Journal of Equine Veterinary Science / Heart Rate Variability in Horses

Posted on June 8, 2011

This article appeared in a journal published by Elsevier. Journal of Equine Veterinary Science 31 (2011) 78-84
Original Research Heart Rate Variability in Horses Engaged in Equine-Assisted Activities

Ellen Kaye Gehrke PhD a, Ann Baldwin PhD b, Patric M. Schiltz PhD c

a Department of Community Health, National University, San Diego, CA b Department of Physiology, College of Medicine, University of Arizona, Tucson, AZ c Department of Health Sciences, National University, San Diego, CA

Journal of Equine Veterinary Science journal homepage: www.j-evs.com
aabstract

Although there has been a recent surge in using horses to treat mental and emotional human health issues, the consequences of horse-assisted interventions on the stress response of horses have not been well documented. Assessment of the autonomic nervous system and its regulation of cardiovascular function has been used as an indi- cator of acute and chronic stress in human beings and horses. Heart rate variability (HRV) is a noninvasive measurement that has been used to assess autonomic nervous system regulation of cardiovascular function. There is evidence to suggest that several factors including the genotype, behavior, environment, temperament, and nutritional status of the horse play a key role in the large inter-individual variations in basal HRV. The present study determined whether 24-hour HRV recordings in horses currently working in equine-assisted therapy (EAT) differ from those previously shown in Thoroughbred horses. Findings from the present study found that in contrast to previous studies in Thoroughbred horses, diurnal and nocturnal low frequency and high frequency powers were not signi␣cantly different in horses that are currently engaged in EAT. Future studies are needed to determine the short- and long-term consequences of horses participating in EAT programs. Findings from this study will provide the basis for the development of a physiological/behavioral assessment criteria to determine the conse- quences of EAT on the well-being of horses as well as to help EAT Centers to improve the bene␣cial effects of EAT in human beings.

! 2011 Elsevier Inc. All rights reserved.

among employees in a variety of corporations [5]. Although horse-assisted interventions have increased in popularity, there are also considerable efforts to strengthen the care and welfare of horses used in these interventions [6]. However, the consequences of horse-assisted interventions on the stress response of horses have not been well documented.

Assessment of the autonomic nervous system (ANS) and its regulation of cardiovascular function have been used as an indicator of acute and chronic stress in human beings [7] and horses [8-10]. The ANS consists of two major branches, the sympathetic system and the parasympathetic system. When changes in the normal balance of these two systems occur because of environmental challenges, the organism becomes vulnerable to pathology [11]. Heart rate variability (HRV) is a noninvasive measurement that has been used to assess ANS regulation of cardiovascular function [7,12]. HRV

Keywords:

Autonomic nervous system Diurnal rhythm Heart rate variability Equine

Horse Sympathetic/parasympathetic balance Equine-assisted therapy

1. Introduction

Animal-assisted interventions have been shown to improve cognitive, psychological, and social functioning, while reducing blood pressure, heart rate (HR), and levels of anxiety [1]. There has been a recent surge in using horses to treat mental and emotional human health issues. For instance, therapeutic horseback riding and equine-assisted therapies (EAT) have been successfully used in individuals with disorders of the central nervous system [2-4]. More- over, horses have also played a co-coaching role in leader- ship, teambuilding, and personal development programs

Correspondence author at: Ellen Kaye Gehrke, PhD, Department of Community Health, School of Health and Human Services, National University, 3678 Aero Court, San Diego, CA.

E-mail address: www.ekayegehrke@nu.edu (E.K. Gehrke).

0737-0806/$ – see front matter ! 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.jevs.2010.12.007

Author’s personal copy

E.K. Gehrke et al. / Journal of Equine Veterinary Science 31 (2011) 78-84 79

has been de␣ned as the complex beat-to-beat variation in HR produced by the interaction between sympathetic and parasympathetic (vagal) neural activity at the sinus node of the heart [10-12]. HRV analysis has also been used as a research tool in farm and companion animal species to assess changes in sympathovagal balance associated to stress, pathology, behavioral dysfunction, training regi- mens, and emotional states [9,10,13-17] and to determine how differences between sympathetic and para- sympathetic contributions to HRV relate to disease, stress, and coping strategies [10].

HRV analyses in human beings and animals are mainly based on time and frequency domain analyses [10,12,18]. Parameters of HRV in time domain analyses include inter- beats interval (IBI), which re␣ects the average time interval between the R peaks of the QRS complex (R-R interval). HRV parameters in time domain analyses also include the standard deviation of all normal to normal (SDNN) IBIs and the root mean square of successive differences between adjacent normal R-R intervals (RMSSD) [10,12]. Although SDNN has been shown to be an accurate predictor of overall variability, RMSSD is an esti- mate of high frequency beat-to-beat variations associated with parasympathetic vagal activity [10]. By contrast, Fast Fourier transformation is used to transform IBI time series data into different frequency bands [10]. Studies in human beings have shown that the power in the high frequency (HF) band (0.15-0.4 Hz) is an index of parasympathetic vagal tone [19,20]. Power in the very low frequency (VLF) band (<0.04 Hz) and low frequency (LF) band (0.04-0.15 Hz) have been associated with sympathetic and/or para- sympathetic activity [21]. However, the functional signi␣- cance of these measurements continues to be a subject of debate [20,22].

HRV has been used to assess stress reactions in horses in response to aggressive handling, inappropriate stabling, feeding, or long-distance traveling as well as stress from pain or illness [8,9]. Although previous studies have provided evidence to suggest that HRV in horses shows good stability across age and a high degree of repeatability when tested over subsequent days [9,17], large inter- individual variations in the basal values of HRV in horses have been also documented [10]. It has been proposed that several factors including the genotype, behavior, environ- ment, temperament, and nutritional status of the horse play a key role in the large inter-individual variations in basal HRV [10]. Most of the previous research on HRV in horses has been performed in young Thoroughbred horses during stall rest or during treadmill activity and no evidence of previous or current EAT experience [23-27]. Horses used for EAT are usually semi-retired working horses of a wide range of breeds and ages, and are typically standing or moving at a slow walk during their interactions with clients. Moreover, horses engaged in EAT have a diet appropriate to their nutritional needs, which is often different than high-performance diets used to feed Thor- oughbred horses. Whether these differences in the nutri- tional status, physical activities, environment, and interactions with human beings have an effect in basal HRV levels is presently unknown.

Although HRV studies in Thoroughbred horses have greatly improved our understanding of the relationship

Table 1

Description of equine subjects

EAT  equine-assisted therapy, X indicates whether the Equine is a Gelding or a Mare.

between autonomic function and stress in horses, the use of HRV as a tool to assess the stress response in horses working in EAT has not been well characterized. Improving the way to scienti␣cally assess the stress response in horses involved in EAT to assist people with physical and/or psychological conditions [28] is essential to determine how the horses are coping with their association with human beings. Because of the close interaction of horses used in EAT with human beings diagnosed with a wide variety of physical and/or psychological conditions [29], character- izing the relationship between HRV and behavior in this horse population could improve current understanding on how horses are coping during EAT.

This report is part of a larger study assessing autonomic function of horses actively working in EAT. The long-term objective of this study is to determine whether HRV could be used as a measure to evaluate the suitability of a given horseehuman pair for partnership in EAT. To accomplish this, the present study ␣rst determined whether 24-hour HRV recordings in horses currently working in EAT and co- coaching leadership training differ from those previously shown in Thoroughbred horses [16]. In this study, we report 24-hour, time and frequency domain analysis of HRV in horses with 2-3 years of experience working in EAT and/ or co-coaching leadership training. Findings from this study will provide the basis for the development of a physiological and/or behavioral assessment criteria to determine the consequences of EAT on the well-being of horses as well as to help EAT Centers to improve the bene␣cial effects of EAT in human beings.

2. Materials and Methods

2.1. Subjects

Three mares and six gelding horses (n  9) with an average age of 15.9 ` 7.7 years (mean ` standard deviation) and most weighing between 900 and 1,100 pounds, and a mule weighing 1,500 pounds participated in the present study. The research subjects were located in the San Diego, California area. Weather temperatures ranged from 45F (night) to 80F (day) during the 24-hour period of data collection from late March through early June. All horses were treated in a humane manner in accordance with the International Guiding Principles for Biomedical Research

Equine Age Gelding Mare Experience in EAT

(Years)

Mule (SeeMore) 35 X Peruvian Paso (Cassie) 20 Peruvian Paso (Vida) 21 Andalusian (Storm) 8 X Andalusian (Tessie) 9

Quarter horse (Rusty) 14 X Quarter horse (Roanie) 25 X Mustang (Tonopah) 14 X Mustang (Shiloh) 7 X

3 X 3 X 3 3 X (9 months 2

pregnant)

3 3 2 3

Author’s personal copy

80 E.K. Gehrke et al. / Journal of Equine Veterinary Science 31 (2011) 78-84

Fig. 1. Photograph showing placement of two electrodes on one side of withers and on chest.

Involving Animals. The age, gender, and breed of the horses are listed in Table 1. The HeartMath Research Institute (Boulder Creek, California) analyzed the electrocardio- graphic recordings of 12 horses. However, three horses were found to show arrhythmias and were eliminated from the study. All remaining horses (n  9) were healthy and free of arrhythmia.

2.2. Experimental Protocol

The horses were allowed to walk freely in a two-acre pasture. The researchers did not interact with the horses in any way other than by attaching the electrodes and feeding the animals during their normal morning, after- noon, and evening feeding times of 6:30-7:00 AM and 6:30-7:00 PM. Two horses at a time were attached to Holter monitors and 24-hour recordings were made. Horses were fed alfalfa in the morning, bermuda grass at noon, and a mix of bermuda and alfalfa in the evening. The researchers recorded their observations of the equine horses from a distance and ␣ndings were carefully docu- mented with entries on an almost hourly basis during the 24-hour period. The primary concern was that the electrodes stay attached and that any deviations from the horses’ “normal” day were noted. The horses were observed walking their two acre pasture, drinking quietly, conducting horse play during the day with their other herd mate, lying down or dozing, standing up, eating, and behaving during the 24-hour cycles in their normal daily patterns. Otherwise, their environment was stable and quiet during the data collection for all nine horses, as con␣rmed by the observations. The only difference in their daily patterns was that the owner did not groom, ride, or have the horses do any EAT on the data collection day. The horses were located near a secondary road and once a day a garbage truck collected waste from the health center across the road. The horses were accustomed to the noise and no corresponding spikes on the HRV recordings were noted. There were no disturbances apart from normal day- to-day interactions with their other herd mate. Nighttime observations were made on 1-2-hour intervals by walking

the outside of the pasture and noting the behavior of the horses. They were always standing quietly at rest or lying next to a herd mate standing next to them.

2.3. Recordings

Electrocardiographic recordings were done at various times between late March and early June of 2006 using 24- hour Holter monitors in nine horses. Four electrodes from the monitors were attached to each equine animal to surround the heart cavity. One electrode was placed on either side of the withers, one in the center of the chest and one at the cranial ventral abdomen. Initially, a veterinarian was on-site for the electrode placement to ensure that the major lead lines were centered along the heart meridians. The Holter monitor was attached to the horse through an adaptation of a neck and shoulder Lycra slinky sheet with a pocket attached to protect the monitor. Four small squares were shaved with horse clippers, and then a razor was used to remove any additional hair from the skin. The four electrode attachment sites were no bigger than 5 cm 5 cm. The electrode attachment sites were on both sides of the withers, the chest, and the underbelly. Iodine was used to clean the area before attaching electrodes to the subjects. In addition to conductive gel, superglue was used to attach the electrodes. The horses stood quietly relaxed during hair shaving and electrode attachment, demonstrating minimal, if any, stress related to monitor hook-up stage. The glue and the electrodes shed from their skin within a few days after the data collection was complete and the hair quickly grew back on each horse. There was no harm or discomfort noted for any of the horses during the data collection process. A photograph of electrode placement is shown in Figure 1.

Holter monitors (Del Mar Avionics) were used which incorporate a time-lock control circuit to ensure that an accurate tape speed is maintained throughout the 24-hour electrocardiogram (ECG) recording period. The 24-hour readings started at approximately 8:00 AM with readings captured until 8:00 AM the following day. Nighttime read- ings were de␣ned as those readings obtained from dark until 0800 the following day. Twenty-four-hour readings

(nighttime and daytime values) were obtained indepen- dently (ie., the 24-hour readings were not the average of the day-time and night-time values). There were slight changes in temperature or hours of daylight during the 3-month data collection phase. The researchers were careful to select days that were mild with cool evenings and their daily rituals of feeding and being with the herd were unaltered. The slinky holding the monitor could get warm and thus cool days were selected for the study. For this reason, and also because the researchers had to monitor the horses on a 24-hour obser- vation period while maintaining full-time jobs, it took almost 3 months to obtain data from nine horses.

2.4. Data Analysis

The data ␣les obtained from the 24-hour readings were sent to the HeartMath Research Institute, Boulder Creek, California, for time and frequency domain analysis. The Holter Monitoring software used (HeartMath, Impresario) is produced by Delmar-Reynolds/Spacelabs, located in Los Angeles, California. Algorithms within the software program provide interactive interpretation of waveforms for the raw ECG collected. Any abnormal beats were removed from the data by a trained human operator. Arti- facts and calibration pulses were also eliminated from the data. Then a text ␣le was generated for HRV analysis. The program used to obtain the frequency domain from the HRV data was DADiSP(r) (Newton, MA). Further details of the methods and software used for the analysis are described in McCraty and Atkinson [30]. HRV was measured using both time and frequency domain analyses for 24 continuous hours, with breakdown of the values obtained at night compared with those observed during the day.

2.4.1. Time Domain Analysis

Time domain data were averaged over nine horses for the 24-hour HR readings as a function of mean HR and mean IBI. Time domain analysis is expressed as beats per minute (bpm) or as IBIs in milliseconds (ms). Time domain data were used to evaluate HRV in terms of three variables all expressed in milliseconds: (i) the standard deviation of all the normal RR intervals in a 24-hour ECG recording (SDNN), (ii) the mean of the standard deviation of all RR intervals for each 5-minute segment of the 24-hour recording (SDNN index), and (iii) the root mean square of successive differences between normal heart beats (RMSSD). The SDNN re␣ects the ebb and ␣ow of all the factors that contribute to the HRV during a 24-hour recording, including the slow oscillations that are believed to re␣ect the intrinsic ability of the heart to respond to hormonal in␣uences [10,12,30]. Although the meaning of this measure with respect to horses has not been evaluated, in human beings, low SDNN is predictive of increased risk of sudden cardiac death and ventricular tachycardia inde- pendent of other established risk factors. The SDNN Index provides an estimate of the variability because of the factors only affecting HRV within a 5-minute epoch and is calculated by ␣rst dividing the 24-hour recording into 288 5-minute segments and calculating the standard deviation of all RR intervals during each segment. The SDNN Index is the average of these 288 numbers. The index is believed to be a measure primarily of autonomic in␣uence on HRV

[30]. The RMSSD value is obtained by ␣rst calculating each successive time difference between the QRS complexes in milliseconds. Each of the values is squared and the result is averaged before the square root of the total is obtained. The RMSSD re␣ects the short-term variance in HR and is the primary time domain measure used to estimate the HF (fast) beat-to-beat variations in HRV that provides an estimate of the parasympathetic regulation of the heart in human beings.

2.4.2. Frequency Domain Analysis

The measurement of power within each of the frequency bands is reported in absolute values of power (ms2). The frequency bands and their corresponding wavelengths selected for investigation were the same as those used for human beings [30]: ultra-low frequency (ULF)  <0.0033 Hz; VLF  0.0033-0.04 Hz; LF  0.04-0.15 Hz; and HF  0.15-0.4 Hz. In human beings, the HF range correlates to parasympathetic activity related to HR, which is largely mediated by the vagus nerve [10]. The LF range can re␣ect both parasympathetic and sympathetic activity in human beings. However, in most cases, LF provides an approxima- tion of sympathetic activity [10,30]. The VLF range is not as well de␣ned as HF and LF and while this frequency range may be in␣uenced by sympathetic activity, other factors may also modulate its power [30,31]. The ULF frequency is highly correlated to the SDNN and low values have been shown to be predictive of mortality in postmyocardial infarction patients. The 24-hour readings are necessary to capture these data [30]. Normalized HF (HF NORM) and normalized LF (LF NORM) were used to minimize the effects of the VLF changes on the HF and LF values. This type of analysis emphasizes the balance between the parasympathetic (HF) and the sympathetic (LF) branches of the ANS [12,30]. LF NORM and HF NORM components were calculated by dividing the power of the LF and HF components by the total power ([TP], from which the VLF power has been sub- tracted), and multiplying by 100. The 5-minute TP is the average of the TP (power in the band, <0.40 Hz) of each 5- minute segment, which includes the ULF, VLF, LF, and HF power spectrum bands. This measure provides an indicator of overall autonomic activity, with sympathetic activity being the primary contributing factor [10]. The 24-hour TP analysis is a spectral analysis of the entire 24-hour record as a whole (ie., it is not divided into 5-minute segments before analysis). This allows for the ULF to be assessed.

2.4.3. Statistical Analysis

All data were subjected to statistical analysis using a paired two-tailed Student’s t-test to determine whether there was a signi␣cant difference between day and night values for each parameter. A probability of less than .05 was considered signi␣cant.

3. Results

3.1. Time Domain Analysis

3.1.1. Heart Rate

Results from the present study found an overall mean HR of 43 ` 8 bpm with a signi␣cant decrease (P < .001) at

Author’s personal copy

E.K. Gehrke et al. / Journal of Equine Veterinary Science 31 (2011) 78-84 81

Author’s personal copy

82 E.K. Gehrke et al. / Journal of Equine Veterinary Science 31 (2011) 78-84

Table 2

Frequency domain analysis of heart rate values

TP  total power; HF  high frequency; LF  low frequency; VLF  very low frequency; NORM  normalized. Power values (TP) are expressed in ms2.

aP < .05 between day and night values.

night (39 ` 8 bpm) compared with day (46 ` 9 bpm). The horses assessed in the present study had an overall mean IBI of 1,484 ` 318 ms. Consistent with the HR data, night values (1,634 ` 360 ms) were found to be signi␣cantly increased (P < .0001), compared with day values (1,413 ` 313 ms).

3.1.2. Heart Rate Variability

The SDNN index during night (205 ` 84 ms) and day (207 ` 91 ms) was not signi␣cantly different. Consistent with the SDNN ␣ndings, a 24-hour SDNN index was 153 ` 57 ms with night (150 ` 58 ms) and day (155 ` 65 ms), values not being signi␣cantly different. In contrast, the horses demonstrated an average RMSSD of 123 ` 69 ms with a signi␣cantly (P < .05) lower value during daytime readings (115 ` 68 ms) compared with nighttime values (152 ` 86 ms).

3.2. Frequency Domain Analysis

Table 2 documents the values for the HF, LF, and VLF ranges measured in horses. Although there was no signif- icant difference between the nighttime and daytime HF values, the nighttime values showed a trend to be higher than those obtained during the day. There was no differ- ence between the values obtained for nighttime or daytime readings in the VLF range.

Table 2 also shows the values obtained for 5-minute TP, LF NORM, and HF NORM averaged over all nine horses for day and night periods. There was no signi␣cant difference between the 5-minute TP values during daytime compared with nighttime. However, daytime values for LF NORM were signi␣cantly greater than at night (P < .05). Moreover, daytime values for HF NORM were signi␣cantly smaller than at night (P < .05). Power values for the ULF band ranged from 11,109 to 67,269 ms2. Values for the TP obtained for the 24-hour period of recording ranged from 33,261 to 133,151 ms2 and the average value was 54,132 ` 35,874 ms2.

4. Discussion

This study reports 24-hour recordings of HRV per- formed on nine horses that were actively engaged in EAT and were allowed to move freely in turnout pasture. The high standard deviations in HF, LF, and VLF readings found

in the present study are likely because of the small sample size, different ages, breeds, and genders [10]. Horses participating in the present study were recruited from a cross-section of hot-blooded and cold-blooded breeds, as well as a blend of age and gender. This sample represents horses commonly used in EAT centers.

The results of this study showed that HRV oscillations in horses engaged in EAT occur within the same frequency ranges as for human beings. Previous studies measured HRV in horses using frequency ranges (LF: 0.01-0.07 Hz; HF: 0.07-0.5 Hz) that were slightly different from those typically used for human beings [16,23-25,27]. For instance, Kuwahara et al. [27] de␣ned LF (0.01-0.07 Hz) for sympa- thetic and/or parasympathetic activity and HF (0.07-0.6 Hz) for parasympathetic activity using intravenous atropine (0.04 mg/kg) and propranolol (0.2 mg/kg) to block auto- nomic nervous activity in horses. In those investigations, LF and HF powers obtained from measurements on horses during stall rest were quite similar to those reported in the present study. Interestingly, the HRV LF and HF frequency ranges used in the present study were based on the ranges used in human beings. These frequency ranges are slightly different from those de␣ned for horses by Kuwahara et al. [27]. The present ␣ndings are consistent with the LF and HF power values obtained by Kato et al. [25], which used the frequency ranges de␣ned by Kuwahara et al. [27]. These results have important practical implications because it supports the use of measurement equipment using human frequency ranges in horses. For instance, these ␣ndings suggest that Polar Equine monitors, which use human frequency ranges and are used to monitor horse HRV during endurance events, can be used to obtain accurate HRV data in horses. Most importantly, validation of the use of human-based HRV frequencies in horses means that direct comparisons can be made between human and horse HRV patterns, providing a quantitative method of measuring interactive responses between the two species. These ␣ndings also suggest that the contributions to TP between horses and human beings are similar.

Findings from the present study demonstrated that the power2values for the ULF band ranged from 11,109 to 67,269 ms . Moreover, values for the TP obtained for the 24-hour period of recording ranged from 33,261 to 133,151 ms2 and the average value was 54,132 ` 35,874 ms2. For human beings, HF and LF power components account for 5% of TP and VLF and ULF account for the remaining 95% [12]; thus, our ␣ndings regarding the spectral frequency distributions of HRV in horses are consistent with ␣ndings in human beings. However, the physiological correlates for the ULF contribution to HRV are unknown.

Findings from the present study reported statistically signi␣cant differences between day and night averages for both time and frequency domain parameters. In particular, LF NORM signi␣cantly decreased and HF NORM signi␣- cantly increased at night when the horses showed larger periods of sleep. Considering that horses are less active and spend more time sleeping at night than during daytime, the observed reduction in HR at night and the concomitant increase in IBI were expected. The increased RMSSD nighttime value shown by the horses in the present study was also anticipated because parasympathetic activity is expected to be increased at night when horses spend more

Frequency 24 Hour Day Night Domain Measures

HF LF VLF 5 minute TP LF NORM HF NORM

1,359 ` 1,179

4,283 ` 3,415 15,221 ` 12,661 20,864 ` 15,187 74.1 ` 11.6

25.9 ` 11.6

1,125 ` 757

4,639 ` 4,150 15,640 ` 12,774 21,495 ` 16,335

76.5 ` 10.8a 23.5 ` 10.8a

1,981 ` 2,899

3,966 ` 2,591 15,885 ` 12,141 21,833 ` 13,302 69.7 ` 14.5

30.3 ` 14.5

time sleeping, compared with daytime. Moreover, LF values showed a trend (not statistically signi␣cant) of an increase during the day, compared with nighttime values. These ␣ndings are expected because sympathetic function has been correlated with LF activity. Consistent with the present ␣ndings, results from previous studies in human beings have shown that LF activity predominates during waking hours and HF activity predominates during sleep [32]. Moreover, the LF range in human beings has been reported to be in␣uenced by sympathetic activity, whereas the HF range is an index of parasympathetic activity [12].

There is evidence to suggest that both LF and HF powers tended to be higher at night than during the day when HRV is recorded over a 24-hour period during stall rest [16]. These ␣ndings are in contrast to our results showing that the non-normalized LF and HF powers showed no signi␣- cant difference between day and night in horses that were allowed freedom of movement. It is possible that these contradicting ␣ndings are because of differences in the age and breed of the horses (2-year-old Thoroughbred horses were used in the study by Kuwahara et al. [16] or previous experiences (eg., Active physical training and nutrition of Thoroughbred horses vs. EAT). In addition, although horses in the study by Kuwahara et al. [16] were at stall rest, horses in the present study were assessed in a more natural environment. Moreover, previous studies on other species have shown that environment can in␣uence HRV diurnal rhythm. For example, when miniature swine were housed together in pairs instead of in isolation, the HRV diurnal rhythm disappeared [33]. A recent study found that when rodents were provided with enrichment items in their cages, such as tunnels, the HRV diurnal rhythm also disappeared [34]. However, whether the genetic or envi- ronmental factors play a role in these differences [10] is presently unclear. Future work is needed to determine the short- and long-term consequences of participation of EAT programs in horses. The baseline data obtained from this study are critical in understanding the signi␣cance of changes that take place in HRV dynamics by participating in EAT programs.

Studies in human beings have found that HRV analyses are useful for evaluating diseases, in particular cardiovas- cular, psychiatric, and psychological disorders [35-38]. In human beings, the dynamics of HRV and the correlation of HRV to emotional states are well documented; different types of emotions being readily distinguished by changes in heart rhythm patterns that are independent of HR [39,40]. In addition, there is theoretical and empirical evidence for the emergence of HRV as an important marker of emotional regulatory ability [41]. The use of HRV analysis in animal research to determine the role of the ANS to disease, stress, and coping strategies [10] is of great relevance for animals such as horses that participate in programs to assist people with physical and/or psychological conditions [28]. Moni- toring how horses are coping with their close association with human beings will not only bene␣t the well-being of the horses, but will provide important information that could improve the ef␣cacy of EAT programs and their bene␣t to human beings. There are already several studies that have used HRV to investigate the physiological basis for the behavior and temperament of horses [9,17,42]. Therefore, HRV seems to be an appropriate variable to be

used in future studies evaluating the suitability of a given horseehuman pair for partnership in EAT.

4.1. Limitations of the Data Collection

The HRV data may have been affected by the responses of the horses to momentary environmental incidents that could have occurred outside of the observation periods. However, the purpose of the study was not to limit the data collection only to periods of nonarousal, but to obtain a record of a normal day in the life of the horse. To some investigators, measuring HRV in animals that are not con- strained but are free to roam, poses a problem. In fact, the Task Force on HRV [12] states that in general, spectral methodology should be applied only to relatively stationary conditions. As stated by Grossman [43], one assumption of the frequency domain procedures is that the time series (variation in IBI over the recording period) has a constant mean and autovariance for the entire epoch. In practice, this assumption is violated when there is a slow drift in the signal (a change in mean), over the recording period. However, it is important to note that there is no consensus on the stationary issue. In addition, according to Grossman [43] large violations of the stationary assump- tion have little effect on most HRV estimations. In the present study, although the horses could move freely, our observations indicated that most of their activities had low effect, and that they rarely moved faster than a walk. Reitmann et al. [9] showed that frequency domain HRV parameters measured in the same horses at rest and at a forward walk showed no signi␣cant difference between the two conditions. For these reasons, it is unlikely that the 24-hour HRV recordings obtained from freely moving horses were signi␣cantly affected by the nonstationary issue. Finally, while the present study and most of the HRV studies in horses have used Holter-type recordings, it is important to point out that an increasing number of studies are also using the Polar Vantage and Polar R-R system [10].

5. Conclusion

The heart and its regulation, in terms of neural control, are similar in human beings and in horses [44]. Therefore, it was not surprising that HRV analysis of horses demon- strated measurable power values for all the frequency ranges used in the analysis of human data, including ULF, VLF, LF, and HF. This ␣nding indicates that HRV analysis of horses should be very similar to human beings, although it is not known whether the values attributed to heart-related disease in human beings are the same in horses. The values obtained from this study are baseline values for horses actively participating in an EAT program and could be used in future studies determining whether the emotional state of the horse is being in␣uenced by external stimulation such as humanehorse interactions. Most importantly, these results indicate that the types of monitors commonly used in horse facilities, which measure HRV in horses using the same frequency ranges as for human beings, are capable of providing valid data for both horses and human beings. In future work, the aim will be to use HRV to improve the consequences of EAT on the well-being of horses and to help EAT Centers to improve the bene␣cial

Author’s personal copy

E.K. Gehrke et al. / Journal of Equine Veterinary Science 31 (2011) 78-84 83

Author’s personal copy

84 E.K. Gehrke et al. / Journal of Equine Veterinary Science 31 (2011) 78-84

effects of EAT in human beings. Findings from this report are the basis of future studies determining whether HRV could be used as a measure to evaluate the suitability of a given horseehuman pair for partnership in EAT.

Acknowledgments

The authors acknowledge the Institute of HeartMath Research Division, Boulder Creek, CA, particularly Rollin McCraty, Ph.D. (Director), Mike Atkinson, PhD., and Jackie Waterman (Medical Technician), for data analysis support and consultative advice.

References

[1] Morrison M. Health bene␣ts of animal-assisted interventions. Complement Health Pract Rev 2007;12:51-62.

[2] All AC, Loving GL, Crane LL. Animals, horseback riding, and impli- cations for rehabilitation therapy. J Rehabil 1999;65:49-57.

[3] Bass MM, Duchowny CA, Llabre MM. The effect of therapeutic horseback riding on social functioning in children with autism. J Autism Dev Disord 2009;39:1261-7.

[4] Sterba JA. Does horseback riding therapy or therapist-directed hip- potherapy rehabilitate children with cerebral palsy? Dev Med Child Neurol 2007;49:68-73.

[5] Gehrke EK. Developing coherent leadership in partnership with horsesda new approach to leadership training. J Res Innovat Teach 2009;2:222-33.

[6] NARHA. Learn about equine welfare. Denver, CO: NARHA; 2010. [7] Porges SW. The polyvagal perspective. Biol Psychol 2007;74:116-43. [8] Rietmann T, Stauffacher M, Bernasconi P, Auer J, Weishaupt M. The

association between heart rate, heart rate variability, endocrine and behavioural pain measures in horses suffering from laminitis. J Vet Med A Physiol Pathol Clin Med 2004;51:218-25.

[9] Rietmann T, Stuart A, Bernasconi P, Stauffacher M. Assessment of mental stress in warmblood horses: heart rate variability in comparison to heart rate and selected behavioural parameters. Appl Anim Behav Sci 2004;88:121-36.

[10] Von Borel E, Langbein J, Despres G, Hansen S, Leterrier C, Marchant- Forde J, et al. Heart rate variability as a measure of autonomic regulation of cardiac activity for assessing stress and walfare in farm animalsda review. Physiol Behav 2007;92:293-316.

[11] Thayer J, Sternberg E. Beyond heart rate variability. Vagal regulation of allostatic systems. Ann NY Acad Sci 2006;1088:361-72.

[12] Task Force of the European Society of Cardiology and the North American Society of Pacing Electrophysiology. Heart rate variability: standards of measurement, physiological interpretation, and clinical use. Circulation 1996;93:1043-65.

[13] Pomfrett CJD, Glover DG, Bollen BG, Pollard BJ. Perturbation of heart rate variability in cattle fed BSE-infected material. Vet Rec 2004;154:687-91.

[14] Bachmann I, Bernasconi P, Hermann R, Weishaupt M, Stauffacher M. Behavioural and physiological responses to an acute stressor in cribbiting and control horses. Appl Anim Behav Sci 2003;82:297-311.

[15] Thayer JF, Hahn AW, Pearson MA, Sollers 3rd JJ, Johnson PJ, Loch WE. Heart rate variability during exercise in the horse. Biomed Sci Ins- trum 1997;34:246-51.

[16] Kuwahara M, Hiraga A, Kai M, Tsubone H, Sugano S. In␣uence of training on autonomic nervous function in horses: evaluation by power spectral analysis of heart rate variability. Equine Vet J Suppl 1999;30:178-80.

[17] Visser E, van Reenen C, van der Werf J, Schilder M, Knaap J, Barnveld A, et al. Heart rate and heart rate variability during a novel object test and a handling test in young horses. Physiol Behav 2002;76:289-96.

[18] Litvack D, Oberlander T, Carney L, Saul J. Time and frequency domain methods for heart rate variability analysis: a methodological comparison. Psychophysiology 1995;32:492-504.

[19] Akselrod S. Components of heart rate variability. In: Malik M, Camm A, editors. Heart rate variability. Armonk, NY: Futura Publishing Company; 1995. p. 147-63.

[20] Malliani A, Lombardi F, Pagani M, Cerutti S. Power spectral analysis of cardiovascular variability in patients at risk for sudden cardiac death. J Cardiovasc Electrophysiol 1994;5:274-86.

[21] Cerutti S. Spectral analysis of the heart rate variability signal. In: Malik M, Camm AJ, editors. Heart rate variability. Armonk, NY: Futura Publishing Company, Inc; 1995. p. 63-74.

[22] Houle MS, Billman GE. Low-frequency component of the heart rate spectrum: a poor marker of sympathetic activity. Am J Physiol Heart Circ Physiol 1999;276:215-23.

[23] Physick-Sheard P, Marlin D, Thornhill R, Schroter R. Frequency domain analysis of heart rate variability in horses at rest and during exercise. Equine Vet J 2000;32:253-62.

[24] Ohmura H, Hiraga A, Aida H, Kuwahara M, Tsubone H. Effects of initial handling and training on autonomic nervous function in young Thoroughbreds. Am J Vet Res 2002;63:1488-91.

[25] Kato T, Ohmura H, Hiraga A, Wada S, Kuwahara M, Tsubone H. Changes in heart rate variability in horses during immersion in warm springwater. Am J Vet Res 2003;64:1482-5.

[26] Norman S, Eager R, Waran N, Jefferey L, Schroter R, Marlin D. Recording of ECG signals on a portable MiniDisc recorder for time and frequency domain heart rate variability analysis. Physiol Behav 2005;83:729-38.

[27] Kuwahara M, Hashimoto S, Ishii K, Yagi Y, Hada T, Hiraga A, et al. Assessment of autonomic nervous function by power spectral analysis of heart rate variability in the horse. J Auton Nerv Syst 1996;60:43-8.

[28] Klontz B, Bivens S, Leinart D, Klontz T. The effectiveness of equine- assisted experiential therapy: results of an open clinical trial. Soc Anim 2007;15:257-67.

[29] NARHA. Who is served. Denver, CO: NARHA; 2010. [30] McCraty R, Atkinson M. Autonomic assessment report. A compre- hensive heart rate variability analysis. Boulder Creek, CA: HeartMath Research Center, Institute of Heart Math Research Center, Institute

of HeartMath; 1996. [31] Ori Z, Monir G, Weiss J, Sayhouni X, Singer D. Heart rate variability.

Frequency domain analysis. Cardiol Clin 1992;10:499-537. [32] Low P. Clinical autonomic disorders. 2nd ed. Philadelphia, PA:

Lippincott-Raven; 1997. [33] Kuwahara M, Tsujino Y, Tsubone H, Kumagai E, Tsutsumi H,

Tanigawa M. Effects of pair housing on diurnal rhythms of heart rate and heart rate variability in miniature swine. Exp Anim 2004;53:303-9.

[34] Brauner A, Kurjiaka D, Ibragimov A, Baldwin A. Impact of cage size and enrichment (tube and shelf) on heart rate variability in rats. Scand J Lab Anim Sci (in press).

[35] Eryonucu B, Bilge M, Guler N, Uygan I. The effect of autonomic nervous system activity on exaggerated blood pressure response to exercise: evaluation by heart rate variability. Acta Cardiol 2000;55:181-5.

[36] Karavidas M, Lehrer P, Vaschillo E, Vaschillo B, Marin H, Buyske S, et al. Preliminary results of an open label study of heart rate vari- ability biofeedback for the treatment of major depression. Appl Psychophysiol Biofeedback 2007;32:19-30.

[37] Makela P, Vahlberg T, Kantola I, Vesalainen R, Jula A. The effects of a 6-month sodium restriction on cardiac autonomic function in patients with mild to moderate essential hypertension. Am J Hypertens 2008;21:1183-7.

[38] Montano N, Porta A, Cogliati C, Constantino G, Tobaldini E, Casali K, et al. Heart rate variability explored in the frequency domain: a tool to investigate the link between heart and behavior. Neurosci Bio- behav Rev 2009;33:71-80.

[39] McCraty R, Atkinson M, Tiller W, Rein G, Watkins A. The effects of emotions on short-term power spectrum analysis of heart rate variability. Am J Cardiol 1995;76:1089-93.

[40] Lane R, McRae K, Reiman E, Chen K, Ahern G, Thayer J. Neural correlates of heart rate variability during emotion. Neuroimage 2009;44:213-22.

[41] Appelhans B, Luekens L. Heart rate variability as an index of regu- lated emotional responding. Rev Gen Psychol 2006;10:229-40. [42] Visser E, Van Reenen C, Rundgren M, Zetterqvist M, Morgan M,

Blokhuis H. Responses of horses in behavioral tests correlate with

temperament assessed by riders. Equine Vet J 2003;35:176-83. [43] Grossman P. Breathing rhythms of the heart in a world of no steady state: a comment on Weber, Molenaar, and van der Molen.

Psychophysiology 1992;29:66-72. [44] Marr C. Cardiology of the horse. Philadelphia, PA: Saunders; 1998.