Cardiac arrhythmia prevalence and risk factors in 24‐h electrocardiograms of sedentary horses

Abstract

Background

Cardiac arrhythmias are common in horses, but their clinical relevance remains controversial.

Objectives

To describe prevalence and identify risk factors for arrhythmias that may warrant additional screening in a sedentary mixed‐breed population of horses.

Study Design

Prospective cross‐sectional.

Methods

Ninety‐four clinically healthy, university‐owned, sedentary horses underwent 24‐h ambulatory electrocardiograms and echocardiograms. Potential risk factors were recorded for all horses. Affected individuals were defined as those with >1 supraventricular premature complex/h or with any ventricular premature complexes for risk factor analysis. Forward stepwise logistic regression was used to identify factors associated with the presence of arrhythmia. The p value <0.15 was considered significant in univariable screening and p ≤ 0.05 was considered significant overall.

Results

During recording, 92.6% (87/94) of horses experienced an arrhythmia. Supraventricular premature complexes were present in 86.2% (81/94) of the horses, and ventricular complexes were present in 24.5% (23/94) of the horses. Of the sample, 38.3% (36/94) were considered affected by arrhythmias. Increased heart girth score was associated with detection of arrhythmias (OR 1.06, 95% CI 1–1.12, p = 0.05) and age was retained in the final model as a confounder.

Main Limitations

The study sample included a limited range of body condition scores and number of stallions for risk factor analysis.

Conclusion

Arrhythmias occurred with high frequency in this group of horses despite no known history or clinical signs of cardiovascular disease. Increased heart girth is a potential risk factor for arrhythmia in the horse.

1. INTRODUCTION

Cardiac arrhythmias in the horse have been associated with poor performance and sudden death. ¹ , ² This poses welfare, safety, and public relations concerns for horses, riders, drivers, and handlers. ³ , ⁴ Although arrhythmias are common, the practical implications on performance and safety as well as the expected frequency of arrhythmias in clinically normal horses remain poorly understood. ⁵ , ⁶ Because arrhythmias can be both intermittent and situational, they pose a diagnostic challenge. The number of arrhythmias detected is dependent upon the duration of observation and the activity performed during the observation period. A continuous 24‐h ambulatory electrocardiogram (aECG) is recommended for characterisation of intermittent arrhythmias. ¹ Previous studies of arrhythmia prevalence have reported supraventricular premature complexes at rates of 4%–85% and ventricular arrhythmias in up to 60% of horses, with variation attributable to differences in recording duration, the population evaluated, the presence of stressors such as exercise or change in management, and the definition of arrhythmias. ⁵ , ⁶ , ⁷ , ⁸ In humans, arrhythmia prevalence at rest has been reported as high as 59.4% during a 24‐h aECG, and in dogs prevalence of up to 39.5% has been reported in a hospital‐based population. ⁹ , ¹⁰ , ¹¹ This variability invokes the need to determine what is to be expected in a larger diverse equine population at rest over a 24‐h recording period. Understanding the prevalence of arrhythmias in healthy horses at rest over 24 h is essential to understanding arrhythmias that warrant further clinical investigation.

To identify horses for arrhythmia screening, a better understanding of which horses are most likely to experience arrhythmias is needed. In humans, increased age, weight, and presence of pre‐existing cardiac disease have been identified as risk factors for arrhythmias. ¹² , ¹³ Sex and increased age have been identified as risk factors for arrhythmias and sudden death in racehorses. ¹⁴ , ¹⁵ , ¹⁶ Valvular regurgitation has been described as a risk factor for the development of arrhythmias in humans. ¹⁷ , ¹⁸ In horses and humans, mitral regurgitation has been described as a risk factor for the development of atrial arrhythmias, specifically atrial fibrillation. ¹⁸ , ¹⁹ , ²⁰ , ²¹ The relationship between aortic, pulmonary, and tricuspid regurgitation and the development of arrhythmias in horses has been investigated to a lesser extent. Arrhythmia prevalence and risk factors for arrhythmias have not been described in a larger sample of sedentary horses. Thus, we set out to identify risk factors for arrhythmias in a general population of horses with the goal of defining horses that would benefit from arrhythmia screening.

The aims of this study were to (1) describe the prevalence of cardiac arrhythmias at rest in a population of sedentary horses and (2) identify risk factors for arrhythmias that may warrant additional diagnostic testing. We hypothesised that age, weight, height, heart girth, breed, body condition score (BCS), sex, cardiac troponin I concentration, and the presence of valvular regurgitation would be associated with the detection of cardiac arrhythmias in a 24‐h aECG performed at rest.

2. MATERIALS AND METHODS

2.1. Study sample

Twenty‐four‐hour aECG recordings were collected from 98 university‐owned sedentary mixed‐breed research horses housed at the UC Davis Center for Equine Health over the course of 8 months (September 2021 to April 2022). Sedentary was defined as horses that were routinely housed in turnout paddocks with variable amounts of grass and no forced exercise. This cohort of horses was selected from 100 horses in the teaching and research herd included in a larger equine precision health initiative that also underwent whole genome sequencing and detailed phenotyping. ²² Horses in the precision health initiative were enrolled 2 years before the start of enrolment in the arrhythmia study. These horses were selected from ~150 horses in the herd to mimic the horse population of the United States. Horses were considered for inclusion in the precision health initiative if they were less than or equal to 15 years of age to allow them to be followed over time. An exception was made for animals that had unique or informative phenotypes if these animals were also less than or equal to 18 years of age (Table S1). Secondary selection was based on sex with the intent to enrol males and females at a 1:1 ratio. All 98 horses from the precision health study still in the herd were considered for inclusion in the present study. Four horses with a history of cardiovascular (horse 33 for an aortocardiac fistula and horse 8 for moderate to severe aortic regurgitation) or muscular disease (horse 10 for PSSM and horse 70 for vitamin E myopathy) were excluded from this study for the purposes of arrhythmia evaluation, resulting in 94 horses available for inclusion (Figure 1).

FIGURE 1.

Flowchart of study sample selection from the teaching and research herd.

Horses were housed individually for an overnight acclimation period before the beginning of their aECG recording. Horses remained individually housed for the duration of their recording and were maintained on their routine diet of grass and/or alfalfa hay morning and evening and had water available ad libitum. Immediately before each aECG, horses underwent a physical examination and jugular vein venipuncture. Heart girth, height, weight, age, and BCS were recorded for each horse at the time of the ECG. Heart girth was measured in centimetres, placing a measuring tape across the highest part of the withers and around the chest of the horse as closely behind the elbows as possible. Height was measured via a height measuring stick (Schneiders) and weight was measured via a platform scale (Transcell). Age was determined based on university records. A BCS was determined utilising the Henneke Body Condition Scoring System. ²³ All procedures were approved by an institutional animal care and use committee at the University of California, Davis (Protocol 23250).

2.2. Ambulatory ECG recordings

A 5‐lead aECG was recorded utilising a Burdick 48 Hr H3+ Recorder with three separate channels (Hill‐Rom Holdings, Inc.). Hair was clipped in ~2″ × 2″ squares and cleaned with 70% isopropyl alcohol at electrode sites. Skintact W‐601 Solid Gel Cloth Electrodes (Leonhard Lang USA, Inc.) were placed with three on the left side of the horse and two on the right (Figure 2). All electrodes were secured with superglue (The Gorilla Glue Company). Elastikon Elastic Tape (Johnson & Johnson) was then placed around heart girth, followed by an adjustable elastic surcingle (Equi‐Essentials) and a training surcingle (Dover Saddlery) to hold the aECG monitor, leads, and electrodes in place. Horses were checked at ~6 and 12 h after the beginning of recording, with leads repositioned as needed. Vision 5 Holter Analysis Software (Hill‐Rom Holdings, Inc.) was utilised for ECG review and analysis. ECGs were reviewed by a single researcher (LM) and erroneous complex identification was corrected before secondary review of the ECGs by a second researcher who also subjectively evaluated supraventricular complexes and removed complexes flagged in areas of sinus arrhythmia or atrioventricular block not consistent with supraventricular complexes (JM). Both the researcher performing the initial review and the researcher performing the secondary review were masked to horse identification and demographics. A full disclosure review was performed by both researchers following initial complex evaluation to ensure no erroneous classifications were missed. The software parameters were set to identify supraventricular premature complexes with deviation >20% from previous R‐R intervals, atrial fibrillation, ventricular complexes, and pauses defined as 3000 ms or greater. Supraventricular premature complexes (SVPCs) were characterised as premature atrial depolarisations with >20% shorter deviation from the previous beat (Figure 3A) consistent with previous literature. ⁸ , ²⁴ , ²⁵ Overall deviation was evaluated subjectively during ECG review and periods of sinus arrhythmia of greater than 20% overall deviation in R‐R interval were removed from the SVPC category. Ventricular complexes (VCs) were defined as ventricular depolarisation characterised by the absence of detectable P‐waves and change in QRS morphology (Figure 3B). Runs of ventricular complexes were defined as more than three consecutive ventricular complexes and were described as ventricular tachycardia or slow ventricular rhythms based on the rate. HR during runs of VCs were measured manually via digital callipers (Sangabery). Total recording duration, total analysed recording duration, minimum and maximum HR in beats per minute (BPM), average HR, number of SVPCs, VCs, and pauses (second degree atrioventricular block [2DAVB] or sinus block) were recorded, as well as frequency of singlets, couplets, or runs of 3+ SVPCs and/or VCs in each recording.

FIGURE 2.

Electrode placement. Leads were connected to Skintact W‐601 Solid Gel Cloth ECG Electrodes (Leonhard Lang USA) and then placed on each horse. (A) On the left side of the horse, one electrode (LL) was placed caudal to the xiphoid process on the ventral midline. One electrode (V) was placed 50 cm down from the dorsal midline, caudal to the long head of the triceps. One electrode (LA) was placed 20 cm down from the dorsal midline, caudal to the scapular cartilage. (B) On the right side of the horse, one electrode (RA) was placed 20 cm down from the dorsal midline, caudal to the scapular cartilage. The final electrode (RL) was placed 40 cm down from the dorsal midline.

FIGURE 3.

Representative tracing of a supraventricular premature complex (SVPC) on electrocardiogram (A) and representative tracing of a run of four ventricular complexes (VC) on electrocardiogram; the first VC is a fusion beat (B). This run of ventricular complexes at a rate between 45 and 60 beats per minute is consistent with a slow ventricular rhythm. Tracings are displayed at 25 mm/s paper speed and 10 mm/mV.

2.3. Blood sample handling and analysis

Blood was collected in heparinised tubes (Becton, Dickinson and Company) and centrifuged at 3000 rpm and 24°C for 15 min. Heparinised plasma samples were aliquoted into cryovials and stored in a −80°C freezer for a maximum of 4 months until analysis. Plasma samples were subsequently shipped on dry ice to the University of Pennsylvania for analysis of Cardiac Troponin I (cTnI) concentration utilising the Stratus CS immunoassay (Stratus CS, Siemens Diagnostic Healthcare) which was previously validated for use with equine cTnI. ²⁶

2.4. Echocardiography

Unsedated transthoracic echocardiograms were performed on all horses enrolled in the study between 16 May 2020 and 13 May 2022 by two experienced clinicians (EW and JM). The hair was clipped bilaterally and prepared with alcohol and ultrasound gel. Cine‐loops and images were obtained with an ultrasound machine (LOGIQ e, Sound) equipped with a 3.5 MHz cardiac probe (3Sc‐RS Sector Transducer) and simultaneous ECG recording. Standard echocardiographic examinations involved 2‐dimensional, M‐mode, and colour Doppler imaging as previously described. ²⁷ Standard views included long axis 2‐dimensional images including views of the right ventricular outflow tract, left ventricular outflow tract, and 4‐chamber view from the right; short axis 2‐dimensional images of the left ventricle at the chordal level, mitral level, and of the aorta from the right; M‐mode images of the aortic valve, mitral valve, and left ventricle at the chordal level from the right short axis views; and long axis 2‐dimensional images of the left atrium, aortic valve, and pulmonic valve from the left. Colour Doppler imaging was performed by scanning from cranial to caudal over the valve of interest from the right side of the horse, or in the case of the mitral valve, from the left and right side of the horse. Colour map and depth were held constant throughout the study; however, gain and frequency adjustments were made on individual horses to accommodate variation in image quality and effects of variable body types. Cine‐loops were captured to optimise any visible regurgitant jets observed during the study. Mitral regurgitation (MR), aortic regurgitation (AR), pulmonary regurgitation (PR), and tricuspid regurgitation (TR) were scored from 0 to 4 by two observers (LM and JM) utilising colour Doppler cine‐loops and a previously described scoring system. ²⁸ A score of 0 signified no regurgitation present, while a score of 4 signified severe regurgitation.

2.5. Data analysis

GraphPad Prism Version 10.1.2 (GraphPad Software) and Stata 17.0 (StataCorp LLC) were used for statistical analysis. Evaluated risk factors for this study were breed, age, height, weight, heart girth, BCS, sex, cTnI concentration, and presence of valvular regurgitation. Descriptive statistics were performed on arrhythmia prevalence and each risk factor. Normality of each risk factor was assessed by a Shapiro–Wilk test. Normally distributed data are reported as mean ± standard deviation and nonparametric data as median, interquartile range (IQR), and minimum‐maximum. Three definitions of arrhythmias were initially considered for risk factor analysis: (1) The presence of >1 SVPC/h or the presence of a VC; (2) The presence of >1 SVPC/h; and (3) The presence of a VC. Due to the low numbers of horses with VCs, the first definition was selected to maximise the power to detect arrhythmia risk factors. Affected individuals were thus defined as horses with >1 SVPC/h or the presence of a VC. Any horse with a frequency of >1 SPVC/h or at least one VC present on recording was considered affected. Unaffected individuals were defined as horses with <1 SVPC/h and no VCs present in their recording. This definition is somewhat arbitrary as there is not a consensus on what constitutes a clinically relevant arrhythmia and warrants further screening. These cut offs were selected because they have been previously discussed in the literature and are conservative, erring on the side of more horses considered affected. ¹ , ²⁹ A conservative approach was elected as the purpose of the study was to identify horses for further screening. Ultimately, additional screening, including exercise testing and an echocardiogram, will be essential before providing clinical recommendations.

For analysis, data were categorised as follows: BCS was grouped as 5–6.9, 7–7.9, and ≥8. Regurgitation (mitral, aortic, tricuspid, and pulmonary) was classified as absent (score = 0) or present (score >0) based on echocardiography. Sex was classified as gelding, mare, and other (stallion n = 2; ovariectomised mare n = 5). Breed was classified as Quarter Horse (n = 29), Thoroughbred (n = 29), Warmblood (n = 20), and Other (n = 16). Forward stepwise logistic regression to identify factors associated with the presence of arrhythmia was performed as follows: individual variables were screened for inclusion and considered eligible for multivariable modelling at p < 0.15. Starting with the best‐fitting model, additional variables were added in descending order of statistical significance and retained if they significantly improved model fit (p ≤ 0.05) based on likelihood ratio testing. Interaction terms among main effects were assessed and retained if they significantly improved model fit. Age, sex, and breed were considered potential confounders. These variables were added singly to the final model and retained if they altered effect estimates by >10%.

3. RESULTS

3.1. Study sample

The 94 horses included consisted of 50 mares (5 of which were ovariectomised), 42 geldings, and 2 stallions. The sample consisted of Quarter Horses (30), Thoroughbreds (29), Warmbloods (20), Standardbreds (6), Iberian (2), Lipizzaners (2), Appaloosas (2), an Arabian (1), a Paint (1), and a horse of unknown breeding (1). Age, height, weight, heart girth, and cTNI descriptive statistics are provided in Table 1. Descriptive statistics for categorical variables (breed, sex, BCS, and valvular regurgitation) are provided in Table 2. A regularly irregular rhythm consistent with second‐degree atrioventricular block was auscultated in two horses before aECG recording. All other horses had a regular rhythm on auscultation, and no murmurs were auscultated during physical examinations.

TABLE 1.

Characteristics of the population of 94 horses included in the sedentary study sample—Continuous variables.

Variable	Minimum	Maximum	Median (IQR)	Mean ± SD
Age (years)	3	20	N/A	12 ± 4
Weight (kg)	429	776	557 (523–600)	N/A
Height (cm)	143.3	183.9	162.6 (152.9–164.1)	N/A
Heart Girth (cm)	174	212	N/A	190 ± 8
cTnI (ng/mL)	0.00	5.45	0.00 (0.00–0.01)	N/A

Abbreviation: cTnI, serum cardiac troponin I concentration.

TABLE 2.

Characteristics of the population of 94 horses included in the sedentary study sample—Categorical variables.

Variable	Category	Number of horses
Sex	Mare	54/94 (47.9%)
	Gelding	42/94 (44.7%)
	Other	7/94 (7.4%)
Breed	Quarter Horse	30/94 (31.9%)
	Thoroughbred	29/94 (30.9%)
	Warmblood	20/94 (21.3%)
	Other	15/94 (16.0%)
Body Condition Score	5–6.9	51/94 (54.3%)
	7–7.9	24/94 (25.5%)
	≥8	19/94 (20.2%)
Aortic regurgitation	No	48/92 (52.2%)
	Yes	44/92 (47.8%)
Tricuspid regurgitation	No	71/90 (78.9%)
	Yes	19/90 (21.1%)
Pulmonary regurgitation	No	64/90 (71.1%)
	Yes	26/90 (28.9%)
Mitral regurgitation	No	83/93 (89.2%)
	Yes	10/93 (10.8%)

3.2. Arrhythmia detection

A total of 94 aECGs were used for analysis; four were repeated, either due to poor‐quality recordings (2), behaviour (1), or technological difficulties (1). The recording duration ranged from 23.9 to 28.7 h with a median duration of 24.5 h (IQR 24.3–24.8). The analysed duration of each recording, after exclusion of regions with artefact that prevented accurate evaluation, ranged from 5.5 to 25.5 h, with a median duration of 24.3 h (IQR 24.1–24.7). The average heart rate throughout the recording ranged from 28 to 44 bpm, with a median average heart rate of 35 bpm (IQR 33–38). During recording, 92.6% (87/94) of horses experienced an arrhythmia (SVPC or VC). At least one SVPC was present in 86.2% (81/94) of the horses, and 19.1% (18/94) had >1 SVPC/h (Figure 4A). SVPCs ranged from 0.0 per recording to 24.6/h with a median of 0.2/h (IQR 0.04–0.5). Ventricular complexes were detected in 24.5% (23/94) of the horses (Figure 4B). VCs ranged from 0.0 per recording to 19.1/h with a median of 0.0/h (IQR 0.0–0.04). Pauses, defined as either 2DAVB or a sinus block, were observed in 88.3% (83/94) of the horses. Pauses ranged from 0.0 per recording to 340.4/h, with a median of 1.9/h (IQR 0.2–36.1) The longest pause per recording ranged from 2.0 to 6.6 s, with a mean of 4.2 ± 1.0 s. The prevalence of individuals with >1 SVPC/h or a VC was 38.3% (36/94). SVPC and VCs occurred as singlets, couplets, or runs of 3+ complexes in a row (Table S3). SVPC couplets were identified in 13.6% (11/81) of horses that experienced an SVPC in their recording, and one horse had a single run of 4 SVPCs. Of the horses that experienced them, the minimum number of SVPC couplets per recording was 1 per recording and the maximum was 9, with a median of 1 (IQR 1–7). VC couplets were identified in 12.5% (3/24) of horses that experienced VCs. Runs were identified in 25.0% (6/24) of horses that experienced VCs. Among the three horses with ventricular couplets, one horse had 16 couplets, one had 8, and one had 2 per recording. Among the six horses that experienced runs of VCs, the minimum number of runs was 1 and the maximum was 30 with a median of 2.5 (IQR 1.8–9.8). The maximum number of VCs per run was 117 complexes and the minimum was 3 complexes with a median of 7 complexes (IQR 4–57). No complex arrhythmias such as polymorphic ventricular tachycardia, supraventricular tachycardia, or atrial fibrillation were identified.

FIGURE 4.

Number of horses with a given frequency of supraventricular (A) and ventricular (B) ectopy in a continuous 24‐h ECG recording (n = 94). Dotted line in A represents the cut‐off of 1 SVPC/h that was considered of potential clinical relevance for the purpose of this analysis.

3.3. Cardiac troponin I

In our sample, 5.3% (5/94) were above the reference range for Cardiac Troponin I (cTnI) in horses (0.00–0.06 ng/mL). ²⁶ One Lipizzaner mare (cTnI of 5.45 ng/mL) had a history of familial narcolepsy and had collapse episodes the day of the ECG. ³⁰ Four horses with increased cTnI had no known history to explain the elevation, including two Warmblood mares (3.07 and 0.07 ng/mL), one Thoroughbred gelding (0.17 ng/mL), and one Standardbred mare (0.12 ng/mL). Neither of the mares with a cTnI concentration >1 ng/mL had an increased number of pathologic rhythm disturbances. The Lipizzaner mare with a cTnI of 5.45 ng/mL had no recorded arrhythmias and the Warmblood mare with a cTnI of 3.07 ng/mL had 1 SVPC recorded.

3.4. Valvular regurgitation

Echocardiography was performed between May 2020 and May 2022. The amount of time that elapsed between echocardiogram and aECG ranged from 1 day to 20 months with a median of 8.9 (IQR 4.5–15.4) months. Valvular regurgitation was identified in 69.1% (65/94) of horses. 10.8% (10/93) of horses had mitral regurgitation, 47.8% (44/92) had aortic regurgitation, 21.1% (19/90) had tricuspid regurgitation, and 28.9% (26/90) had pulmonic regurgitation. Of the 47.8% of horses that had aortic regurgitation, 18.2% (8/44) of those horses had a score of 2 or higher. Aortic regurgitation was the only type of regurgitation that received a score of a 3 (n = 1). All other types of regurgitation received a score of 2 or below. No scores of 4 were recorded for any type of regurgitation. Horses with inadequate Cine‐loop quality to identify regurgitation were excluded from this analysis of individual valves as indicated.

3.5. Arrhythmia risk factors

Affected horses (n = 36) were defined as having the presence of >1 SVPC/h or the presence of a VC in their recording for the purpose of risk factor evaluation. Results of univariable logistic regression analysis are shown in Table S4. Variables eligible for inclusion in a multivariable model included heart girth and BCS. Starting with a univariable model that included heart girth, no additional variables significantly improved model fit. Age qualified as a confounder. The final logistic regression model is shown in Table 3.

TABLE 3.

Final multivariable logistic regression model of factors associated with arrhythmia in 94 sedentary horses from a single premise in California.

Variable	Coefficient	Odds ratio	95% CI	p value
Intercept	−2.22	0.11	0.02–0.57	<0.01
Heart girth (cm)	0.06	1.06	1.00–1.12	0.05
Age (y)	0.09	1.09	0.96–1.23	0.17

Note: The intercept represents the baseline odds of arrhythmia in 3‐year‐old horses with heart girth of 174 cm.

4. DISCUSSION

Cardiac arrhythmias occur at a high prevalence in this sedentary mixed breed equine population without a history of cardiac disease. During recording, 92.6% of horses experienced at least one arrhythmia: 86.2% of them had at least one supraventricular premature complex and 24.5% of them had at least one ventricular complex. The prevalence of individuals in the sample with >1 SVPC/h or a VC, which we considered of potential clinical relevance, was 36/94 (38.3%). This is the first study evaluating 24‐h aECG findings in a sedentary mixed breed population of this size. Previous reports cite infrequent detection of arrhythmias at rest and have evaluated small uniform populations of horses. ⁷ , ²⁸ , ²⁹ , ³¹ Arrhythmia prevalence reported during, and immediately post exercise, ranges from 27.8% to 92.0% and is generally higher than that reported at rest. These studies have focused on high‐level performance horses including racehorses and other athletes such as 3‐day eventers, dressage horses, and jumpers rather than the more general sample population studied here. ⁵ , ²⁴ , ²⁵ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ Risk factor analysis identified increased heart girth as a potential risk factor for increased arrhythmia prevalence and increased age as a confounder (Figure 5). These data fill a literature gap by describing arrhythmia prevalence in a large, mixed breed, sedentary herd of horses over 24 h and exploring potential risk factors in a herd of horses not trained for any specific purpose.

FIGURE 5.

Box and whisker plot of heart girth. Minimum, maximum, and median in unaffected horses (<1 SVPC/h and no VC present on recording) compared with affected horses (>1 SVPC/h or VC present on recording). Lower limit and upper limit of box represent 25th and 75th percentile, respectively. *p = 0.06 in univariable logistic regression and 0.05 in the final multivariable model with age included as a confounder.

The prevalence of SVPCs (86.2%) in this sample was higher than previously reported in sedentary Standardbreds (15.2%) or Thoroughbreds (0%), despite those breeds being represented in our population. ³⁷ Higher SVPC prevalence comparable to what was observed here (70%–86.4%) has been reported in athletic populations of 10–22 horses at rest and in a hospital‐based population after recovering from anaesthesia (85%). ²⁹ , ³⁸ Lower prevalence of SVPCs has also been reported in athletic populations including dressage horses (5%) and showjumpers (32%) at rest. ²⁴ , ²⁵ However, these are based on very brief resting ECGs of 2–5 min in duration compared with the 24‐h recording period used in those reported above, which likely explains the decrease in arrhythmia detection. The prevalence of horses with >1 SVPC/h in 24‐h recordings (5%–18%) is similar to what we report (19.1%). ²⁹ , ³⁸ Prevalence of VCs in this sample (24.5%) is increased compared with previous reports in a sedentary population at rest (0.0%–3.5%). ⁷ , ³⁷ , ³⁹ However, the prevalence of VCs is decreased compared with reports in both athletic (50%) and hospital‐based (60%) populations. ⁸ , ²⁹ A benefit of the 24‐h aECG monitor is to allow for an increased likelihood of detecting intermittent arrhythmias due to the increased recording time and periods of increased vagal tone during overnight recordings. It is possible that our reported prevalence is high due to the subjective nature of evaluating overall beat‐to‐beat variation. Increased information on prevalence of arrhythmias in 24‐h recordings will aid in interpretation of the clinical relevance of intermittent cardiac arrhythmia findings.

Increased heart girth was identified as a potential risk factor for cardiac arrhythmias in this study such that for every 1 cm increase in heart girth, a horse had a 6% increase in the odds of having cardiac arrhythmias when adjusted for age. BCS was also identified as a possible risk factor on univariable screening but did not improve the model fit compared with heart girth alone. Both of these values are measures of overall barrel size and could be associated with increased body condition. Metabolic syndrome has been associated with cardiovascular disease and the development of arrhythmias in humans. ⁴⁰ , ⁴¹ Obesity and insulin resistance have been associated with changes in cardiac structure and function in horses and dogs. ⁴⁰ , ⁴¹ However, the single arrhythmia study in Equine Metabolic Syndrome (EMS) did not find an increase in arrhythmias in EMS cases compared with breed‐matched controls. ⁴¹ Due to small sample size, this finding should be interpreted with caution. No other risk factors were identified consistent with several previous reports in athletic horses at rest. ²⁴ , ²⁵ , ³² Age was identified as a confounder in this study and, although not a statistically significant risk factor on its own, improved the overall fit of the model, suggesting that increased age increases the odds of having cardiac arrhythmias. Some studies have identified age and sex as potential risk factors for sudden death or post‐race arrhythmias in racehorses. ¹⁴ , ¹⁵ However, it is possible that fitness level has a relationship with these risk factors, as age has been detected as a risk factor in racehorses but not in other populations. The population studied here is much older than that associated with racehorses and therefore age likely has different ramifications. Fitness has also been associated with an increase in SVPCs, with eventing and endurance horses having more SVPCs than pleasure horse controls. ²⁹ It is likely that fitness and obesity and/or metabolic derangements represent separate physiologic stresses that increase the risk of arrhythmogenesis, as observed in humans. ¹³ Our findings combined with the existing literature suggest that increased heart girth, which can be associated with metabolic derangements, may put a horse at risk for the development of arrhythmias and warrants additional investigation.

This study had a few limitations related to the subjective nature of data classification in this field. The use of a subjective evaluation component in the secondary review of aECGs may have affected the arrhythmia detection rate in our sample compared with a strict cut‐off. However, we believe this reflects the methods used in clinical practice and accounts for horse‐to‐horse variability in rhythm. The two horses with increased cTnI may have had transient myocardial damage that was not associated with an arrhythmia. Alternatively, serial testing and evaluation of potentially cross‐reactive enzymes, although not possible in this case, would provide insight into potential laboratory error associated with these results. Valvular regurgitation scores may have varied during the elapsed time between aECG and echocardiography recordings during the study. Furthermore, horses with pre‐existing moderate or severe valvular regurgitation and consistent murmurs were intentionally excluded from this study, thus our findings should not be extrapolated to a clinically affected population with evidence of cardiac disease. Whilst the relationship between the presence of arrhythmias and heart girth warrants additional investigation, our sample only had BCSs ranging from 5 to 9, not giving a true representation of BCS grades any lower than what is considered to be ‘ideal’ (5). While the horses in this study were largely overconditioned, an increase in heart girth can also represent a variation in conformation, and future studies including metabolic testing are needed to clarify these relationships. In humans, sex is associated with cardiac disease, with men being at higher risk of developing cardiac arrhythmias and experiencing life‐threatening cardiac events. ⁴² , ⁴³ , ⁴⁴ In Thoroughbred racehorses, one recent study found that intact colts were at lower risk of experiencing sudden cardiac death when compared with mares and geldings. ¹⁴ , ¹⁵ As there were only two stallions in our study sample, our sex analysis was limited to geldings and mares, so we were unable to provide insight into this discrepancy. Investigation into the prevalence of arrhythmias in horses with lower BCSs as well as in stallions would provide additional insight into what may put horses at risk for the development of arrhythmias.

In conclusion, heart girth was found to be a potential risk factor for a higher prevalence of arrhythmias over the course of a 24‐h recording. Arrhythmias occurred at a high prevalence in this sample despite a lack of clinical signs. While additional research is needed to determine what puts a horse at increased risk of developing arrhythmias and at what point these arrhythmias are clinically relevant, these data provide valuable insight to what prevalence of arrhythmias in a population may be considered normal when performing 24‐h ECG monitoring in sedentary horses.

FUNDING INFORMATION

This project was supported by the Center for Equine Health with funds provided by the State of California satellite wagering fund and contributions by private donors and by Morris Animal Foundation grant number (D21EQ‐050).

CONFLICT OF INTEREST STATEMENT

The authors have declared no conflicting interests.

AUTHOR CONTRIBUTIONS

Lauren T. Maas: Writing – original draft; investigation; methodology; writing – review and editing; formal analysis; data curation; conceptualization; funding acquisition. Elizabeth Williams Louie: Investigation; writing – review and editing. Carrie J. Finno: Funding acquisition; writing – review and editing; conceptualization; resources. Callum G. Donnelly: Conceptualization; writing – review and editing. Joshua A. Stern: Resources; writing – review and editing; conceptualization; funding acquisition; methodology. Ashley E. Hill: Conceptualization; funding acquisition; writing – review and editing; formal analysis; methodology. Jessica M. Morgan: Conceptualization; investigation; funding acquisition; writing – review and editing; methodology; project administration; supervision; formal analysis; resources.

DATA INTEGRITY STATEMENT

Lauren T. Maas and Jessica M. Morgan had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of data analysis.

ETHICAL ANIMAL RESEARCH

The animal study protocol was approved by an Institutional Animal Care and Use Committee (Protocol 23250).

INFORMED CONSENT

Not applicable.

ANTIMICROBIAL STEWARDSHIP POLICY

Not applicable.

Supporting information

Table S1. Horses in population selected for a distinct phenotype. Excluded from risk factor analysis.

Table S2. Supraventricular premature complex (SVPC) profile for horses that experienced any SVPCs during their recording. Horses in bold were excluded from risk factor analysis due to history of neuromuscular disease.

Table S3. Ventricular complex (VC) profile for horses that experienced any VCs during their 24‐h continuous ambulatory ECG. Excluded from risk factor analysis.

Table S4. Univariable analysis of factors potentially associated with arrhythmia in 94 sedentary horses from single premises in California assessed in 2023.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are openly available in DRYAD at https://datadryad.org/dataset/doi:10.5061/dryad.0gb5mkmd9.