J. Vet. Med.
A 51, 218–225 (2004)

2004 Blackwell Verlag, Berlin
ISSN 0931–184X
Equine Hospital, Faculty of Veterinary Medicine, University of Zurich, Zurich, Switzerland
The Association between Heart Rate, Heart Rate Variability, Endocrine and
Behavioural Pain Measures in Horses Suffering from Laminitis*
T. R. Rietmann1, M. Stauffacher2, P. Bernasconi3, J. A. Auer1 and M. A. Weishaupt1,4
Addresses of authors: 1Equine Hospital, Faculty of Veterinary Medicine, University of Zurich, Winterthurerstrasse 260, CH-
8057 Zurich; 2Physiology and Animal Husbandry, Institute of Animal Sciences, Swiss Federal Institute of Technology (ETH),
Zurich, CH-8092 Zurich; 3Via Retica 26, CH-7053 Samedan, Switzerland; 4Corresponding author: Tel.: +41-1-63-58434;
fax: +41-1-63-58905; E-mail: mweishaupt@vetclinics.unizh.ch
With 2 figures and 3 tables
Summary
The objective of this study was to compare the stress response of
horses suffering from laminitis after short- and long-term
treatment with the intent to evaluate power spectral analysis of
heart rate variability (HRV) for pain monitoring. Data were
collected from 19 horses with acute or chronic exacerbating
laminitis without known primary disease before and after
treatment with non-steroidal anti-inflammatory drugs
(NSAID). Recordings were carried out the day after admission
to the equine hospital. Measurements were repeated on day 7 of
the treatment. The recorded parameters included a clinical
orthopaedic index (OLPI: Obel-grade plus hoof tester score),
frequency of weight-shifting between contralateral limbs, mean
beat-to-beat interval (R-R) duration, standard deviation of
continuous R-R intervals, low- (LF) and high-frequency (HF)
components of HRV, sympatho-vagal balance (LF/HF), and
plasma concentration of cortisol, adrenalin and noradrenalin.
The LF represents mainly sympathetic influences on the heart
whereas HF is mediated by the parasympathetic tone.Weight-
shifting and OLPI decreased significantly with treatment. The
LF normalized units (n.u.) decreased after NSAID from
60.41 ± 21.42 to 51.12 ± 19.81 and was 49.33 ± 22.64 on
day 7, whereas HF n.u. increased from 35.07 ± 20.02 to
43.14 ± 18.30 and was 45.98 ± 23.00 on day 7. Hormone
levels showed no tendency to change with treatment. The OLPI
was only correlated with LF/HF, LF and HF (R ¼ 0.57, 0.55
and )0.54 respectively). Significant negative correlations existed
between HFn.u. and weight-shifting frequency (R ¼ )0.37),
HFn.u. and adrenalin (R ¼ )0.47), and HFn.u. and nor-
adrenalin (R ¼ 0.33). The LFn.u. only correlated positively
with adrenalin. Cortisol levels were poorly associated with the
other parameters. Determination of the sympatho-vagal influ-
ences on cardiac function may offer complementary informa-
tion for reliable assessment of pain and may represent a valuable
alternative method to catecholamine measurements.
Introduction
Horses may not have a conscious concept of what has to be
attributed to the term pain or ÔhurtingÕ. There is even still a
debate whether and to which extent animals can experience
*This report represents a part of the thesis submitted by
T. R. Rietmann to the Faculty of Veterinary Medicine, University of
Zurich, Switzerland, as required for the DrMedVet degree.
U.S. Copyright Clearance Center Code Statement: 0931–184X/2004/5105–0218 $15.00/0
Received for publication October 27, 2003
pain because it is defined as an unpleasant sensory or
emotional experience associated with actual or potential tissue
damage or described in terms of such damage (Flecknell,
2000). Pain management plays a central role within the
curative activity of a veterinarian and is an important focus
of animal welfare: (1) Eliciting pain and blocking nociception
are important diagnostic tools in equine practice; (2) evolution
of the pain intensity during treatment influences medical or
surgical therapy as it manifests the success of therapy; (3) pain
itself can have deleterious effects on convalescence; (4) today,
adequate patient care requires effective pain management and
transparency towards the owner (Flecknell, 2000; Hellebre-
kers, 2000; Taylor et al., 2002). Therefore, an attempt should
be made to elucidate the different forms of behaviour
responses to pain and to assess the correspondent physiolo-
gical variables.
A commonly addressed difficulty in the pain assessment in
animals is the lack of verbal self-report and a rather poor
sensitivity or specificity of behavioural and physiological pain
parameters (Flecknell, 2000; van Dijk et al., 2001; Taylor
et al., 2002). The behavioural pain response in horses can
resemble that caused by other sources of distress, e.g.
restlessness or unwillingness to accept riding aids. Other
responses are rather pain specific such as non-weight-bearing
lameness. Most of them are not gradually expressed but are
manifested as the pain exceeds a certain (individual) threshold.
Physiological responses to pain provide a quantitative assess-
ment and may have a better sensitivity, but lack specificity (van
Dijk et al., 2001). Most of the currently used parameters assess
the general stress response, which intends to re-establish the
body’s homeostasis (Seyle, 1936; Wall, 1992; McEwen, 1999;
Keeling and Jensen, 2002). Traditionally, the activation of two
neuroendocrine axes is studied in stress situations: the stimu-
lation of the hypothalamo-pituitary-adrenocortical axis, which
leads to an increase in cortisol production, and the activation
of the sympatho-adrenomedullar system resulting in the
increased secretion of catecholamines. However, the initiation
of the stress response can also be caused by physical activity or
psychological strain (Keeling and Jensen, 2002).
Both, the physiological stress response and the behavioural
signs of severe and acute pain are well described in horses and
include elevated heart and respiratory rates, increased body
temperature and blood pressure and typical facial expression
among many other specific behavioural signs (Dobromylskyj
et al., 2000; Schatzmann, 2000). In contrast, studies on
www.blackwell-synergy.com
Pain Monitoring with HRV
moderate and chronic pain revealed that different veterinarians
differ considerably in the judgement on the degree of pain
perceived by horses (Price et al., 2002). Additionally, it seems
that decisive placebo effects take place when animals are
evaluated after analgesic procedures (Hardie, 2000). To reduce
bias in clinical settings, the use of standardized pain rating
scales is recommended (Hardie, 2000). Furthermore, the
usefulness of new methods that specify the stress response to
pain should be investigated, as there is no single parameter
available to objectively assess pain (Taylor et al., 2002).
Heart rate variability (HRV) describes the variation between
consecutive heartbeats of an ambulatory electrocardiogram
(ECG), and is a reliable quantitative marker of autonomic
nervous activity (Task Force of the European Society of
Cardiology and the North American Society of Pacing and
Electrophysiology, 1996). It can be used as a tool for assessing
the integrity of the autonomic nervous system, the interaction
between psychological states and autonomic control, and the
pathophysiology of diseases. It has been shown that the neural
regulation of HRV, caused by sympathetic and parasympa-
thetic activity, accounts for a significant proportion of this
variability (Akselrod et al., 1981). By means of power spectral
analysis (PSA), two major peaks in the frequency content of
HRV have been identified in horses as well as in humans. A
low-frequency (LF) peak is centred on 0.1 Hz, reflecting
mainly sympathetic and to a lesser extent parasympathetic
activity, whereas high-frequency (HF) oscillations around
0.25 Hz are because of respiratory sinus arrhythmia mediated
by tonic vagal activity. Accordingly, pain as well as other
stressors would alter the power spectrum of HRV towards an
elevated sympatho-vagal balance with an increased LF and
reduced HF component. In previous studies in horses it was
shown that HRV varied significantly between basal conditions
and psychological stress (Visser et al., 2002; Voss et al., 2002;
Bachmann et al., 2003; Rietmann et al., 2004) and between rest
and moderate treadmill exercise (Physick-Sheard et al., 2000;
Voss et al., 2002). The effect of pain on HRV has been
examined in human neonates and infants, where behavioural
pain scores correlated with HRV and also blood pressure (van
Dijk et al., 2001). In the equine literature there are, to our
knowledge, no studies published that investigate the direct
influence of pain on HRV (Perkins et al., 2000).
The aim of the present study was to evaluate the method of
PSA of HRV for pain monitoring in horses and to determine
the association between sympathetically and vagally mediated
frequency components of HRV and clinical, behavioural and
physiological parameters: an orthopaedic pain score, beha-
viour during data collection, heart rate, plasma cortisol and
catecholamines. Horses with laminitis were used to examine
the decrease in pain and pain-associated stress after a single
administration of a non-steroidal anti-inflammatory drug and
after 1 week of treatment.
Materials and Methods
Horses and laminitis therapy
Nineteen horses, nine geldings and 10 mares between 3 and
28 years of age (mean ± SD: 13.6 ± 6.2 years), suffering from
laminitis were included in this study. They were all admitted to
the Equine Hospital, Faculty of Veterinary Medicine, Univer-
sity of Zurich, Switzerland. Horses in which laminitis developed
219
secondary to a primary disease (e.g. colic, retention of the
placenta, orthopaedic problem in the contralateral limb,
Cushing disease) were not considered with the intent to limit
the origin of the pain to the affected hooves. The diagnosis of
laminitis was based on a clinical examination including the
assessment of posture, gait, pulsation of the digital arteries,
superficial hoof temperature and pressure susceptibility of the
hoof sole, and the radiological findings such as rotation or
sinking of the distal phalanx (Stashak, 1987; Swanson, 1999). If
no radiological alterations were present at the time of admis-
sion, the case was judged as acute. All other cases, as they were
emergency patients, were judged as exacerbation of a chronic or
subclinical disease state (Hood, 1999). The therapy was
performed according to the current standard protocol of the
hospital, including the immediate administration of NSAID,
unfractioned Heparin (Fragmin, Dalteparinum, natricum,
Pharmacia AG, Dübendorf, Switzerland, at a dose of
50 IU/kg), support of the palmar hoof structures with thick,
soft hoof bandages and deep, soft bedding. Because controlled
administration of the anti-inflammatory agent was required in
this study, the medications were administered intravenously.
To prevent thrombophlebitis of the jugular vein, Phenylbuta-
zone was replaced by Vedaprofen (Quadrisol, Veterinaria,
Zürich, Switzerland) at a dose of 2 mg/kg body weight (bw) i.v.
b.i.d. After the first two measurements the NSAIDs were
administered orally, using either Vedaprofen (Quadrisol) at a
dose of 2 mg/kg b.i.d. or Phenylbutazone (Equipalazone Gel,
Dr E. Gräub AG, Berne, Switzerland) at a dose of 3 mg/kg bw
b.i.d. Two horses (nos 6 and 17) had a different drug regimen
because of poor response to the standard therapy [Phenyl-
butazone i.v. at a dose of 4 mg/kg bw b.i.d. and Finadyne
(Flunixin meglumine, Berna Veterinärprodukte AG, Berne,
Switzerland) i.v. at a dose of 1.1 mg/kg bw or b.i.d.].
Study design
The horses were given time to acclimatize to the novel
environment until the afternoon of the day after admission
(day 2). The experimental observation period lasted from day 2
until day 7. The horses served as their own controls. Data from
day 2 were used to monitor the pain reducing effect of a single
administration of NSAID (short-time effect of therapy). These
data were collected from 1 h before (measurement 1) until 1 h
after NSAID application (measurement 2). The trials started
always at 17:00 hours to guarantee a quiet stall environment.
Data of measurement 1 were also compared with a third
measurement taking place on day 7 at the same time of the day
(long time effect of therapy).
Clinical examination and orthopaedic laminitis pain index
The horses were clinically examined in the morning of day 2
following a standardized protocol. Rectal temperature, heart
and respiratory rates, sweating, time spent in recumbency,
degree of pulsation of the digital arteries, posture in the stall,
ability to lift the legs, gait according to the Obel-grading
system (Obel, 1948; Stashak, 1987; Owens et al., 1995; Morgan
et al., 1999; Hood et al., 2001) and lameness when turning in
small circles were recorded. At least on day 2 and day 7 the
orthopaedic examination was carried out without the hoof
bandages, which allowed the assessment of the sensitivity of
220 T. R. Rietmann et al.

Table 1. Clinical examination on day 2 and orthopaedic laminitis pain The intervals between the R-peaks of the ECG were stored
index (OLPI) continuously in milliseconds. These R-R intervals were down-
Day 2 Day 7
loaded to a portable computer (Macintosh PowerBook G3,
Apple Computer Inc., Cupertino, CA, USA) and analysed with
OBEL Hoof OBEL Hoof ProBeat (ProBeat, Bellinzona, Switzerland) and ProFit
Horse score tester OLPI score tester OLPI (Quantum Soft, Zurich, Switzerland).
Heart rate was calculated as mean duration between the
1 3 2 71 1 1 29
2 2 1 43 1 0 14 R-R peaks. As indicator of the overall parasympathetic
3 3 2 71 1 1 29 influence on the HRV, the root mean square of successive
4 2 1 43 1 1 29 R-R intervals (SDRR) was used (Task Force of the European
5 2 1 43 0 1 14 Society of Cardiology and the North American Society of
6 4 3 100 – – –
Pacing and Electrophysiology, 1996; Perkins et al., 2000).
7 1 1 29 – – –
8 2 1 43 – – –
9 3 2 71 – – –
10 2 2 57 0 0 0 Power spectral analysis (PSA)
11 2 3 57 1 2 43 The R-R interval data of 384 consecutive heartbeats were
12 1 1 29 0 1 14
13 2 1 43 1 1 29 subjected to PSA using the autroregressive model (Bernasconi
14 2 2 86 0 0 0 et al., 1998). This method was chosen instead of Fast Fourier
15 4 2 86 3 2 71 Transform because it allows the comparison of short-time
16 2 1 43 1 0 14 ECG sequences (Bernasconi et al., 1998), which is advanta-
17 4 3 100 3 3 86
geous for horses because of their low resting heart rate. The LF
18 3 2 71 0 0 0
19 3 3 86 2 2 43 and HF were calculated in normalized units (n.u.), which
allows the comparison of different measurements and subjects
Hoof tester applied in the toe region of the hoof sole: (Bernasconi et al., 1998). The ratio LF/HF represents the
Hoof tester 0: no withdrawal of the leg; hoof tester 1: withdrawal with sympatho-vagal balance. Frequency component thresholds
firm pressure; hoof tester 2: withdrawal with light pressure; hoof tester
3: withdrawal with hand-pressure.
were set at 0.01–0.15 Hz for LF and 0.15–0.5 for HF
Obel-grading system (adapted from Stashak, 1987). (Rietmann et al., 2004).
Grade 0: no gait abnormality; grade 1: incessant alternate foot lifting
at rest (weight-shifting). No lameness at walk. Short stilted gait at
walk; grade 2: walking willingly but stilted. Lameness at the trot. A Behaviour recording and analysis
foot still can be lifted off the ground; grade 3: movement is very
reluctant. Trotting impossible. Vigorous resistance for lifting up a foot; A video camera (Panasonic WV BP-500, lens Panasonic
grade 4: refusal of movement unless force is used. WV-LA 21003 2.1 mm 1:1.0, Matsushita Electric Ind. Co,
OLPI was determined as percentage of the total score (Obelmax + hoof Ltd, Japan) was installed in the box of the laminitis patients.
testermax ¼ 7 ¼ 100%). Behaviour was recorded continuously from 17:00 until
20:00 hours on SVHS video tapes with a 24 h time-lapse
the corium by use of a hoof tester. Out of these parameters video recorder (Panasonic AG 6024, Matsushita Electric Ind.
those with the most relevance to laminitic pain were chosen to Co, Ltd, Japan). Preliminary analyses of laminitic horses
set up a clinical score, the orthopaedic laminitis index (OLPI). indicated that food intake remained unaffected by the disease
The definitive OLPI included the Obel-grading score and the and that movement in the stall was highly dependent on the
hoof tester result (Table 1). Data screening revealed that the presence or absence of food. Therefore, one single reliable
other parameters did not contribute to the good differentiation parameter observed in all animals was chosen as behavioural
of the various pain degrees (e.g. digital pulsation was elevated variable: the time frequency of weight-shifting between
in all horses every day) and were therefore no longer contralateral limbs (Hood et al., 2001). The limb had to be
considered for analysis. The clinical findings were compared lifted from the ground thereby unloading the hoof shortly. The
with those made by the treating veterinarian to reduce bias. time frequency of weight-shifting was determined manually
while screening the video tapes. Intervals of analysis coincided
precisely with the ECG intervals selected for the heart rate
Heart rate and HRV recording and analysis determination and PSA of HRV.
The ECG data were collected continuously from 17:00 until
20:00 hours. The horses were loosely tied at a halter in their
stalls from 17:20 to 17:50 hours and from 18:30 to 19:00 hours Blood sampling and hormone assay
on days 2 and 7, because preliminary analyses of HRV showed On day 2, blood samples were taken 60 and 30 min before and
that feeding, free movement and increased or decreased mental at 15, 35, 60, 90 and 120 min after the administration of the
activity (e.g. increased alertness or dozing) alters HRV results. NSAID via a jugular catheter (SecalonT, Becton Dickinson,
Within these time windows, the ECG was analysed in periods Basel, Switzerland) after the first 20 ml of blood were
of approximately 10 min, during which the horses were quietly discarded. For cortisol determination, blood was collected
standing with the nucal region higher but the nostrils lower into heparinized tubes; for the catecholamines, ethylenediam-
than the withers and without lateral neck movements. Two inetetraacetic acid (EDTA) tubes (7 ml BD VacutainerTM,
electrodes were placed on wet skin on the left lateral thorax wall Becton Dickinson, Basel, Switzerland) were used. The blood
and fixed thoroughly with a stable girth. The electrodes were samples were immediately centrifuged at 4C and 3000 g
directly connected to the storage device (R-R Recorder, Polar during 15 min. Subsequently the plasma was removed, divided
Electro Oy, Kempele, Finland), which was fixed onto the girth. into 2 ml aliquots and stored until processing at )80C.
Pain Monitoring with HRV 221

Plasma cortisol was analysed by radioimmunoassay (RIA) as
described by Bachmann et al. (2003). Plasma catecholamines
(adrenalin, noradrenalin) were determined using a RIA (Tri-
CatTM, IBL Hamburg, Germany) and analyses were carried
out as described by Braun et al. (1999) and Wassell et al.
(1999). The kit had a detection limit of 10 pg/ml for adrenalin
and 20 pg/ml for noradrenalin and was applicable for all
mammalian species because of the extraction step described in
the instructions for use (http://www.ibl-hamburg.com/prod/
re_29395_m.htm#sale).
Statistical analyses
Statistical analyses were carried out with SigmaStat 2.03
(SPSS Science, Chicago, IL, USA). Descriptive statistics
(mean, SD) were computed for all parameters. Data were
tested for normality with the Kolmogorov–Smirnov test.
Differences between measurements 1 and 2, and between
measurements 1 and 3 were calculated with a paired t-test. Out
of the blood samples only sample nos 2 and 5 were compared
for differences with a paired t-test, because they were time-
matched with the other parameters. The Spearman correlation
matrix was used to compare OLPI, behaviour, physiological
and endocrine data of day 2. Differences of heart rate, HRV,
endocrine parameters and behaviour between the clinically
determined pain levels were calculated with a one-way anova.
If not indicated otherwise, significance level was set at
P < 0.05.
hoof sole. Thirteen of the 19 horses examined were dismissed
from the hospital on day 8, the day after the final
measurements. Three horses were killed and one horse was
dismissed before the end of the observations. Because of
ECG artefacts, recordings of horses 8 and 10 on day 7 had to
be discarded. Therefore, day 2 data included 19 horses and
day 7 data 13 horses.
Clinical, behavioural and physiological short- and long-term
pain reduction
The day after admission, 47% of the horses showed an Obel-
grade 3 or higher. The OLPI decreased from 62 ± 23% of the
total score on day 2 to 28 ± 25% on day 7 (mean ± SD).
There was also a significant reduction in the frequency of
weight-shifting from measure 1 to measure 2 and from
measure 1 to measure 3. The mean R-R interval over 348
consecutive heartbeats was 1286 s corresponding to a heart
rate of 47 beats/min at measure 1. Heart rate was significantly
lower at measures 2 and 3 as displayed by the larger R-R
intervals. Mean values and SD of all parameters at measure 1,
measure 2 and measure 3 are summarized in Table 2. The
increase of the R-R intervals, SDRR and HFn.u. as well as the
decrease of LFn.u. and LF/HF demonstrates the reduction of
the physiological stress response with short- and long-term
treatment although only the changes of the R-R intervals
reached significance. From horse no. 6 weight-shifting fre-
quency could not be obtained because the more affected left
hind foot was at all times off the ground.

Results
The bw of the 19 horses was 446 ± 135 kg (mean ± SD).
Seventeen horses were subjectively judged to be moderately
to highly overweigh at the time of admission. The group
included horses of different breeds (Swiss mountain horses
n ¼ 3, Icelandic horses n ¼ 3, Thoroughbreds n ¼ 3, Welsh
Ponies Typ A n ¼ 3, Ponies of unknown origin n ¼ 3,
Haflinger n ¼ 2, Warmbloods n ¼ 2). Five horses (horse nos
2, 3, 4, 8 and 10) had acute laminitis at the time of admission,
14 suffered from an exacerbation of chronic laminitis. Fifteen
cases had all 4 feet involved, although clinical and radiolo-
gical examination indicated pain predominantly in the front
hoofs. Only one horse showed laminitis symptoms mainly in
the hind limbs. One horse had additionally an abscess of the
Cortisol and catecholamines
The course of cortisol and catecholamine plasma levels after
NSAID administration on day 2 is illustrated in Fig. 1.
Medication with NSAID did not reduce any of the hormone
concentrations significantly.
Spearman correlation of the parameters
The Spearman (rank) correlation matrix was applied to the
data of measures 1 and 2 on day 2. Significant correlations
were found between HRV-parameters, behavioural, clinical
and endocrine parameters (Table 3). Cortisol correlated
poorly with other parameters.

Table 2. Mean values and standard

deviation at measure 1, 30 min Measurement 1 Measurement 2 Measurement 3
before NSAID application on day 2,
at measure 2, 1 h after NSAID R-R (ms) 1286 (344) 1425 (270)* 1434 (265)**
application on day 2 and at measure SDRR (ms) 92.34 (46.27) 103.94 (54.80) 100.88 (55.80)
3, 5 days later, 30 min before LF n.u. 60.41 (21.42) 51.12 (19.81) 49.33 (22.64)
NSAID application HF n.u. 35.07 (20.02) 43.14 (18.30) 45.98 (23.00)
LF/HF 4.37 (6.81) 2.06 (3.27) 1.82 (1.84)
OLPI 62 (23) – 28 (25)
Weight-shifting 46 (34) 34 (24)* 31 (24)**
frequency
(n per 10 min)
Cortisol (nmol) 77.25 (48.19) 65.19 (42.60) –
Adrenalin (pg/ml) 58.50 (30.68) 60.71 (23.98) –
Noradrenalin (pg/ml) 160.97 (143.19) 117.82 (102.10) –

*Significant difference between measurement 1 and measurement 2 (P < 0.05).

**Significant difference between measurement 1 and measurement 3 (P < 0.05).
222 T. R. Rietmann et al.

350 Cortisol 350 increase of weight-shifting frequency between front limbs with
Adrenalin increasing severity of the disease in chronically affected horses
Noradrenalin
300 300 and attributed this phenomenon to the digital discomfort of
bilateral laminitis. In horses with acute-induced laminitis
250 250
following carbohydrate overfeed, the same author recorded
an increase in weight-shifting frequency. Similarly, Garner
200 200
et al. (1977), Harkema et al. (1978) and Clarke et al. (1982)
nmol/l

150
100
50
0 –60 –30 15 35 60 90 120
Sampling time (min)
Fig. 1. Catecholamine (pg/ml) and cortisol concentrations (nmol) at
60 and 30 min before and at 15, 35, 60, 90 and 120 min after the
administration of the NSAID.
Differences between different degrees of pain
Based on the OLPI determined on day 2 before NSAID
administration, the 19 horses were categorized in three groups:
group 1 had an OLPI between 100 and 86% (high pain
intensity; n ¼ 5), group 2 of 71% (moderate pain intensity;
n ¼ 4) and group 3 £ 57% (low pain intensity; n ¼ 10). Mean
values and SD of weight-shifting frequency, LF n.u., HF n.u.
noradrenalin and adrenalin for each pain group are listed in
Fig. 2. Testing the three groups separately for differences in
the mean values between measurements 1 and 2 did not reveal
any significance in none of all measured parameters. Data
from measurement 3 were not analysed for differences between
pain levels because of the limited number of animals in the
respective subgroups.
Discussion
Short- and long-term pain reduction: clinical, behavioural and
physiological parameters
The clinical data of this study demonstrate that half of the
horses suffered from a considerable level of pain as indicated in
the Obel-grade of 3 the day after admission. One hour before
the analgesic treatment weight-shifting was performed
46 ± 34 times per 10 min. Hood et al. (2001) reported an
pg/ml
reported a significant stress response the first 24 h after onset
150
of Obel-grade 3 lameness. These changes were all interpreted
100
clearly as a response to pain. The HRV-parameters of
measurement 1 also referred to a stress-activated autonomous
50 nervous system when comparing those values with mean values
of 36 healthy horses measured at rest (Rietmann et al., 2004).
0 Especially the vagal tone (HFn.u. ¼ 35 ± 20 versus 50.6 ±
18.8) was lowered.
One hour after NSAID administration, the reduction of
digital pain was evident and manifested by a decrease in the
weight-shifting frequency. Also mean heart rate decreased. The
short-term pain reduction was confirmed by the increase in
vagal tone (HF) and decrease of sympathetic activity (LF).
However, because of the large SDs measured in the HRV-
parameters, differences before and after treatment did not
reach significance.
On day 7, the orthopaedic index (OLPI) and the weight-
shifting frequency decreased indicating an improvement of the
symptoms. The response of the autonomic nervous system
corresponded to the behavioural and clinical findings: HFn.u.
increased and LFn.u. decreased, indicating that the sympatho-
vagal balance returned to a state closer to that expected at rest.
Interestingly, the weight-shifting movements on day 7 were still
performed by the horses to the same extent as during
measurement 2. The same accounts for the mean R-R interval
and the spectral components of HRV. These findings imply
that digital pain and the corresponding stress response was
comparable between the recordings 1 h before the medication
on day 7 and 1 h after medication on day 2. As there are no
data available from the horses after full recovery or before
illness, it is difficult to conclude whether the similarity of the
data from measurement 2 and measurement 3 may be due to a
good response to the analgesic treatment 1 h after application
or manifest the fact that laminitis generally is not completely
cured after 7 days of treatment. At this point still all horses
were receiving analgesic treatment, but the dose was subse-
quently reduced, dependent of the individual clinical signs.

Table 3. Spearman correlation coefficients

OLPI Weight-shifting R-R SDRR LF n.u. HF n.u. LF/HF Adr NAdr Cortisol

OLPI 1.00
Weight-shifting ns 1.00
R-R ns )0.58 1.00
SDRR ns )0.39 0.39 1.00
LF n.u. 0.55 ns ns ns 1.00
HF n.u. )0.54 )0.37 ns ns )0.85 1.00
LF/HF 0.57 ns ns ns 0.94 )0.95 1.00
Adr ns 0.65 )0.44 ns 0.33 )0.47 0.42 1.00
NA ns 0.53 )0.38 ns ns )0.33 ns 0.65 1.00
Cortisol ns ns ns ns ns ns ns ns 0.50 1.00

Significant correlation coefficients. ns, not significant; Adr, adrenalin; NAdr, noradrenalin; OLPI, orthopaedic laminitis index; SDRR, successive
R-R intervals; SDRR, root mean square of successive R-R intervals; LF n.u., low frequency; HF n.u., high frequency.
Pain Monitoring with HRV 223

Group 3; low pain intensity Group 2; moderate pain intensity Group 1; high pain intensity

R–R (ms) Weight shifting (1/10 min)

1800
1600 70
60
1400
50
1200
40
1000
30
800 20
600 10
measure 1 measure 2 measure 1 measure 2

HF (n.u.) LF (n.u.)
70
90
60
80
50
70
40 60
30 50
20 40
10 30
measure 1 measure 2 measure 1 measure 2

Adrenalin (pg/ml) Noradrenalin (pg/ml)

140 500
430
120
360
100
290
80
220
60 150
Fig. 2. Mean values and standard
deviations of the three pain groups 40 80
of measurements 1 and 2. measure 1 measure 2 measure 1 measure 2

Endocrine parameters et al. (1994) discussed such a superposition of psychological

No changes were observed in the hormone levels, neither for
cortisol nor for adrenalin and noradrenalin. In other studies,
the experimental induction of laminitis led to an elevation of
cortisol and catecholamines with peak concentrations 8–16 h
after onset of Obel-grade 3 lameness (Clarke et al., 1982; Hood,
1999). In our study, time of blood sampling was beyond 16 h
after onset of lameness. However, all horses experienced still
various degrees of weight-bearing pain (OLPI ¼ 66%). It is
hypothesized that disease duration plays an important role for
the regulation of the hypothalamo-pituitary-adrenocortical
axis. A depression of the cortisol release in horses with
longstanding disease may be possible. Nonetheless, compared
with healthy individuals, the absolute concentrations measured
in this study were elevated taking into account the diurnal
variation. It is known that horses in their familiar environment
display a diurnal cortisol fluctuation with lowered concentra-
tions in the evening hours (Zolovick et al., 1966; Hoffsis et al.,
1970; Irvine and Alexander, 1994). Our values were comparable
with those obtained in the morning hours by Bachmann et al.
(2003), a study with identical blood collection and in which
plasma processing and analysing procedures were performed at
the same laboratory and by the same person. A doubt stills
remains, whether the slightly elevated concentrations of cortisol
in the present study were exclusively caused by pain, since one
day acclimation to the novel environment may be not long
enough to exclude this as source of the abolishment of the
diurnal cortisol variation (Irvine and Alexander, 1994). Ley
stress over foot pain in sheep. If this were the case in our study,
it would probably have affected also the HRV results. For this
reason, the animals were surveyed with a video camera, which
enabled the analysis of head, and neck position that occurred
simultaneously to the analysed ECG sequences, as they are
believed to correlate with the level of excitement (Rietmann
et al., 2004). This study confirms the findings of Raekallio et al.
(1997) that cortisol is not a good indicator for pain in clinical
equine studies. The catecholamine concentrations were thought
to have the same tendency to display the decrease of the stress
response after medication as the frequency components of
HRV. Hence, both are believed to react promptly to stimuli
inducing an activation of the sympathetic nervous system.
Catecholamines are circulating in the blood stream within
1 min after the triggering stimulus and have a biological half-
life < 1 min (Thews et al., 1999). Since with our data only the
noradrenalin concentrations display a tendency to decrease, it
may be concluded that it has a stronger specificity to pain than
adrenalin (Hodson et al., 1986). However, the very large SDs
interfere with any final conclusion.
Correlations
Generally, physiological responses only have a weak correla-
tion power to pain (Wall, 1992). Respiratory frequency was
judged the best for post-operative pain in dogs (Conzemius
et al., 1997). In a comparable study, heart rate showed the
224
strongest association with a correlation coefficient of 0.168
(Holton et al., 1998). Raekallio et al. (1997), investigating
post-operative pain in horses, did not find any significant
correlation between heart rate and a multifactorial pain scale
or with catecholamines and cortisol. Zierz and Wintzer (1996)
described an elevation of catecholamines with increasing
multifactorial pain scores of horses suffering from various
diseases. In human neonates HRV has been shown to correlate
with a multifactorial pain scale. In our study, the frequency
components of HRV were significantly and consistently
correlated (0.54–0.57) with changes of the OLPI. Furthermore,
HF n.u. correlated with weight-shifting frequency, adrenalin
and noradrenalin. In contrast, the LF component was only
correlated with adrenalin. This may indicate a stronger
sensitivity of the HF component for monitoring stress response
than LF, i.e. the changes of the vagal tone would represent the
initial regulatory mechanisms able to react to potential
harmful, low-intensity stimuli. In a previous study, it was
already observed that the parasympathetic part of the
autonomic nervous system had primary regulatory function
in the response to low-intensity exercise (Kuwahara et al.,
1996), whereas the LF component of HRV (the sympathetic
activity) increased only when the behavioural stress response
was expressed with substantial flight or fight reactions
(Rietmann et al., 2004). Interestingly, adrenalin was strongly
correlated with weight-shifting (R ¼ 0.65). There is a tempta-
tion to explain this finding with the more physical action
unloading the affected hoofs moving head and neck. It is
nonetheless not sure whether adrenalin is a more sensitive
parameter for physical or for psychological stress.
Differences between low-, moderate- and high-pain score (OLPI)
In all determined parameters large SDs were observed. This
variability was attributed to the individual extent of the
disease. Effects of age and breed on pain were not considered,
but nearly all influences on the measures were excluded by
careful assessment of the parameters and the time-matched
analysis of HRV, heart rate, plasma hormone levels and
weight-shifting. Moreover, the horses were restricted from
moving around and eating. Hunger, excitability or sleepiness
was prevented and the environment was undisturbed. The
investigation of differences between low-, moderate- and high-
pain scores (OLPI £ 57%, OLPI ¼ 71% and OLPI ¼ 86–
100%) did not produce significant results (Fig. 2). Despite the
catecholamine concentration being up to five times higher in
the high-pain group than in the low-pain group, the lack of
significance is not surprising as the high- and moderate-pain
group contained only five and four individuals, respectively.
Interestingly, the low- and moderate-pain group had similar
weight-shifting frequencies at measure 1, whereas at measure 2
the moderate- and high-pain group had similar values. The
physiological measures were strongly correlated with weight-
shifting. In contrast to the findings of Hood et al. (2001) our
clinical pain score correlated poorly with that disease-specific
parameter. The difference might be due to the addition of the
hoof tester results to the Obel-grade, the higher percentage of
animals with a lower pain index than in the study of Hood
et al. (2001), or to the time spent between the assessment of the
OLPI and the recordings of weight-shifting.
Among the HRV-parameters again HF was the clearest to
manifest the differences between the pain groups (Fig. 2). The
T. R. Rietmann et al.
clinical relevance of determining the spectral components of
HRV for pain assessment in horses should therefore be tested
in further studies using a larger number of individuals and a
more homogeneous patient distribution as it is not excluded
that the duration of the disease and animal breed influence
HRV. In this study there were, for example, two Icelandic
horses, aged 22 and 28 years, which were scored with a low
OLPI of 57 and 43% but had high heart rates and catechol-
amines, a very low vagal tone and high sympathetic values.
The third confounding case was a 6-year-old Warmblood
horse, with a severe non-compensated chronic laminitis on all
4 feet (OLPI 86%), which showed a very low physiological
stress response, but a high weight-shifting frequency.
Conclusions
Recognizing and classifying severity of pain always starts with
observing the animal. The same observations, however, can
result in multiple interpretations because of differences in
experience, training and ÔphilosophyÕ between veterinarians,
owners and people involved with the care of the horse. When
pain-scoring systems are applied to objectify the significance of
the observations, video recordings and profound knowledge of
the behaviour and physiology of recorded parameters are
essential. The HRV can be decisively affected by other factors
than pain, such as movement, eating and emotionality. For the
use in practice repeated measures are recommended and
factors such as excitability and unfamiliar surroundings need
to be considered. When these factors are controlled, we may
conclude that the decrease of the HF component (vagal tone)
is attributable to an increase in pain. For gaining more insides
on the autonomic response to different stressors, the non-
invasive method described in this study may be more practic-
able and less costly than stress hormone analysis. Especially
the continuous sampling of an ECG is much more advan-
tageous compared with the point sampling of blood. There-
fore, the PSA of HRV might be a helpful technique for
measuring the stress response to pain.
Acknowledgements
The authors wish to thank the Vet.-med.-Labor, Zurich,
Switzerland, namely Pascal Marti, for the analyses of the
catecholamines and Daniel Sicher from the Veterinaria AG,
Zurich, Switzerland, for placing the medicaments at our
disposal. Authors are grateful to Anthony Moses for analysing
the plasma cortisol concentrations and to Thomas Wiestner
for support during the statistical analysis. The study was
supported by a grant (973.254) from the Swiss Federal
Veterinary Office (FVO).
References
Akselrod, S., D. Gordon, F. A. Ubel, D. C. Shannon, R. D. Berger,
and R. J. Cohen, 1981: Power spectrum analysis of heart rate
fluctuations: a quantitative probe of bear-to-beat cardiovascular
control. Science 213, 220–222.
Bachmann, I., P. Bernasconi, R. Herrmann, M. A. Weishaupt, and M.
Stauffacher, 2003: Behavioural and physiological responses to an
acute stressor in crib-biting and control horses. Appl. Anim. Behav.
Sci. 82, 297–311.
Bernasconi, P., E. Messmer, A. Bernasconi, and A. Tholen, 1998:
Assessment of the sympatho-vagal interaction in central serous
Pain Monitoring with HRV
chorioretinopathy measured by power spectral analysis of heart rate
variability. Graefes Arch. Clin. Exp. Ophthalmol. 236, 571–576.
Braun, S. L., A. Barankay, and K. Löschenkohl, 1999: Determination
of Endogeneous Catecholamines by Radioimmunoassay during
Dopamine Infusion. 14th Ann. Meeting Europ. Assoc. Cardioth-
orathic Anaest.
Clarke, L. L., H. E. Garner, and D. Hatfield, 1982: Plasma volume,
electrolyte, and endocrine changes during onset of laminitis hyper-
tension in horses. Am. J. Vet. Res. 43, 1551–1555.
Conzemius, M. G., C. M. Hill, J. L. Sammarco, and S. Z. Perkowski,
1997: Correlation between subjective and objective measures used to
determine severity of postoperative pain in dogs. J. Am. Vet. Med.
Assoc. 210, 1619–1622.
van Dijk, M., J. B. de Boer, H. M. Koot, H. J. Duivenvoorden,
J. Passchier, N. Bouwmeester, and D. Tibboel, 2001: The associ-
ation between physiological and behavioral pain measures in 0- to
3-year-old infants after major surgery. J. Pain Symptom Manage.
22, 600–609.
Dobromylskyj, P., P. A. Flecknell, B. D. Lascelles, H. Livingston, P.
Taylor, and A. Waterman-Pearson, 2000: Pain assessment. In:
Flecknell, P.A. (eds), Pain Management in Animals, pp. 53–79.
W.B. Saunders, London, UK.
Flecknell, P. A. 2000: Animal pain – an introduction. In: Flecknell, P.
A. (eds), Pain Management in Animals, pp. 1–7. W.B. Saunders,
London, UK.
Garner, H. E., A. W. Hahn, C. Salem, J. R. Coffman, D. P. Hutche-
son, and J. H. Johnson, 1977: Cardiac output, left ventricular
ejection rate, plasma volume, and heart rate changes in equine
laminitis-hypertension. Am. J. Vet. Res. 38, 725–729.
Hardie, E. M., 2000: Recognition of pain behaviour in animals. In:
Hellebrekers, L. J. (eds), Animal Pain, pp. 51–70. Van der Wees,
Utrecht, NL.
Harkema, J. R., N. E. Robinson, and J. B. Scott, 1978: Cardiovas-
cular, acid-base, electrolyte, and plasma volume changes in ponies
developing alimentary laminitis. Am. J. Vet. Res. 39, 741–744.
Hellebrekers, L. J. 2000: Pathophysiology of pain in animals and its
consequence for analgesic therapy. In: Hellebrekers, L. J. (eds),
Animal Pain, pp. 71–85. Van der Wees, Utrecht, NL.
Hodson, N. P., J. A. Wright, and J. Hunt, 1986: The sympatho-
adrenal system and plasma levels of adrenocorticotropic hor-
mone, cortisol and catecholamines in equine grass sickness. Vet.
Rec. 119, 148–150.
Hoffsis, G. F., P. W. Murdick, V. L. Tharp, and K. Ault, 1970: Plasma
concentrations of cortisol and corticosterone in the normal horse.
Am. J. Vet. Res. 31, 1379–1387.
Holton, L. L., E. M. Scott, A. M. Nolan, J. Reid, and E. Welsh, 1998:
Relationship between physiological factors and clinical pain in dogs
scored using a numerical rating scale. J. Small. Anim. Pract. 39,
469–474.
Hood, D. M. 1999: Laminitis as a systemic disease. Vet. Clin. North
Am. Equine Pract. 15, 481–494.
Hood, D. M., I. P. Wagner, D. D. Taylor, G. W. Brumbaugh, and
M. K. Chaffin, 2001: Voluntary limb-load distribution in horses
with acute and chronic laminitis. Am. J. Vet. Res. 62, 1393–1398.
Irvine, C. H. G., and S. L. Alexander, 1994: Factors affecting the
circadian rhythm in plasma cortisol concentrations in the horse.
Domest. Anim. Endocrinol. 11, 227–238.
Keeling, L., and P. Jensen, 2002: Behavioural disturbances, stress and
welfare. In: Jensen, P. (eds), The Ethology of Domestic Animals: An
Introductory Text, pp. 79–98. CABT Publishing.
Kuwahara, M., S. Hashimoto, K. Ishii, Y. Yagi, T. Hada, A. Hiraga,
M. Kai, K. Kubo, H. Oki, H. Tsubone, and S. Sugano, 1996:
Assessment of autonomic nervous function by power spectral ana-
lysis of heart rate variability in the horse. J. Auton. Nerv. Syst. 60,
43–48.
Ley, S. J., A. E. Waterman, A. Livingston, and T. J. Parkinson, 1994:
Effect of chronic pain associated with lameness on plasma cortisol
concentrations in sheep: a field study. Vet. Sci. 57, 332–335.
225
McEwen, B. S. 1999: Stress and hippocampal plasticity. Annu. Rev.
Neurosci. 22, 105–122.
Morgan, S. J., D. A. Grosenbaugh, and D. M. Hood, 1999: The
pathophysiology of chronic laminitis. Pain and anatomic pathology.
Vet. Clin. North Am. Equine Pract. 15, 395–417.
Obel, N. 1948: Studies on the Histopathology of Acute Laminitis.
Almqvist and and Wiksells Boktryckteri AK, Uppsala, Sweden.
Owens, J. G., S. G. Kamerling, S. R. Stanton, and M. L. Keowen,
1995: Effects of ketoprofen and phenylbutazone on chronic hoof
pain and lameness in the horse. Equine Vet. J. 27, 296–300.
Perkins, J. D., I. M. Bowen, R. W. Else, C. M. Marr, and I. G.
Mayhew, 2000: Functional and histopathological evidence of car-
diac parasympathetic dysautonomia in equine grass sickness. Vet.
Rec. 146, 246–250.
Physick-Sheard, P. W., D. J. Marlin, R. Thornhill, and R. C. Schroter,
2000: Frequency domain analysis of heart rate variability in horses
at rest and during exercise. Equine Vet. J. 32, 253–262.
Price, J., J. M. Marques, E. M. Welsh, and N. K. Waran, 2002: Pilot
epidemiological study of attitudes towards pain in horses. Vet. Rec.
151, 570–575.
Raekallio, M., P. M. Taylor, and R. C. Bennett, 1997: Preliminary
investigations of pain and analgesia assessment in horses adminis-
tered phenylbutazone or placebo after arthroscopic surgery. Vet.
Surg. 26, 150–155.
Rietmann, T. R., A. E. A. Stuart, P. Bernasconi, M. A. Weishaupt,
M. Stauffacher, and J. A. Aver, 2004: Assessment of mental stress
in warmblood horses: heart rate variability in comparison to heart
rate and selected behavioural parameters. Appl. Anim. Behav. Sci.,
in press.
Schatzmann, U. 2000: Practical analgesic treatment in horses. In:
Hellebrekers, L. J. (eds), Animal Pain, pp. 161–179. Van der Wees,
Utrecht, NL.
Seyle, H. 1936: A syndrome produced by diverse nocuous agents.
Nature 138, 32.
Stashak, T. S. 1987: Lameness. In: Stashak, T. S. (eds), AdamsÕ Lame-
nesses in Horses, pp. 486–785. Lea and Febiger, Philadelphia, USA.
Swanson, T. D. 1999: Clinical presentation, diagnosis, and prognosis
of acute laminitis. Vet. Clin. North Am. Equine Pract. 15, 311–320.
Task Force of the European Society of Cardiology and the North
American Society of Pacing and Electrophysiology, 1996: Heart rate
variability: standards of measurement, physiological interpretation
and clinical use. Circulation 93, 1043–1065.
Taylor, P. M., P. J. Pascoe, and K. R. Mama, 2002: Diagnosing and
treating pain in the horse. Where are we today. Vet. Clin. North
Am. Equine Pract. 18, 1–19.
Thews, G., E. Mutschler, and M. A. Vaupel, 1999: Anatomie, Physi-
ologie, Pathophysiologie des Menschen. Wissenschaftliche Ver-
lagsgesellschaft mbH, Stuttgart.
Visser, E. K., C. G. van Reenen, J. T. N. van der Werf, M. B. H.
Schilder, J. H. Knaap, A. Barneveld, and H. J. Blokhuis, 2002:
Heart rate and heart rate variability during a novel object test and a
handling test in young horses. Physiol. Behav. 76, 289–296.
Voss, B., E. Mohr, and H. Krzywanek, 2002: Effects of aqua-treadmill
exercise on selected blood parameters and on heart-rate variability
of horses. J. Vet. Med. A 49, 137–143.
Wall, P. 1992: Defining Ôpain in animalsÕ. In: Short, C., and A. van
Poznak (eds), Animal Pain, pp. 63–79. Churchill Livingstone, New
York.
Wassell, J., P. Reed, J. Kane, and C. Weinkove, 1999: Freedom from
drug interference in new immunoassays for urinary catecholamines
and metanephrines. Clin. Chem. 45, 2216–2223.
Zierz, J., and H. J. Wintzer, 1996: Acute pain in the horse and one
possibility for its objective evaluation. Tierarztl. Prax. 24, 108–112
(article in German).
Zolovick, A., D. W. Upson, and B. E. Eleftheriou, 1966: Diurnal
variation in plasma gluco-corticosteroid levels in the horse (Equus
caballus). J. Endocrinol. 35, 249–253.