Abstract
Vaccination is the most efficient tool for preventing West Nile virus (WNV) disease in horses. The objectives of this retrospective study were to: i) estimate the provincial and national horse population in Canada; ii) estimate the annual vaccination coverage for WNV in each province; and iii) compare estimates of provincial annual vaccination coverage. Horse population data (2006–2023) and the number of vaccine doses sold in Canada (2016–2020) were obtained by province. Based on these data and expert consultations, the annual number of vaccine doses administered, and provincial horse populations were modeled at the provincial level using Program Evaluation and Review Technique (PERT) distributions, whereas the probability of primary vaccination in vaccinated horses was modeled with a uniform distribution. The annual number of horses vaccinated was estimated using a binomial distribution. Monte Carlo simulations were used to estimate provincial annual vaccination coverage with 95% credible intervals (CI). National horse population and vaccination coverage were estimated by summing the provincial estimates. The national horse population in Canada (2016–2020) was estimated at 795 420 horses, with a vaccination coverage of 16% (95% CI: 13 to 20%). The vaccination coverage estimates for WNV ranged from 10% (95% CI: 6 to 20%) in the Atlantic provinces to 35% (95% CI: 22 to 53%) in Manitoba. The coverage was significantly higher (P ≤ 0.01) in Manitoba than in the Atlantic provinces, Saskatchewan, Alberta, and British Columbia. The study revealed low annual WNV vaccination coverage across Canada, emphasizing the need to promote vaccination and collect more comprehensive data on horse population, as uncertainties in these data drive the uncertainty in vaccination coverage based on model sensitivity analysis.
Résumé
La vaccination est l’outil le plus efficace pour prévenir la maladie causée par le virus du Nil occidental (VNO) chez les chevaux. Une étude rétrospective a été menée au Canada pour i) estimer les populations provinciales et nationales de chevaux, ii) estimer la couverture vaccinale provinciale annuelle pour le VNO et iii) comparer les couvertures vaccinales annuelles entre les provinces. Les données sur la population équine au Canada (2006–2023) et le nombre de doses de vaccin vendues (2016–2020) ont été collectés par province. Avec ces données et l’avis d’experts, le nombre annuel de doses de vaccin administrées et les populations provinciales de chevaux ont été modélisées avec des distributions PERT, alors que la probabilité de primovaccination parmi les chevaux vaccinés a été modélisée avec une distribution uniforme. Le nombre de chevaux vaccinés annuellement a été estimé avec une distribution binomiale. Des simulations de Monte Carlo ont permis d’estimer la couverture vaccinale annuelle avec des intervalles de crédibilité (IC 95 %). La population nationale de chevaux (2016–2020) a été estimée à 795 420 chevaux avec une couverture vaccinale de 16 % (IC 95 % : 13 %–20 %). Une couverture vaccinale allant de 10 % (IC 95 % : 6 %–20 %) dans les provinces de l’Atlantique à 35 % (95 % IC : 22 %–53 %) au Manitoba a été estimée, avec une couverture significativement plus élevée (P ≤ 0,01) au Manitoba qu’en Colombie-Britannique, Alberta, Saskatchewan et dans les provinces d’Atlantique. Les résultats de l’étude indiquent une faible couverture vaccinale annuelle contre le VNO à travers le Canada, illustrant la nécessité de promouvoir la vaccination et d’obtenir des données plus complètes sur la population équine. En effet, l’incertitude des données de population est celle qui influence le plus l’incertitude des résultats de couverture vaccinale selon l’analyse de sensibilité du modèle.
(Traduit par les auteurs)
Introduction
Recent epidemiological studies on West Nile virus (WNV) disease in horses have highlighted the need to better understand WNV vaccination coverage in Canada (1,2). From 2003 to 2023, 898 cases of WNV were reported in Canada to the Canadian Food Inspection Agency (CFIA), with the number of cases fluctuating each year, but with at least 1 case reported annually (1,3). Cases, which consist of horses exhibiting neurological signs and with a positive laboratory test confirming active infection, must be reported to the CFIA. A retrospective study of these cases reported a mortality rate of 31.9% and that 96% of the cases occurred in unvaccinated horses (1).
Cases of WNV in horses in Canada have been reported to form clusters mostly on a regional scale, indicating that regional factors influencing the incidence of disease, such as meteorological conditions or landscape, could play a role (2). Clusters of cases and incidence of WNV in horses could also be largely impacted by variations in vaccination coverage (2). Gaining a comprehensive understanding of the extent of horses vaccinated in Canada may provide crucial insight for establishing tailored recommendations to prevent WNV disease in horses.
The American Association of Equine Practitioners (AAEP) recommends the WNV vaccine as a core vaccine for all horses in North America (4). Vaccines against WNV are highly effective in limiting the development of clinical disease (4,5), the intensity of clinical signs, and mortality (4,6).
In 2003, it was estimated that 75% of horses in Saskatchewan were vaccinated against WNV (7). Surveys conducted in the United States by the National Animal Health Monitoring System Unit suggest that approximately 75% of all equine operations surveyed had some kind of vaccination program in place for their resident horses (8). Among those operations that vaccinate, 56.8% vaccinated their horses against WNV in 2015 (8). In the survey, the risk of exposure to a pathogen was reported as the main factor in deciding whether to vaccinate a horse against a disease. Other factors influencing vaccination rates could be limited veterinary access, lack of knowledge and/or awareness about WNV disease (9), possible side effects of the vaccine, and the cost of vaccination (8).
Accurately estimating horse populations in Canada is challenging as there is no mandatory verified identification or national traceability program for horses. The incomplete and uncertain nature of horse population data in Canada has been mentioned in previous studies (1,2,10,11). The Statistics Canada Census of Agriculture and national surveys conducted by Equestrian Canada (EC) are the 2 primary sources of information about the total population of horses in Canada.
The Statistics Canada Census of Agriculture is conducted every 5 y and only accounts for horses living on census farms on a designated census day (12). It therefore provides only a minimum baseline number of horses in Canada. The exact proportion of horses living on properties covered by the census is unknown and the definition of a farm changed in the last census (2021), making it difficult to compare recent census data to that obtained from previous ones (12).
Equestrian Canada conducted national surveys in 1998, 2003, 2010, and 2021 (13,14). Estimates of horse population obtained from EC are based on national surveys completed voluntarily by people in the equine industry; responses are compiled to estimate the provincial proportion of horses missed by the Census of Agriculture. Equestrian Canada national surveys indicate that the national horse population increased from 1998 to 2010, followed by an almost 50% decline from 2010 to 2021 (13,14). Other sources of information from horse breed registries, provincial associations, and government authorities provide additional information on the horse population in Canada but are not always up to date.
The objectives of this retrospective study were to: i) estimate the provincial and national horse population in Canada; ii) estimate WNV annual vaccination coverage in each province based on currently available information on provincial horse populations and vaccine sales; and iii) compare estimates of provincial annual vaccination coverage.
Materials and methods
A Monte Carlo simulation retrospective study was conducted using data from various Canadian sources, including surveys, censuses, vaccine sale information, and opinions of experts. The target populations were defined as the average horse population for each province from 2016 to 2020. The horse population was modeled by province using Program Evaluation and Review Technique (PERT) distributions, which are defined by minimum, maximum, and most likely estimate (MLE) parameter values.
Data from the 2016 Canadian Census of Agriculture were used to establish the minimum parameter value for each PERT distribution. Only these data were used for this parameter because the new farm definition in the 2021 Census of Agriculture (12) excluded sectors previously reported (e.g., horse boarding, riding schools, and required declared revenues or expenses) and previous censuses were outside the study period.
Various sources (see Supplementary Material, Table SI) were consulted to estimate the maximum parameter value in each province. The highest number of horses reported in a timepoint across all provincial sources was used as the maximum parameter for the distributions. Data mined included any/all data available from various sources (e.g., equestrian associations, breed registries) for the period from 2006 (10 y before the study period) and 2023, which was considered reasonable given that horses live for approximately 25 y. As opposed to census data, surveys are generally not conducted periodically, often use different methodologies, and only provide estimates. Multiple years (2006 to 2023) were therefore included to maximize the chances of capturing the largest reasonable number of horses living in Canada during the study period.
For each province, a most likely horse population value was initially estimated for 2018 (midpoint of study) using a simple linear regression of the previous (in 2010) and subsequent (in 2021) EC survey population estimates (13,14). These 2018 provincial estimates were presented to the Canadian Animal Health Surveillance System (CAHSS) Equine Surveillance Network (a Canadian group of experts from equine veterinary practices, government, industry, academia, and regional surveillance partners) through a confidential and anonymous survey.
These experts were provided with the minimum and maximum horse population values for each province, described earlier, as well as the 2018 population estimates. They were then asked to estimate the most likely population size by indicating their best estimate on a scale of 7 equal intervals based on the provincial maximum and minimum values. If all experts agreed on the provincial estimate for 2018, which was the midpoint of the survey, this was used as the MLE value. Otherwise, the midpoint of the best interval estimate of each expert was averaged and used as the MLE.
The annual number of WNV vaccine doses sold by province from 2016 to 2020 was obtained, under agreement, from all pharmaceutical companies that marketed a WNV vaccine in Canada (Boehringer Ingelheim, Zoetis, and Merck) during that period. It was assumed that every dose sold was administered to a horse in the year it was sold, and these doses were therefore referred to as doses administered. The variability in the annual number of vaccine doses administered in a province was modeled as a stochastic variable using a PERT distribution with the minimum, the mean (representing the MLE), and the maximum number of annual doses administered in the province over the 5-year period.
Since the primary vaccination requires a booster 4 to 6 wk after the first vaccination (4), whereas the following annual booster only requires a single dose prior to the mosquito season every year, the number of horses vaccinated is lower than the number of vaccine doses administered. Therefore, a parameter representing the probability of primary vaccination [p(primary)] was included in the model and used to estimate the number of horses vaccinated (Figure 1).
Figure 1.
Schematic illustration of the Monte Carlo simulation to estimate vaccination coverage for West Nile virus in horses, with example of distributions for the horse population (Nb horses), the annual number of vaccine doses administered (Nb doses), and the probability of primary vaccination [p(primary)] and of a binomial distribution for the number of horses vaccinated.
A convenience sample of 5 equine practitioners from Alberta, Ontario, and Québec who were members of the CAHSS Equine Surveillance Network were consulted and asked to estimate the proportion of horses that would be vaccinated for the first time in a given year, therefore requiring 2 doses. One of the experts estimated that less than 5% of the horses were vaccinated for the first time, whereas the other 4 estimated approximately 10%. However, one of these mentioned that a significant proportion of the horses vaccinated for the first time were not receiving their booster dose.
Based on this information and the limited number of experts interviewed, the probability of primary vaccination was modeled as a uniform distribution with a minimum of 0 and a maximum of 0.1. A Monte Carlo simulation, consisting of 10 000 iterations of the following steps, was used to estimate the annual vaccination coverage for each province using R Statistical Software (Version 4.2.1; R Core Team 2022, Vienna, Austria).
As shown in Figure 1, for each iteration, a single value was randomly sampled from each of the 3 distributions: horse populations (Nb horses), annual number of vaccine doses administered (Nb doses), and for each province, probability of primary vaccination [p(primary)]. The latter 2 parameters were used to estimate the number of horses vaccinated, with a binomial distribution, where “Nb doses” was the sampled number of vaccine doses administered and “p” represented the probability that a given dose was either the first of a series of a primary vaccination or an annual booster. Therefore, horses receiving 2 doses of vaccine (primary vaccination) were only counted once.
Finally, the vaccination coverage was calculated by dividing the annual number of horses vaccinated by the horse population. The outputs from the Monte Carlo simulations were probability distributions of annual vaccination coverage values for each province, which were summarized by their median and 95% credible intervals (CI). The national horse population and overall number of horses vaccinated in Canada were estimated by the summation of all the provincial estimates for each iteration. For comparing vaccination coverage among provinces, the output for each pair of provinces was subtracted from one another at each iteration. If the 99% CI (to account for multiple comparisons) of the difference distribution over all iterations included zero, the provinces compared were not considered significantly different. The R code is published in GitHub.
A sensitivity analysis was conducted to explore the influence of the input variables on the output values. Spearman’s rank order correlation coefficients among each of the inputs (provincial horse populations, provincial doses administered, and probability of primary vaccination) and the outputs (vaccination coverages) were obtained from each iteration and plotted in a tornado chart (15). Scatter plots of the randomly selected values for the inputs against the corresponding output value for each iteration were examined.
Results
Parameters of the PERT distributions used to estimate the annual number of vaccine doses administered and the horse population from 2016 to 2020 are presented in Table I. The maximum value for every province, except for Ontario and the Atlantic provinces, was reported in the 2006 EC survey (see Supplementary Material, Table SI). The maximum value was from a 2006 provincial survey in Ontario and the 2010 EC survey in the Atlantic provinces. Data from the 2021 EC survey and the most recent surveys by provincial equestrian federations (Alberta Equestrian Federation, Cheval Québec, and the Horse Council of British Columbia) all showed a decline in the horse population in the last few years.
Table I.
Parameters of the PERT distributions used to estimate the annual number of vaccine doses administered and horse population, Canada, 2016–2020.
| Province | Annual number of vaccine doses administered | Horse population estimates | ||||
|---|---|---|---|---|---|---|
|
|
|
|||||
| Minimum | MLE | Maximum | Minimum | MLE | Maximum | |
| Atlantica | 1084 | 1742 | 2970 | 4741 | 18 500 | 25 850 |
| Quebec | 13 138 | 14 603 | 15 760 | 19 661 | 95 000 | 131 981 |
| Ontario | 42 806 | 47 451 | 52 234 | 64 536 | 200 500 | 379 412 |
| Manitoba | 11 472 | 12 919 | 13 896 | 20 662 | 32 500 | 67 287 |
| Saskatchewan | 11 640 | 13 115 | 13 842 | 39 896 | 85 000 | 127 395 |
| Alberta | 29 778 | 32 821 | 36 616 | 108 702 | 303 000 | 335 980 |
| British Columbia | 10 284 | 10 787 | 11 250 | 33 363 | 87 000 | 125 570 |
MLE — Most likely estimate.
New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador.
The survey of experts regarding the MLE of provincial horse populations was completed by 8 respondents who provided a total of 12 estimates, with a minimum of 1 answer for every province and a maximum of 3 answers for the Atlantic provinces (see Supplementary Material, Table SII). Most answers (10/12) indicated that the 2018 population estimates based on the regression from 2010 to 2021 were too low based on experts’ opinions. One respondent answered “I don’t know” for all the provinces.
The number of doses administered annually remained similar over the study years in most provinces, except in the Atlantic provinces where there was a 2.7-fold increase between the minimum and maximum (Table I).
The median provincial horse population ranged from 17 674 (95% CI: 9737 to 24 060) in the Atlantic provinces to 282 209 (95% CI: 188 576 to 329 837) in Alberta. For the national horse population, the median was 795 420 horses (95% CI: 644 363 to 944 094) (Table II).
Table II.
Median values with 95% credible intervals (CI) of the number of horses vaccinated, horse population, and vaccination coverage in horses from 10 000 Monte Carlo simulations for West Nile virus in Canada, 2016–2020.
| Province | Number of horses vaccinated (95% CI) | Horse population (95% CI) | Vaccination coverage (95% CI) (%) |
|---|---|---|---|
| Atlantica | 1652 (1130 to 2321) | 17 674 (9737 to 24 060) | 10 (6 to 20)1 |
| Quebec | 13 237 (12 265 to 14 189) | 90 029 (46 370 to 123 583) | 15 (11 to 30)1,2 |
| Ontario | 43 147 (39 830 to 46 610) | 206 135 (102 528 to 324 232) | 22 (14 to 44)1,2 |
| Manitoba | 11 686 (10 780 to 12 529) | 35 490 (23 246 to 54 263) | 35 (22 to 53)2 |
| Saskatchewan | 11 830 (10 966 to 12 588) | 84 363 (53 128 to 114 873) | 15 (11 to 24)1 |
| Alberta | 29 934 (27 607 to 32 475) | 282 209 (188 576 to 329 837) | 11 (9 to 17)1 |
| British Columbia | 9801 (9348 to 10 272) | 85 008 (49 952 to 114 881) | 12 (9 to 21)1 |
| Canada (total) | 127 040 (119 530 to 135 101) | 795 420 (644 363 to 944 094) | 16 (13 to 20) |
New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador.
Provincial vaccination coverages with different superscripts are statistically different at P ≤ 0.01.
The estimates of provincial vaccination coverage for WNV in horses ranged from 10% (95% CI: 6 to 20%) in the Atlantic provinces to 35% (95% CI: 22 to 53%) in Manitoba (Table II). The national annual vaccination coverage for WNV in horses was estimated at 16% (95% CI: 13 to 20%). The vaccination coverage in Manitoba was significantly higher (P ≤ 0.01) than in Saskatchewan, Alberta, British Columbia, and the Atlantic provinces.
The tornado chart (see Supplementary Material, Figure S1) and the scatter plots indicated that uncertainty about the provincial horse populations had the most impact on the estimate of annual vaccination coverage for Canada, especially for the 2 provinces with the largest horse population (Ontario and Alberta). Comparatively, the uncertainty about the probability of primary vaccination and the variability in the number of doses administered did not have much impact on the results.
Discussion
This study highlights low annual vaccination coverage for WNV in horses in all the Canadian provinces based on horse populations derived from stochastic modeling and vaccine sales data. The provincial estimates of horse population presented here are the first ones published in Canada to consider the uncertainty surrounding the horse population when estimated with a multisource approach.
An internal survey conducted by CAHSS, based mostly on personal communications with national horse organizations in Canada, estimated the national population at 771 995 horses, which falls within the 95% credible interval of the national horse population presented in this study (CAHSS, personal communication, 2023).
Experts consulted suggested that the 2021 population estimate from EC might be underestimated due to limited survey outreach. It is therefore likely that the horse population reported in the EC surveys exaggerated the decline in population estimates from 2010 to 2021 (13,14) compared to reality. This concern drove the use of expert opinion instead of survey data to set the MLE for provincial horse populations.
The present study indicates that there is a low national annual vaccination coverage for WNV in horses. In 2003, vaccination coverage of up to 75% was estimated in Saskatchewan (7). West Nile virus vaccination coverage in Canada could have been higher in 2003 due to the novelty of the disease, with the first cases reported in 2002 (16), and the new availability of a vaccine (17). Annual vaccination rates may then have declined over time, as infection with the virus became seasonally endemic with fluctuating case numbers (1).
Since risk perception by the horse’s owner is an important factor in the decision to vaccinate (8), it is likely that the annual vaccination coverage follows fluctuations in the number of reported cases. This was not observed in this study, however, probably due to the short study period. Vaccination data obtained over a longer period of time would be required in order to explore this hypothesis.
The Atlantic provinces have the lowest annual vaccination coverage in Canada, which may be explained by the absence of reported cases in these provinces (1,3). Vaccination coverage appears to be highest in Manitoba, which is intriguing when compared to the low coverage in other western provinces. Although annual incidence rates of WNV in Manitoba have not surpassed those of other provinces (1), there has been a notable increase in WNV activity across Canada since 2011, culminating in a peak incidence rate in Manitoba in 2016 (1). It is possible that horse owners in Manitoba perceive a higher risk of WNV than in other provinces, resulting in greater annual vaccination coverage. This could be due to veterinarians actively promoting vaccination or increased awareness among horse owners in this province.
In the latest socioeconomic study conducted by EC (13), the primary use of horses (e.g., recreational, sport, and breeding) was similar in proportion among provinces and would not, therefore, explain the observed provincial differences in annual vaccination coverage. Studies investigating the factors influencing WNV vaccination in Canada, along with historical vaccination sales data, would be required in order to determine whether Manitoba has consistently exhibited higher coverage or if a significant increase occurred at a particular time. It may be challenging to compare annual vaccination coverage with the number of reported WNV cases and incidence rates, as vaccination prevents clinical disease and case numbers can drive vaccination uptake.
This same EC study (13) mentioned the difficulty in accessing veterinary services in several regions across Canada, including Quebec, northern Ontario, central British Columbia, and New Brunswick, which could make it difficult to access and promote WNV vaccination in these areas. Moreover, experts from the CAHSS Equine Surveillance Network mentioned that there are underserviced regions in Alberta. Other factors that could be a barrier to vaccination are a lack of knowledge and/or awareness of WNV (9) and the cost of vaccination.
Presenting vaccination coverages as 95% CI with a stochastic model confers a range of values wherein the true vaccination coverage most probably lies within the limits of the data available. This method accounts for uncertainties in both horse population and horses vaccinated for WNV (based on vaccine doses sold) and offers a more comprehensive integration of the data available. It makes it possible to conclude that the vaccination coverage for WNV in horses is low in Canada, even in the absence of complete data on horse population. As revealed by the sensitivity analysis, more complete data on horse population could help improve the accuracy of the CI estimates and may be considered as more information becomes available.
The validity of the stochastic model outputs relies on some assumptions. First, it was assumed that every horse vaccinated for the first time (primary vaccination) received a series of 2 doses and that the annual booster consisted of only 1 dose. Losses due to broken vials and multidose packages not administered entirely were considered as negligible. Vaccination coverages presented could be overestimated if these assumptions were not met.
It was also assumed that the experts consulted were knowledgeable about the MLE of the horse population and the probability of primary vaccination in their respective provinces. In some provinces where only 1 expert provided input, the level of agreement among experts regarding this parameter could not be evaluated.
The method used in this study highlighted the usefulness of stochastic modeling and the strength of a collaborative approach among equine practitioners, industry, academia, and government authorities in estimating vaccination coverage for WNV and horse populations in Canada. The methods and results may also be useful for other investigations dealing with incomplete animal populations and a high level of uncertainty in the data.
The findings of this study underscore the need to collect more comprehensive data on horse population across Canada. Low annual vaccination coverage for WNV in horses may be due to the perception that the risk or consequences of WNV disease in horses do not justify vaccination. However, WNV is widely distributed among horses in Canada (except in the Atlantic provinces) (2), cases have significantly increased since 2011, and almost one-third of horses that develop WNV disease are likely to die (1). It is therefore recommended that an effort be made to increase awareness of WNV and promote vaccination in Canada, despite the limited access to veterinary services in some areas.
Supplementary Data
Table SI.
Estimates of provincial horse population in Canada from various sources.
| Provinces | Horse population | Year | Source |
|---|---|---|---|
| Atlantica | 4430 | 2022 | Canadian Quarter Horse Association (1) |
| 9008 | 2021 | Equestrian Canada (2) | |
| 25 850 | 2010 | Equestrian Canada (3) | |
| 24 178 | 2007 | Equestrian Canada (3) | |
| 24 602 | 2006 | Equestrian Canada (3) | |
| Québec | 25 318 | 2022 | Cheval Québec (4) |
| 55 058 | 2021 | Equestrian Canada (2) | |
| 129 500 | 2010 | Equestrian Canada (3) | |
| 117 495 | 2007 | Filière Cheval Québec (5) | |
| 126 341 | 2007 | Equestrian Canada (3) | |
| 131 981 | 2006 | Equestrian Canada (3) | |
| Ontario | 127 981 | 2021 | Equestrian Canada (2) |
| 212 500 | 2010 | Equestrian Canada (3) | |
| 221 585 | 2007 | Equestrian Canada (3) | |
| 229 243 | 2006 | Equestrian Canada (3) | |
| 379 412 | 2006 | OMAFRA b | |
| Manitoba | 24 034 | 2021 | Equestrian Canada (2) |
| 56 000 | 2010 | Equestrian Canada (3) | |
| 62 222 | 2007 | Equestrian Canada (3) | |
| 67 287 | 2006 | Equestrian Canada (3) | |
| Saskatchewan | 64 381 | 2021 | Equestrian Canada (2) |
| 108 800 | 2010 | Equestrian Canada (3) | |
| 119 298 | 2007 | Equestrian Canada (3) | |
| 127 395 | 2006 | Equestrian Canada (3) | |
| Alberta | 117 001 | 2022 | Canadian Quarter Horse Association (1) |
| 142 000 | 2021 | Alberta Equestrian Federation (6) | |
| 188 692 | 2021 | Equestrian Canada (2) | |
| 313 500 | 2010 | Equestrian Canada (3) | |
| 315 710 | 2007 | Equestrian Canada (3) | |
| 335 980 | 2006 | Equestrian Canada (3) | |
| British Columbia | 39 224 | 2021 | Equestrian Canada (2) |
| 57 580 | 2019 | Horse Council of British Columbia (7) | |
| 116 000 | 2010 | Equestrian Canada (3) | |
| 123 667 | 2007 | Equestrian Canada (3) | |
| 125 570 | 2006 | Equestrian Canada (3) |
The maximum is identified in bold.
New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador.
Numbers from the Ontario Ministry of Agriculture, Food and Rural Affairs (OMAFRA) that estimated the total horse population from the Census of Agriculture (3.9× census numbers), personal communication with OMAFRA, 2021.
Most likely estimate survey question sent to experts (Table SII)
“For each province, we provide a range of possible estimates based on the minimum population (2016 Census of Agriculture) and maximum population (based on all available information on horse population in Canada). A most likely estimate (proposed MLE) is provided for each province based on the estimated 2018 population. This population estimate was obtained by using a simple linear regression between the 2010 Equestrian Canada study and the one from 2021. Please refer to the presentation to the CAHSS Equine Network on September 29 for more details.
Please answer questions for all provinces for which you have some knowledge on the equine population.
Question is: With the various estimates provided, according to your expertise, do you believe the MLE (most likely estimate) is accurate, too high, or too low? If deemed too high or too low, where would your estimate be situated?”
Example:
Atlantic provinces (PEI, NL, NS, NB)
What would be the most likely horse population to your knowledge?
*** Most likely estimate according to regression: 13 500
○ 4000–7000
○ 7001–10 000
○ 10 001–13 000
○ 13 001–16 000 ***
○ 16 001–19 000
○ 19 001–22 000
○ 22 001–25 000
○ I don’t know
Table SII.
Results from an anonymous survey sent to experts from the equine industry (members of the Canadian Animal Health Surveillance System Equine Network) to gather information on the most likely estimate (MLE) for provincial horse populations, Canada 2016–2020 (N = 8). Final MLE represents the midpoint of the best interval estimate or the average of the midpoints of the best interval estimates when more than one answer was obtained for a single province.
| Province | Proposed MLE (2018) | Expert opinion | Final MLE | ||
|---|---|---|---|---|---|
|
| |||||
| Answer 1 | Answer 2 | Answer 3 | |||
| Atlantica | 13 500 | 13 001 to 16 000 | 16 001 to 19 000 | 22 001 to 25 000 | 18 500 |
| Québec | 75 000 | 80 001 to 95 000 | 95 001 to 110 000 | 95 000 | |
| Ontario | 151 000 | 190 001 to 211 000 | 200 500 | ||
| Manitoba | 32 750 | 25 001 to 30 000 | 35 001 to 40 000 | 32 500 | |
| Saskatchewan | 76 500 | 80 001 to 90 000 | 85 000 | ||
| Alberta | 223 000 | 288 001 to 318 000 | 303 000 | ||
| British Columbia | 60 000 | 69 001 to 81 000 | 93 001 to 105 000 | 87 000 | |
MLE — Most likely estimate.
New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador.
Spearman’s rank correlation coefficient tornado plot of all the provincial input variables (horse populations and vaccine doses administered) and the probability of primary vaccination in vaccinated horses for the national annual vaccination coverage for West Nile virus in horses, Canada, 2016–2020.
AB — Alberta; ATL — Atlantic provinces (New Brunswick, Nova Scotia, Prince Edward Island, and Newfouland and Labrador); BC — British Columbia; MB — Manitoba; ON — Ontario; QC — Québec; SK — Saskatchewan; Horse_pop — Horse population; primo: probablity of primary vaccination in vaccinated horses; doses: vaccine doses administered.
References
- 1.Canadian Quarter Horse Association. Statistics [Internet] [Last accessed January 16, 2025]. Available from: https://cqha.ca/resources/statistics/
- 2.Equestrian Canada. The Canadian equine sector: socio-economic insights 2023. 2023. [Last accessed January 16, 2025]. p. 90. Available from: https://www.equestrian.ca/cdn/storage/resources_v2/fkeEaAYTbatJJwYjJ/original/fkeEaAYTbatJJwYjJ.pdf.
- 3.Evans 2010 Canadian equine industry profile. Equine Canada; 2011. [Last accessed January 16, 2025]. p. 396. Available from: https://www.equestrian.ca/cdn/storage/resources_v2/mzpQQ3p39NRcMys6K/original/mzpQQ3p39NRcMys6K.pdf. [Google Scholar]
- 4.Groupe AGÉCO. Accompagnement pour élaborer une méthodologie d’inventaire des chevaux du Québec. Cheval Québec. 2022. [Last accessed January 16, 2025]. p. 41. Available from: https://www.cheval.quebec/uploads/6/5/5/4/65541265/rapportfinal_inventairechevauxduquebec_groupeageco_20220126.pdf.
- 5.Filière cheval du Québec. Outil d’inventaire des chevaux actifs au Québec. 2008. p. 24. Available from: https://www.cheval.quebec/uploads/6/5/5/4/65541265/inventaire-final2008.pdf.
- 6.Alberta Equestrian Federation. The economic impact of Alberta’s equine community (2021) 2022. [Last accessed January 16, 2025]. p. 48. Available from: https://www.albertaequestrian.com/resources/recreation-and-participation/economic-impact-study.
- 7.Horse Council of British Columbia. Equine industry economic impact study. 2019. [Last accessed January 16, 2025]. p. 52. Available from: https://hcbc.ca/wp-content/uploads/2020/06/3.-HCBC-2019-Economic-Impact-Study-Final-Written-Report-March-2.pdf.
Acknowledgments
The authors thank Boehringer Ingelheim, Zoetis, and Merck for sharing their WNV vaccination data and the equestrian associations that collaborated with the authors. The authors also thank members of the Canadian Animal Health Surveillance System (CAHSS) Equine Surveillance Network for reviewing the method and contributing to the most likely estimates. The authors also extend their appreciation to the Equine Population Working Group, a subdivision of the CAHSS Equine Surveillance Network. Thanks also to Josh Persi for his help in developing the R code.
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;62:469–476. [PMC free article] [PubMed] [Google Scholar]
- 2.Levasseur A, Arsenault J, Paré J. Distribution of West Nile virus cases in horses reveals different spatiotemporal patterns in eastern and western Canada. J Am Vet Med Assoc. 2023:1466–1474. doi: 10.2460/javma.23.05.0259. [DOI] [PubMed] [Google Scholar]
- 3.CAHSS Equine Network — Equine Diseases Dashboard [database on the Internet] Ottawa, Ontario: Canadian Animal Health Surveillance System; [Last accessed January 7, 2025]. [updated December 2023]. Available from: https://cahss.ca/cahss-networks/equine. [Google Scholar]
- 4.American Association of Equine Practioners. West Nile Virus Vaccination Guidelines. AAEP; 2024. [Last accessed January 7, 2025]. Available from: https://aaep.org/resource/west-nile-virus-vaccination-guidelines/ [Google Scholar]
- 5.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;39:498–503. doi: 10.2746/042516407X248476. [DOI] [PubMed] [Google Scholar]
- 6.Cendejas PM, Goodman AG. Vaccination and control methods of West Nile virus infection in equids and humans. Vaccines (Basel) 2024;12:485. doi: 10.3390/vaccines12050485. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.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;55:411–416. doi: 10.1111/j.1865-1682.2008.01051.x. [DOI] [PubMed] [Google Scholar]
- 8.United States Department of Agriculture. Trends in Vaccination Practices on U.S Equine Operations 1998–2015. Animal and Plant Health Inspection Service; 2018. [Last accessed January 7, 2025]. p. 3. Available from: https://www.aphis.usda.gov/animal_health/nahms/equine/downloads/equine15/Equine15_is_Vaccination.pdf. [Google Scholar]
- 9.Galvan R, Rene A, Bae S, Singh K. An Analytical Study of the Perceptions, Prevention Srategies, Treatment and Economic Impact of Equine West Nile Virus. USDA; 2004. [Last assessed January 7, 2025]. Available from: https://www.nass.usda.gov/mexsai/Papers/westnilep.pdf. [Google Scholar]
- 10.Higgins SN, Howden KJ, James CR, Epp T, Lohmann KL. A retrospective study of owner-requested testing as surveillance for equine infectious anemia in Canada (2009–2012) Can Vet J. 2017;58:1294–1300. [PMC free article] [PubMed] [Google Scholar]
- 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;76:263–272. doi: 10.1016/j.prevetmed.2006.05.008. [DOI] [PubMed] [Google Scholar]
- 12.Statistics Canada. Guide to the Census of Agriculture, 2021. Catalogue number 32260002, issue 2021001. Ottawa, Ontario: Government of Canada; 2022. [Last accessed January 7, 2025]. p. 42. Available from: https://www150.statcan.gc.ca/n1/pub/32-26-0002/322600022021001-eng.htm. [Google Scholar]
- 13.Equestrian Canada. The Canadian Equine Sector: Socio-economic Insights 2023. 2023. [Last accessed January 7, 2025]. p. 90. Available from: https://www.equestrian.ca/cdn/storage/resources_v2/fkeEaAYTbatJJwYjJ/original/fkeEaAYTbatJJwYjJ.pdf.
- 14.Evans 2010 Canadian equine industry profile. Equine Canada; 2011. [Last accessed January 7, 2025]. p. 396. Available from: https://www.equestrian.ca/cdn/storage/resources_v2/mzpQQ3p39NRcMys6K/original/mzpQQ3p39NRcMys6K.pdf. [Google Scholar]
- 15.Vose D. Risk Analysis: A Quantitative Guide. 3rd ed. Chichester, England: Wiley; 2008. [Google Scholar]
- 16.Drebot MA, Lindsay R, Barker IK, et al. West Nile virus surveillance and diagnostics: A Canadian perspective. Can J Infect Dis. 2003;14:105–114. doi: 10.1155/2003/575341. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Veterinary Biologics Licensed in Canada [database on the Internet] Ottawa, Ontario: Canadian Food Inspection Agency; [Last accessed January 7, 2025]. [updated June 12, 2024]. Available from: https://apps.inspection.canada.ca/webapps/veterinary-biologics-product-list?lang=en. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table SI.
Estimates of provincial horse population in Canada from various sources.
| Provinces | Horse population | Year | Source |
|---|---|---|---|
| Atlantica | 4430 | 2022 | Canadian Quarter Horse Association (1) |
| 9008 | 2021 | Equestrian Canada (2) | |
| 25 850 | 2010 | Equestrian Canada (3) | |
| 24 178 | 2007 | Equestrian Canada (3) | |
| 24 602 | 2006 | Equestrian Canada (3) | |
| Québec | 25 318 | 2022 | Cheval Québec (4) |
| 55 058 | 2021 | Equestrian Canada (2) | |
| 129 500 | 2010 | Equestrian Canada (3) | |
| 117 495 | 2007 | Filière Cheval Québec (5) | |
| 126 341 | 2007 | Equestrian Canada (3) | |
| 131 981 | 2006 | Equestrian Canada (3) | |
| Ontario | 127 981 | 2021 | Equestrian Canada (2) |
| 212 500 | 2010 | Equestrian Canada (3) | |
| 221 585 | 2007 | Equestrian Canada (3) | |
| 229 243 | 2006 | Equestrian Canada (3) | |
| 379 412 | 2006 | OMAFRA b | |
| Manitoba | 24 034 | 2021 | Equestrian Canada (2) |
| 56 000 | 2010 | Equestrian Canada (3) | |
| 62 222 | 2007 | Equestrian Canada (3) | |
| 67 287 | 2006 | Equestrian Canada (3) | |
| Saskatchewan | 64 381 | 2021 | Equestrian Canada (2) |
| 108 800 | 2010 | Equestrian Canada (3) | |
| 119 298 | 2007 | Equestrian Canada (3) | |
| 127 395 | 2006 | Equestrian Canada (3) | |
| Alberta | 117 001 | 2022 | Canadian Quarter Horse Association (1) |
| 142 000 | 2021 | Alberta Equestrian Federation (6) | |
| 188 692 | 2021 | Equestrian Canada (2) | |
| 313 500 | 2010 | Equestrian Canada (3) | |
| 315 710 | 2007 | Equestrian Canada (3) | |
| 335 980 | 2006 | Equestrian Canada (3) | |
| British Columbia | 39 224 | 2021 | Equestrian Canada (2) |
| 57 580 | 2019 | Horse Council of British Columbia (7) | |
| 116 000 | 2010 | Equestrian Canada (3) | |
| 123 667 | 2007 | Equestrian Canada (3) | |
| 125 570 | 2006 | Equestrian Canada (3) |
The maximum is identified in bold.
New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador.
Numbers from the Ontario Ministry of Agriculture, Food and Rural Affairs (OMAFRA) that estimated the total horse population from the Census of Agriculture (3.9× census numbers), personal communication with OMAFRA, 2021.
Table SII.
Results from an anonymous survey sent to experts from the equine industry (members of the Canadian Animal Health Surveillance System Equine Network) to gather information on the most likely estimate (MLE) for provincial horse populations, Canada 2016–2020 (N = 8). Final MLE represents the midpoint of the best interval estimate or the average of the midpoints of the best interval estimates when more than one answer was obtained for a single province.
| Province | Proposed MLE (2018) | Expert opinion | Final MLE | ||
|---|---|---|---|---|---|
|
| |||||
| Answer 1 | Answer 2 | Answer 3 | |||
| Atlantica | 13 500 | 13 001 to 16 000 | 16 001 to 19 000 | 22 001 to 25 000 | 18 500 |
| Québec | 75 000 | 80 001 to 95 000 | 95 001 to 110 000 | 95 000 | |
| Ontario | 151 000 | 190 001 to 211 000 | 200 500 | ||
| Manitoba | 32 750 | 25 001 to 30 000 | 35 001 to 40 000 | 32 500 | |
| Saskatchewan | 76 500 | 80 001 to 90 000 | 85 000 | ||
| Alberta | 223 000 | 288 001 to 318 000 | 303 000 | ||
| British Columbia | 60 000 | 69 001 to 81 000 | 93 001 to 105 000 | 87 000 | |
MLE — Most likely estimate.
New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador.
Spearman’s rank correlation coefficient tornado plot of all the provincial input variables (horse populations and vaccine doses administered) and the probability of primary vaccination in vaccinated horses for the national annual vaccination coverage for West Nile virus in horses, Canada, 2016–2020.
AB — Alberta; ATL — Atlantic provinces (New Brunswick, Nova Scotia, Prince Edward Island, and Newfouland and Labrador); BC — British Columbia; MB — Manitoba; ON — Ontario; QC — Québec; SK — Saskatchewan; Horse_pop — Horse population; primo: probablity of primary vaccination in vaccinated horses; doses: vaccine doses administered.