J Vet Res. 2007;71(4):256-263.
- 10.↑
Wittich CA, Ward MP, Fosgate GT, Srinivasan R. Identification of hyperendemic foci of horses with West Nile virus disease in Texas. Am J Vet Res. 2008;69(3):378-384. doi:10.2460/ajvr.69.3.378
- 11.
Corrigan RLA, Waldner C, Epp T, et al. Prediction of human cases of West Nile virus by equine cases, Saskatchewan, Canada, 2003. Prev Vet Med. 2006;76(3-4):263-272. doi:10.1016/j.prevetmed.2006.05.008
- 12.↑
García-Bocanegra I, Belkhiria J, Napp S, Cano-Terriza D, Jiménez-Ruiz S, Martínez-López B. Epidemiology and spatio-temporal analysis of West Nile virus in horses in Spain between 2010 and 2016. Transbound Emerg Dis. 2018;65(2):567-577. doi:10.1111/tbed.12742
- 13.↑
Ward MP. Spatio-temporal analysis of infectious disease outbreaks in veterinary medicine: clusters, hotspots and foci. Vet Ital. 2007;43(3):559-570.
- 14.↑
Thompson M, Berke O. Evaluation of the control of West Nile virus in Ontario: did risk patterns change from 2005 to 2012? Zoonoses Public Health. 2017;64(2):100-105. doi:10.1111/zph.12285
- 15.↑
Sugumaran R, Larson SR, Degroote JP. Spatio-temporal cluster analysis of county-based human West Nile virus incidence in the continental United States. Int J Health Geogr. 2009;8(1):43. doi:10.1186/1476-072X-8-43
- 16.↑
Chen CC, Jenkins E, Epp T, Waldner C, Curry PS, Soos C. Climate change and West Nile virus in a highly endemic region of North America. Int J Environ Res Public Health. 2013;10(7):3052-3071. doi:10.3390/ijerph10073052
- 17.↑
Ogden NH. Climate change and vector-borne diseases of public health significance. FEMS Microbiol Lett. 2017;364(19):fnx186. doi:10.1093/femsle/fnx186
- 18.↑
Wudel B, Shadabi E. A Short Review of Literature on the Effects of Climate Change on Mosquito-Borne Illnesses in Canada. National Collaborating Centre for Infectious Diseases; 2016.
- 19.↑
Hongoh V, Berrang-Ford L, Scott ME, Lindsay LR. Expanding geographical distribution of the mosquito, Culex pipiens, in Canada under climate change. Appl Geogr. 2012;33:53-62. doi:10.1016/j.apgeog.2011.05.015
- 20.↑
Bakonyi T, Haussig JM. West Nile virus keeps on moving up in Europe. Euro Surveill. 2020;25(46):2001938. doi:10.2807/1560-7917.ES.2020.25.46.2001938
- 21.↑
Paré J, Moore A. West Nile virus in horses - what do you need to know to diagnose the disease? Can Vet J. 2018;59(10):1119-1120.
- 22.↑
Ecumene. Dictionary, Census of Population, 2016. Statistics Canada. Accessed June 30, 2022. https://www12.statcan.gc.ca/census-recensement/2016/ref/dict/geo023-eng.cfm
- 23.↑
2016 Census - boundary files. Statistics Canada. Updated November 13, 2019. Accessed June 15, 2022. https://www12.statcan.gc.ca/census-recensement/2011/geo/bound-limit/bound-limit-2016-eng.cfm
- 25.↑
Kulldorff M, Heffernan R, Hartman J, Assunção R, Mostashari F. A space-time permutation scan statistic for disease outbreak detection. PLoS Med. 2005;2(3):e59. doi:10.1371/journal.pmed.0020059
- 26.↑
Reisen WK, Wheeler SS. Overwintering of West Nile Virus in the United States. J Med Entomol. 2019;56(6):1498-1507. doi:10.1093/jme/tjz070
- 27.↑
Karki S, Brown WM, Uelmen J, Ruiz MOH, Smith RL. The drivers of West Nile virus human illness in the Chicago, Illinois, USA area: fine scale dynamic effects of weather, mosquito infection, social, and biological conditions. PLoS One. 2020;15(5):e0227160. doi:10.1371/journal.pone.0227160
- 28.↑
Rocheleau J-P, Kotchi S-O, Arsenault J. Can local risk of West Nile virus infection be predicted from previous cases? A descriptive study in Quebec, 2011-2016. Can J Public Health. 2020;111(2):229-238. doi:10.17269/s41997-019-00279-0
- 29.↑
Evans V. 2010 Canadian equine industry profile. Equestrian Canada. Accessed June 20, 2020. https://www.equestrian.ca/industry/about
- 30.↑
Other livestock inventories on farms, Census of Agriculture, 2011 and 2016, inactive. Table 32-10-0427-01. Statistics Canada. Updated June 20, 2020. Accessed March 22, 2021. https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=3210042701&request_locale=en
- 31.↑
Levasseur A, Arsenault J, Paré J. Epidemiology of West Nile virus in horses in Canada from 2003 to 2020: a look at the geographic distribution and vaccination coverage. Poster presented at: 16th International Symposium of Veterinary Epidemiology and Economics; August 7-12, 2022; Halifax.
