Survey of the equine broodmare industry, abortion, and equine herpesvirus-1 vaccination in Ontario

Abstract

A survey of Ontario horse breeders was conducted in 2016 and retrospectively asked farm-level management questions regarding herd sizes, vaccination, respiratory disease, and abortion over the years 2014 to 2015. A total of 88 farm owners completed the survey, reporting 857 broodmares. Using logistic regression modelling, management influences on vaccine use, and the reporting of respiratory disease or abortion was investigated. Having veterinary records and the reasons for breeding were significantly associated with the odds of an abortion even after controlling for broodmare herd size. The odds of having respiratory illness on the farm were significantly greater if the farm was the primary source of income even after controlling for farm size. Respondents with primary breeding operations were more likely to utilize vaccines against equine herpesvirus 1 (EHV-1), independent of herd size. Veterinarians were more involved with abortions than with respiratory disease, leaving a significant gap in the opportunities for client education.

Introduction

Reproductive efficiency and low abortion rates are good indicators of success in the equine breeding industry (1), while outbreaks causing abortions on equine breeding farms can cause catastrophic loss of time, money, and life in both mares and foals, as was evidenced by mare reproductive loss syndrome experienced by breeders in Kentucky in 2001 (2). Regions that are heavily reliant on horse breeding will use seasonal analysis of reproductive performance outcomes such as pregnancy rate and live foal rate to ensure a healthy industry, and to raise alarms during potential outbreaks (1,3). Despite improvements in management knowledge and in conception rates, abortions are still the leading cause of financial loss to the equine industry (3). Overall abortion rates per breeding season have ranged from 9.1% to 21.7% (1,3–6).

Known host risk factors for equine abortion that have been previously investigated include: age, breed, parity, method of insemination, rate of twinning (7), racing or competition (8,9), vaccination (1,10), and endocrine disease (11–13). Environmental factors that have been investigated include season of breeding, light exposure, and stress (4,14). One of the environmental risk factors that has not been addressed is overall herd size and the influence herd size has on management style. In addition to host and environmental effects, pathologies of the umbilicus or placenta causing insufficient nutrition of the fetus, and infectious agents have been implicated when analyzing aborted tissues (15,16). In consideration that many of these infectious agents are contagious, herd management recommendations have also evolved to include improved mare housing, sanitation, and infection control to limit pathogen spread (17,18). Herd management recommendations are made based on regional risks and diseases. Both large and small breeding operations have been affected by abortion stemming from any of the risks listed, despite intensive management.

One of the main infectious agents involved in reproductive loss is equine herpesvirus-1 (EHV-1), which has been implicated in as many as 8.9% of abortions in Germany (19) and 18% of abortions submitted for analysis in 2003 to 2004 in Ontario, Canada (20). Of all the infectious causes of abortion, EHV-1 is the only pathogen for which we have a vaccine labeled for protection during gestation. EHV-1 can lead to reproductive, neurological, and respiratory disease in horses worldwide, and affects any age group. EHV-1 can be transmitted to foals by their dams, and can later recrudesce in response to unknown stimuli (21). Widespread vaccine use is assumed, as EHV-1 vaccination is considered a core vaccine by the American Association of Equine Practitioners (AAEP) and is recommended during gestation as well (22). However, despite vaccination, outbreaks have continued to occur. For example, in 2017, a farm in Peterborough, Ontario with a small breeding operation was affected by all 3 forms of EHV-1, neurologic and respiratory disease, as well as late-term abortions (C. Cooper, 2017, personal communication). Larger farms have also experienced EHV-1-associated abortion storms in Ontario as recently as 2019 (L. Arroyo, 2019, personal communication), and in Germany (23). Most recently, the MHC class 1B2 allele was implicated as a risk factor in mares which abort due to EHV-1 when controlling for environmental and host factors (24), adding a new facet of mare genetics to the risk factors of EHV-associated reproductive loss.

At this time, there are no publications on the health of the Ontario breeding industry outside of what has been published by the Thoroughbred and Standardbred breeding associations. Despite the high periodic incidence of EHV-1-associated abortion in Ontario, the proportion of farms using the EHV-1 vaccine is unknown, as is the level of concern regarding EHV-1 among breeders. The objectives of this study were to use survey data to describe the current state of the Ontario broodmare industry in terms of sources of income, herd sizes, and abortion incidence and to investigate how various farm-level management factors influence the reporting of vaccine use, respiratory disease and abortions.

Materials and methods

Database of breeders

A database of Ontario horse breeders was created by identifying breeders through Internet searches and public online forums. Keyword searches included: “Ontario horse breeder,” “equine breeder Ontario,” “foal sale Ontario,” “Ontario foals,” “Ontario breeder+breed,” “breed+,” “association Ontario,” and their variations. Social media, specifically Facebook, was also used to contact breeders, as many farm webpages identified by Internet searches had Facebook links on their homepage. A recruitment script was approved by the University of Guelph Research and Ethics Board (REB, Approval #16FE021) and included an anonymized link to the survey.

A farm was added to the database if: foal sales were advertised, broodmare management was identified explicitly, horses under the age of 2 y were sold, or images of foals were on the website. Once a farm was identified by name, searches were conducted using the Internet, publicly accessible social media pages, and public membership registry lists to create the database of breeds, farm addresses, city, owner or manager name, any associated phone numbers, e-mail addresses, and primary veterinarian (if available). E-mail was used for ease of survey dissemination. All farms that were identified, but did not have an associated e-mail address, were contacted by telephone to complete the survey.

From the initial database of 664 businesses, 80 had no valid contact information, 5 were not interested in participating, 33 farms reported no current breeding activity and were excluded, and 67 could not receive e-mails and had no public phone number. Consequently, 479 businesses were contacted by e-mail or telephone to complete the survey.

Survey

An online questionnaire titled “Surveillance of Ontario broodmares and abortion occurrences” was created using Qualtrics Survey Software (Seattle, Washington, USA) (Supplementary Item 1, available from the authors). The survey was approved by the University of Guelph REB (Approval #16FE021).

The questionnaire was disseminated by e-mail directly to the breeders identified in our database, through social media, and by a uniform resource locator (URL) from February to May 2016. The questionnaire, containing 23 questions, was designed to take, on average, less than 10 min to complete. Questions were primarily in multiple selection format, but a text box for further comment was available when “Other” was selected. Questions regarding herd demographics and abortions were numeric short answer. Medical records, veterinary treatments, and sample submission were investigated using a 5-point ordinal scale of potential responses: “never,” “rarely,” “sometimes,” “often,” and “always.” Short answer space was also made available for identifying the geographic location of the farm.

Additional information was gathered concerning the number of horses, broodmares, and foals on the property per year, as well as abortions if they occurred. Farm management questions asked about medical records management, veterinary involvement, EHV-1 vaccination, and the submission of diagnostic samples from ill horses and abortive tissues to investigate etiologic agents. EHV-1 vaccination was further investigated by identifying which vaccination protocols were most commonly used. The most common options were included for selection in the questionnaire: “once” and “twice” annual vaccination are self-evident, and the “pre-foaling protocol” was used to describe the AAEP recommendation of vaccination at 5, 7, and 9 mo of gestation using vaccines labelled for EHV-1 protection against abortion.

Lastly, permission to contact breeders concerning future studies relating to EHV-1 and abortion investigations was requested, and a short answer space for contact information was provided. After final survey dissemination, at least 3 attempts at contact were made to ensure maximum participation. Of those who wished to participate, 7 requested to complete the survey over the phone rather than online.

Statistical analysis

Data were downloaded from the Qualtrics software and stored in a spreadsheet using Microsoft Office Excel (2010, Microsoft Corp., Redmond, Washington, USA). Descriptive statistics including medians, ranges, and interquartile ranges (IQR) were reported for continuous variables, and proportions with their 95% confidence intervals (CI) were generated for categorical variables. Exact logistic regression was used to examine the association between herd size and the likelihood of being a business, using the farm revenue as a primary source of income, and medical record management. Exact logistic regression was also used to examine the association between the respondents’ demographic and management styles and the following dependent variables: respiratory illness and vaccination. To examine the associations between these factors and the odds of an abortion, we used multi-level logistic regression models with a random intercept for herd/facility. For all these models, the results of univariable models were compared to models in which the effect of herd size (broodmare herd size for abortions) was controlled. All measures of herd size were categorized into tertiles, and only the measure of population size most relevant to the outcome of interest was explored in our analyses. For each model, the odds ratio (OR) and 95% CI of each variable were reported. For the multi-level logistic regression models, the intra-class correlation coefficient (ICC) was also reported. In addition, for these multi-level models, model fit was assessed by examining the assumptions of homogeneity of variance and normality of the best linear unbiased predictors (BLUPS) graphically; Pearson residuals were also examined to identify outlying observations/covariate patterns. If the assumptions concerning the BLUPS were not met, we compared the fit of the multi-level model to an ordinary logistic regression model to determine if model fit was improved by the inclusion of the random effect using Akaike’s information criterion (AIC). No statistical analyses were conducted on the EHV-1 incidence in samples submitted from sick or aborted tissues due to the rarity of the outcome being reported. All statistical tests were conducted using STATA (STATA Intercooled 15.0; StataCorp, College Station, Texas, USA).

Results

Respondents

Surveys were collected from 101 farms, but only 88 met the inclusion criteria and were complete. Most of the respondents were from southern Ontario (94.3%, 83/88), with 39.8% (33/83) in southwestern Ontario, 30.1% (25/83) specifically from the Golden Horseshoe region of southwestern Ontario (the most densely populated region of Ontario that stretches from the western end of Lake Ontario, south to Lake Erie and north to Lake Scugog), 19.3% (16/83) from eastern Ontario, and 10.8% (9/83) from the central Ontario region. Five respondents were from northern Ontario (Figure 1).

Figure 1.

Map of Ontario broodmare survey (2014–2015) respondents’ geographical farm locations.

For 78.4% (69/88) of respondents, breeding was for business purposes, but only 35.2% (31/88) of respondents indicated their operations to be strictly breeding facilities. Most farms (58.0%; 51/88) gained income from other sources, such as farming, training, or boarding (Table 1). When “other” was selected, crop farming, off-site jobs, retirement funds and veterinary services were included as additional sources of income (Table 1).

Table 1.

Summary of Ontario broodmare farm survey (2014–2015) results — categorical variables.

Questions	Options	N	Percent	95% Confidence interval
Reason for breeding	Hobby	12	13.6	7.8 to 22.7
	Business	69	78.4	68.4 to 85.9
	Other	7	8.0	2.8 to 16.0
Primary income	Yes	31	35.2	25.8 to 45.9
	No	57	64.8	54.1 to 74.2
Sources of income	Strictly breeding	31	35.2	25.8 to 45.9
	Boarding	24	27.3	18.9 to 37.7
	Training	14	15.9	9.6 to 25.3
	Farming	11	12.5	7.00 to 21.4
	Other	30	34.1	24.3 to 45.0
	Unknown	3	3.4	0.7 to 9.6
Medical records	By vet	50	56.8	46.1 to 66.9
	On farm	79	89.8	81.3 to 94.7
	None	1	1.1	0.2 to 7.9
Vaccinate EHV-1	Yes	64	72.7	62.3 to 81.1
	No	24	27.3	18.9 to 37.7
Vaccine protocols used on farm	Once per year	40	62.5	49.8 to 73.7
	Twice per year	15	23.4	14.5 to 35.7
	Pre-foaling	50	78.1	66.0 to 86.8
Respiratory illness	Yes	36	40.9	33.3 to 57.9
	No	49	55.7	42.1 to 66.7
	Unknown	3	3.4	0.7 to 9.6
Treated by veterinarian	Never	4	11.1	3.1 to 26.1
	Sometimes	25	69.4	51.9 to 83.7
	About half	4	11.1	3.1 to 26.1
	Most of the time	1	2.8	0.1 to 14.5
	Always	2	5.6	0.7 to 18.7
Viral testing pursued	Yes	9	25.0	12.1 to 42.2
	No	25	69.4	51.9 to 83.7
	Unknown	2	5.6	0.7 to 18.7
EHV-1 positive	None	8	88.9	51.8 to 99.7
	1	0	0.0	0.0 to 33.6
	2	1	11.1	0.3 to 48.3
Veterinary assistance for abortions	Never	9	10.2	4.8 to 18.5
	Sometimes	7	8.0	3.3 to 15.7
	About half	0	0.0	0.0 to 4.1
	Most of the time	5	5.7	1.9 to 12.8
	Always	64	72.7	62.2 to 81.7
	Unknown	3	3.4	0.7 to 9.6
Abortions in 2 years	Yes	35	39.8	29.9 to 50.5
	No	53	60.2	49.5 to 70.1
Submission of abortive tissues	Yes	17	48.6	31.4 to 66.0
	No	18	51.4	34.0 to 68.6
EHV-1 positive	2015	1	9.1	0.2 to 41.3
	2014	4	40.0	12.2 to 73.8
	None	7	63.6	30.8 to 89.1
	Unknown	1	9.1	0.2 to 41.3

The size of the farm varied widely, from 3 to 200 horses (Table 2). The median herd size was similar for the 2 y investigated: total herd size median was 19.5 in 2015 and 20.5 in 2014, broodmares and foal numbers remained stable with medians of 5 and 3, respectively, for both years. Median total herd size was divided into small (< 13.5), medium (13.5 to 30.0), and large (> 30.0). Similarly, median broodmare herd size was divided into small (< 4), medium (4 to 8), and large (> 8) for subsequent analyses using exact logistic regression.

Table 2.

Summary of Ontario broodmare farm survey (2014–2015) results — continuous variables.

Variable	Year	N	Median	Range (IQR)
Total herd sizea	2015	2468	19.5	3–200 (4–80)
	2014	2449	20.5	0–200 (6–80)
Broodmaresa	2015	857	5	0–58 (2–35)
	2014	827	5	0–75 (2–35)
Foalsa	2015	589	3	0–60 (0–30)
	2014	540	3	0–50 (0–26)
Total number of abortions	2015	55	1	0–8 (0–3)
	2014	30	0.5	0–5 (0–3)
Early embryonic failures	2015	28	0	0–8 (0–2)
	2014	13	0	0–3 (0–2)
Mid-gestation	2015	6	0	0–2 (0–1)
	2014	8	0	0–4 (0–1)
Late gestation	2015	20	0	0–5 (0–1)
	2014	10	0	0–5 (0–1)
Perinatal deaths	2015	2	0	0–1 (0–0)
	2014	4	0	0–3 (0–0)

^aContinuous variables were described using median, range and interquartile range (IQR — Interquartile range — Included as Q1–Q3) due to skewed data.

N — Total number.

Farms that reported their breeding operations as businesses included 63.3% (19/30) of the small, 79.3% (23/29) of the medium, and 93.1% (27/29) of the largest herd sizes. Only farms run as businesses reported their breeding operation as a primary source of income. The largest herds were more likely to be a primary source of income compared to the small (OR = 5.67, 95% CI: 1.78 to 18.08, P = 0.003) and medium (OR = 3.72, 95% CI: 1.24 to 11.17, P = 0.019) herd sizes.

Veterinary involvement

Based on the sources of income for the respondents, 2 veterinarians participated in the survey as they identified “veterinary services” as other sources of income to the breeding operation. Medical records were most often kept on farm [89.8%, (79/88)], although veterinary records were also popular [56.8%, (50/88)]. Only 1 respondent indicated no medical records were kept (Table 1). Most respondents had farm records regardless of herd size or function; however, when the total herd size was large, farms were less likely to rely on their veterinarian’s records compared to small (OR = 0.31, 95% CI: 0.11 to 0.89, P = 0.029), and medium sized herds (OR = 0.32, 95% CI: 0.11 to 0.94, P = 0.038).

A veterinarian was only “sometimes” requested to examine upper respiratory illness on 69.4% (25/36) of farms, and at least “about half ” of the time in 19.44% (7/36) of farms. In contrast, when abortions occurred, a veterinarian would be involved “always” on 72.73% (64/88) of farms.

Abortions

The farms reported 56 abortions having occurred in 2015, 28 of which were early embryonic losses, 6 occurred during mid-gestation, 20 during late gestation, and 2 were perinatal deaths (Table 2). Similar values were reported for 2014, when 35 abortions occurred, 13 embryonic losses, 8 mid-gestation and 10 late-term abortions, and 4 perinatal deaths (Table 2). The overall incidence of abortion per reported pregnancy was 8.5% in 2015 and 6.1% in 2014. Of the 39.8% (35/88) of farms which had experienced abortions in the past 2 y, 48.6% (17/35) had submitted fetal or placental tissues for pathogen testing. Only 1 was attributed to EHV-1 infection in 2015, and 4 in 2014.

Based on unadjusted multi-level models and those adjusted for broodmare herd size, the odds of an abortion were significantly higher if the facility used veterinary records, and they were significantly lower for facilities for which the reason for breeding was classified as “business” compared to “other” (Table 3). The intra-class correlation coefficient for abortion status for mares within a herd were relatively high and exceeded 30% in our all models (Table 3). The assumption of constant variance of the BLUPS was met, but the BLUPS did not meet the assumption of normality in most of the models. However, the fit of the model, based on lower AICs, was always improved with the inclusion of the random effect for farm.

Table 3.

Results^a of multi-level logistic regression models examining the associations between a mare aborting and farm demographics and management based on the 2014–2015 survey.

Variable	Categories	Unadjusted				Adjusted for broodmare herd size
		OR	95% CI	P-value	ICC (95% CI)	OR	95% CI	P-value	ICC (95% CI)
Broodmare herd sizeb	Referent = Small herd (< 4)				0.41 (0.25 to 0.59)
	Medium herd (4 to 8)	1.58	0.39 to 6.46	0.524
	Large herd (> 8)	1.19	0.31 to 4.56	0.801
Reason for breedingc	Referent = Other				0.34 (0.19 to 0.54)				0.35 (0.19 to 0.54)
	Hobby	0.23	0.03 to 1.96	0.178		0.25	0.03 to 2.28	0.220
	Business	0.11	0.02 to 0.64	0.014		0.07	0.01 to 0.51	0.008
Primary income	Yes versus No	0.58	0.23 to 1.46	0.244	0.39 (0.23 to 0.57)	0.51	0.18 to 1.46	0.209	0.39 (0.23 to 0.58)
Primarily breeding	Yes versus No	1.23	0.47 to 3.22	0.677	0.41 (0.25 to 0.59)	1.25	0.47 to 3.32	0.649	0.41 (0.25 to 0.59)
Use of veterinary records	Yes versus No	2.77	1.06 to 7.27	0.038	0.39 (0.24 to 0.57)	2.98	1.03 to 8.57	0.043	0.40 (0.24 to 0.58)
Use of farm records	Yes versus No	0.65	0.11 to 3.63	0.619	0.41 (0.25 to 0.59)	0.57	0.09 to 3.52	0.542	0.41 (0.25 to 0.59)
Use of EHV-1 vaccine	Yes versus No	2.46	0.68 to 8.93	0.171	0.42 (0.26 to 0.59)	2.55	0.70 to 9.28	0.154	0.41 (0.25 to 0.59)
Respiratory illness	Yes versus No	1.10	0.42 to 2.97	0.848	0.41 (0.26 to 0.59)	1.10	0.42 to 2.86	0.852	0.41 (0.25 to 0.59)

^aResults in the table include unadjusted and adjusted models for broodmare herd size. All models include a random intercept for herd and the intra-class correlation coefficient (ICC) concerning the correlation in abortion status among mares that were pregnant in the same herd.

^bMedium versus Large: OR = 1.33, 95% CI: 0.46 to 3.86, P = 0.600.

^cHobby versus Business: i. Unadjusted: OR = 2.06, 95% CI: 0.51 to 8.41, P = 0.312; ii. Adjusted: OR = 3.51, 95% CI: 0.65 to 18.88, P = 0.143.

CI — Confidence interval.

Bold indicates statistical significance P < 0.05.

Illness

Respiratory disease, defined as nasal discharge, cough and/or fever, was present on 40.9% (36/88) of farms. Only 25.0% (9/36) of farms with respiratory illness were submitting samples for viral pathogens testing. Of those farms with illness reported, 52.8% (19/36) had experienced abortions in the previous 2 y.

When exploring the unadjusted univariable models, total herd size and farms used as primary sources of income were significantly associated with the presence of respiratory illness on farm (Table 4). The odds of respiratory illness were significantly greater in the larger herds compared to smaller ones (Table 4). Operations that were primary sources of income had increased odds of reporting illness than those that had additional sources of income (Table 4). When controlling for herd size, the significant association between primary income source and increased odds of reporting respiratory disease persisted, and no significant difference was identified between herd sizes (Table 4). In contrast to abortions, the use of veterinary medical records was not significantly associated with increased odds of having had respiratory illness on the farm when herd size was controlled (Table 4). Neither strict breeding facilities nor farms considered businesses were associated with increased odds of illness, and vaccination did not significantly affect the odds of reporting illness (Table 4).

Table 4.

Results^a of exact logistic regression models examining the associations between farm demographics and management and presence of respiratory illness on farm based on the 2014–2015 survey.

Variable	Categories	% Positive (n/n)	Unadjusted			Adjusted for farm size
			OR	95% CI	P-value	OR	95% CI	P-value
Total herd size	Referent = Small herd (< 13.5)	30.0 (9/30)
	Medium herd (13.5 to 30)	31.0 (9/29)	1.05	0.30 to 3.68	> 0.99
	Large herd (> 30)	62.1 (18/29)	3.73	1.14 to 13.02	0.019
Reason for breeding	Referent = Other	28.6 (2/7)
	Hobby	41.7 (5/12)	1.73	0.18 to 25.41	0.656	1.37	0.13 to 20.29	> 0.99
	Business	42.0 (29/69)	1.80	0.27 to 20.14	0.694	1.03	0.14 to 12.24	> 0.99b
Primary income	Yes versus No	64.5 (20/31)	4.57	1.66 to 13.25	0.001	3.43	1.18 to 10.26	0.013
Primarily breeding	Yes versus No	48.4 (15/31)	1.60	0.60 to 4.27	0.365	1.57	0.57 to 4.39	0.323b
Use of veterinary records	Yes versus No	46.0 (23/50)	1.63	0.63 to 4.32	0.284	2.79	0.93 to 9.60	0.060b
Use of EHV-1 vaccine	Yes versus No	45.3 (29/64)	2.00	0.67 to 6.52	0.225	1.85	0.58 to 6.37	0.304b

^aResults in the table include unadjusted univariable analyses and analyses adjusted for farm size.

^bThe odds of reporting respiratory disease on a farm were significantly greater in large herds compared to medium and small herds in these models.

OR — Odds ratio; CI — Confidence interval.

Bold indicates statistical significance P < 0.05.

Vaccination

In total, 72.7% (64/88) of respondents were vaccinating for EHV-1. The most popular vaccination protocol was the “twice annual” and “pre-foaling” series. There was no significant association between the herd size and the odds of vaccination (Table 5). Furthermore, neither using the farm as a primary source of income nor as a business was associated with vaccination (Table 5). However, strict breeding operations had 5.35 times greater odds of vaccinating than farms that had other uses (95% CI: 1.39 to 30.83, P = 0.011), and this association persisted after controlling for total herd size (Table 5). No significant differences between herd sizes was identified. As seen with the presence of abortion and respiratory illness, the odds of vaccination were greater in farms which used veterinary medical records, once we controlled for total herd size (Table 5). After controlling for use of veterinary records, farms with large herds had significantly greater odds of vaccinating for EHV-1 compared to medium herds (OR = 3.78, 95% CI: 0.96 to 17.07, P = 0.041), but there was no significant difference in the odds of vaccine use when compared to small herds (OR = 1.31, 95% CI: 0.28 to 6.18, P = 0.745).

Table 5.

Results^a of exact logistic regression models examining the associations between farm demographics and management and presence of EHV-1 vaccination on farm based on the 2014–2015 survey.

Variable	Categories	% Positive (n/n)	Unadjusted			Adjusted for farm size
			OR	95% CI	P-value	OR	95% CI	P-value
Total herd size	Referent = Small herd (< 13.5)	80.0 (24/30)
	Medium herd (13.5 to 30)	58.6 (17/29)	0.36	0.09 to 1.29	0.095
	Large herd (> 30)	79.3 (23/29)	0.96	0.22 to 4.17	> 0.99
Reason for breeding	Referent = Other	85.7 (6/7)
	Hobby	66.7 (8/12)	0.35	0.01 to 4.93	0.603	0.36	0.01 to 5.27	0.597
	Business	72.5 (50/69)	0.44	0.01 to 4.04	0.668	0.46	0.01 to 4.71	0.656
Primary income	Yes versus No	83.9 (26/31)	2.57	0.79 to 9.95	0.132	2.57	0.73 to 10.61	0.114
Primarily breeding	Yes versus No	90.3 (28/31)	5.35	1.39 to 30.83	0.011	5.19	1.33 to 30.25	0.010
Use of veterinary records	Yes versus No	80.0 (40/50)	2.31	0.81 to 6.84	0.095	2.94	0.95 to 9.82	0.044b

^aResults in the table include unadjusted univariable analyses and analyses adjusted for farm size.

^bThe odds of a farm vaccinating for EHV-1 were significantly greater in large herds compared to medium herds (OR = 3.78, 95% CI: 0.96 to 17.07, P = 0.041), but not significantly different than small herds (OR = 1.31, 95% CI: 0.28 to 6.18, P = 0.745) in this model.

Bold indicates statistical significance P < 0.05.

Discussion

Since the census performed by Equine Canada in 2011 (25), no industry-wide questionnaires have been completed to assess the current status of horse breeding in Ontario. The participation of 88 farm owners in this survey provided a small window of information on the current state of the industry in terms of operation characteristics, health management, and disease and abortion experiences.

In the Equine Canada census of 2011 (25), 5225 participants reported owning mares of breeding age (average 6.61 mares/owner), approximately 34 538 broodmares in Ontario, but only 30 175 horses in total were reported as “breeding stock” (including stallions), and only 3779 foals were reportedly born in 2006. At the end of the census, participants had anticipated a reduction in foal production of approximately 15% for 2011. A 2016 Census for Agriculture identified that consolidation is also occurring across Canada: 193 492 farms were counted in 2016, down 5.9% from the previous census in 2011 (26).

The database for this project identified only 664 Ontario breeding operations in 2016. A total of 88 farm owners completed the survey and reported a total of 857 broodmares. By providing equal access to the survey to all identified farms, this small subset potentially provided a representative sample of the industry in Ontario. In total, 589 foals were born on the properties which participated in the survey over the 2-year period of study. The risk of abortion in our 2015 population was 8.5%, like the total gestational loss values of other countries, such as France, with 9.1% (4), and Japan, with 8.7% (5). Only 1 of the abortions in 2015 was positive for EHV-1, whereas 4 were identified in 2014. Based on the results, it appeared that the odds of a mare having experienced an abortion were significantly lower if the primary reason for breeding was business compared to those in which the primary reason was other than “breeding” and “business.” While we are cautious based on the small number of farms given the “other” classification, we suspect biosecurity measures may differ among farms based on their reasons for breeding.

Of those who participated, 72.7% of farms reported vaccinating for EHV-1. Interestingly, 85.7% of abortions (30/35), and 80.6% of respiratory illness (29/36) occurred in these vaccinated herds. Without more consistent submission of both respiratory and abortive tissue samples, it would be impossible to distinguish the underlying causes, but farms which confirmed EHV-1-associated abortions in this survey were also vaccinating. In other studies, foals continued to be infected with EHV-1 at a similar rate despite adoption of widespread vaccination (27). Even after repeated vaccination for EHV, 9/55 abortions were caused by EHV-1 in Germany (28). Similarly, it has been reported that “winter pyrexia,” blamed partially on EHV-1, was reduced when 95% of the population was vaccinated, but outbreaks continued to occur (29). More likely, incomplete protection by vaccination, as seen by repeated outbreaks within vaccinated herds, may be involved (10,30). It is possible that farms which experienced abortion or respiratory disease began vaccinating to control for EHV-1 empirically. Vaccination, however, was not associated with the odds of abortions or the odds of illness occurring on the farm once we controlled for the effect of herd size. Off-label use may also have affected the quality of immune response, as some farms were not using vaccines recommended for control of abortion and/or not using the manufacturer-recommended vaccine protocols. The outcomes reviewed in this study were non-specific (i.e., abortion versus abortion due to EHV-1), which inherently could lead to non-differentiated misclassification of disease and result in bias towards the null.

Furthermore, herd size was not associated with the use of vaccination against EHV-1; only primary breeding facilities had increased odds of reporting vaccine use compared to farms which had other sources of revenue, no matter their size. To improve the efficacy of vaccines in broodmares, vaccination every 2 mo during the higher risk period of pregnancy is recommended (22). As EHV-1 is only 1 of the many causes of abortion, it would make most sense that facilities focused on breeding would expend as much effort as possible, including vaccinating, to reduce any risk to the foal crop. However, monthly vaccination for 3 mo in 2 racetrack populations in Japan provided no additional increase in viral neutralization titers, and only an additional 2% to 3% of horses seroconverted with additional vaccinations (29). More recently, serum titers were followed in mares after each vaccination, and titers even declined despite repeated vaccination (31). Previous groups found that only 33% to 35% of vaccinated horses seroconverted after vaccination (29,32), which is consistent with the findings of up to 69.8% “non-responders” in the study by Foote et al (33). Despite this evidence to the contrary, vaccination is still considered as a key factor in mitigating the effect of EHV-1-associated disease in horses by the industry.

In Canada, vaccines are only available through a veterinarian; therefore, the increased association between farms which vaccinate, and veterinary medical record use is expected. Veterinarians were often involved when abortions occurred but less frequently when respiratory illness was reported; therefore, the increased odds of reporting an abortion in farms with veterinary records was also expected. Few of the participating farms elected to submit respiratory samples or aborted tissues for testing during the 2 years of the survey, limiting the ability to confirm the threat of EHV-1 on the property and the association with future illness or abortions. The lack of sample submissions also limits the quality of our laboratory surveillance programs and may influence the significance of reports of EHV-1 in Ontario, as only samples with higher suspicion may be submitted.

A major limitation of this database was the use of the Internet as a primary source for identifying potential breeders, eliminating those who do not use e-mail or publish their business information online. In addition, organizations occasionally require fees from their members, limiting those who register their business on the organization’s website. Telephone recruitment was used to contact those with no e-mail address, and as a follow-up to improve participation. As with any survey with voluntary participation, farms had control of their responses (misclassification bias), and the option to not participate at all (non-response bias). As with any other retrospective survey, misclassification in the form of a recall bias may have skewed any specific numbers provided in the survey. A limitation of using cross-sectional study designs is being unable to decipher the relative timing of exposures, in this case vaccination, and disease or abortion. Similarly, only farms with interest in research, or those which have had issues with abortions or disease may have wanted to participate in the survey, and if this difference in participation was linked to different exposures it could lead to a selection bias. The wording of the survey title and description attempted to limit this possibility by emphasizing the general management practice interest, rather than abortion or vaccination specifically. Furthermore, the accessibility of the survey was maximized using e-mail and Internet links, and telephone follow-up in case of limited access.

In summary, this study provides information about a subset of the equine breeding industry in Ontario: its herd sizes and management styles. Abortions occurred on 39.8% of the properties, most of which were in herds vaccinated for EHV-1, but the overall abortion risk of 8.5% for individual mares per reported pregnancy in this survey was similar to that in other countries. Vaccination was not associated with the odds of reporting abortions or respiratory illness, once we controlled for herd size. Only strict breeding operations were significantly more likely to vaccinate for EHV-1, regardless of herd size. Veterinarians were more often involved when abortions occurred compared to respiratory disease, despite there being increased odds of abortions on farms which reported illness, especially in large herds. Few respiratory samples and abortive tissues were submitted for diagnostic testing, limiting the surveillance ability of our regional laboratories. From the 2011 Equine Canada census, only 31% of participants said they “would expect to learn about a contagious disease from their veterinarian”, and 41% said they would learn about it “by word-of-mouth.” There remains a significant opportunity for veterinarians to educate their clients about infectious disease and the mitigation of abortions on breeding farms. Further investigations into the protective effect of EHV-1 vaccination and the underlying EHV-1 prevalence in Ontario are warranted. CVJ