Hartpury University Repository

Lewis, Victoria

Bicanardi, Thelma

Douglas, Jenni

Dumbell, Lucy

English

A Preliminary Study Investigating Functional Movement Screen Test Scores in Novice and Advanced Female Show Jumping Riders

de Haan, Donna

Female participation in the only sex-integrated summer Olympic sport

Clark, C. C. T.

Draper, Stephen

A detailed comparison of oxygen uptake kinetics at a range of exercise intensities

Tabor, Gillian

Williams, Jane

Elliott, Ami

Mann, Natasha

 1 
 
Equine Posture Analysis: Development of a simple tool to record equine thoracolumbar posture 1 
Gillian Tabora*, Ami Elliotta, Natacha Manna, Jane Williamsa 2 
 3 
aEquestrian Performance Research and Knowledge Exchange Arena, Hartpury University, Gloucester, 4 
UK, GL19 3BE *Corresponding author gillian.tabor@hartpury.ac.uk 5 
 6 
Abstract 7 
Background: Musculoskeletal assessment of horses includes assessment of conformation, symmetry 8 
and postural alignment of the equine spine however objective methods to analyse the spine have high 9 
costs and are logistically complex.   10 
Objective: This study aimed to assess the intra- and inter-rater reliability of simple methods designed 11 
to objectively measure thoracolumbar (TL) posture from photographs.   12 
Methods: A sample of horses (n=190) were photographed with a digital camera in two positions of 13 
stance and also a static image was captured from video at the toe off phase of the right hind in walk.  14 
Measurements of TL angle, depth and area of lordosis were tested for intra and inter reliability.  15 
Results: Repeated measures of TL angle showed no significant differences between measurements 16 
taken by one observer but TL depth and surface area measures were not found to be consistently 17 
reliable.  Inter-rater reliability was poor for all measurements across three observers.   18 
Conclusion: TL angle method of recording thoracolumbar posture in horses has the potential to be 19 
used to gain an objective measure of posture when standardised positioning is applied by a single 20 
experimenter or clinician.   21 
 22 
Highlights 23 
 Musculoskeletal assessment of horses includes assessment of posture 24 
 Objective, simple measures of equine thoracolumbar posture have not been tested for 25 
reliability 26 
 Calculation of equine thoracolumbar angle is reliable with one observer but not with 27 
multiple observers 28 
 Calculation of equine thorocolumbar depth or area is not reliable for either one or more 29 
observers. 30 
 The thoracolumbar angle method could be used to objectively measure outcome of 31 
treatment interventions if repeated by a single observer 32 
 33 
 34 
 35 
 2 
 
Introduction 36 
Musculoskeletal assessment of horses includes assessment of conformation, symmetry and postural 37 
alignment of the equine spine [1,2,3].  In the quadruped mammal, the thoracolumbar (TL) spine is 38 
aligned horizontally, with a position of relative extension described as lordotic and in the horse,if this 39 
lordosis is increased,  may contribute to osseous and/or soft tissue injuries [4]. Vertebral problems 40 
have been associated with changes in neck posture [5] as well as roundness of neck and back shape 41 
[6], and higher muscle activity (via surface electromyography) has been measured at the level of 42 
affected vertebrae [7].  These changes may also TL lordosis and therefore the relative TL posture.  43 
Anecdotally TL lordotic posture has been related to spinal pain and pathology in the TL regions.  44 
However objective methods to analyse the spine for pathology underlying a change in posture, such 45 
as radiography, ultrasonography or scintigraphy have high costs, are logistically complex and require 46 
extensive training for the clinician as well as presenting risks such as exposure to radiation. An 47 
objective, simple and economic measure of equine spinal posture that could be incorporated into daily 48 
practice ‘in the field’ would be beneficial, allowing assessment of posture change with following 49 
treatment and rehabilitative training.   50 
Posture of the spine is routinely assessed in human patients to examine its relationship to low back 51 
pain [8,9] and a pilot study, with a small sample size, of equine posture also assessed sagittal view 52 
photographs in neutral ‘square’ stance [1], finding good reliability of measures of thoracolumbar 53 
posture with a single observer.   However, positioning a horse in a single standardised position is 54 
challenging, time consuming and therefore for ease of practical application it would be more 55 
convenient to be able to measure TL posture in horses during relaxed stance or during gait, similar to 56 
neck posture in ridden [5] and in standing horses [7]. The reliability of such data have not been tested 57 
to date, therefore this study aimed to assess the intra- and inter-rater reliability of simple methods 58 
designed to objectively measure thoracolumbar posture from a single image image of a horse in 59 
 3 
 
neutral and relaxed standing postures, and between images of stance and a single point during the 60 
walk cycle..    61 
Methods 62 
The equine subjects were opportunistically selected from private stable yards and two equestrian 63 
colleges. Inclusion criteria were: horses aged four years or above, in ridden exercise without any 64 
current veterinary management for injury, lameness or illness. All horses wore their normal head 65 
collar or bridle for controlling movement.   66 
Two groups of horses were photographed with a Smartphone camera (8-megapixel resolution; camera 67 
height level with mid thoracic spine, aligned to the longitudinal axis and 2 metres from horse), with a 68 
visible marker of known length in the frame to allow for image calibration and accurate digital 69 
measurement for measurement of TL depth in millimetres [10] (figure 1).  Group one (n=90) horses 70 
were stood on a concrete surface two positions, in random order: feet aligned or feet non-aligned 71 
both positions with head aligned vertically, chin level with shoulder.  For collection of a dynamic image 72 
of posture group two (n=100) horses, after being photographed in stance with feet aligned, were also 73 
walked, across a flat hard surface, perpendicular to the camera at approximately five metres from the 74 
camera and a video recording (frame rate: 60 frames per second) was collected.  Videos were viewed 75 
in Windows Movie Maker (Microsoft Corporation, Redmond, WA, USA) to obtain a static image with 76 
the horse’s right hind leg in the toe off phase of gait, these were extracted for analysis [11]. 77 
Digital images were exported to ImageJ™ software (NIH, Bethesda, Maryland, USA) for analysis.  The 78 
dorsal angle of thoracolumbar spine (TLangle) (Angle ACB°), the depth of the thoracolumbar lordosis 79 
(TLdepth) (Length DC: cm) and the area of the lordosis (TLsa) (Area between line AB and curve of the 80 
dorsal profile AB: cm²) (figure 1) were measured three times with the observer blinded to previous 81 
measurements and a median calculated [1].   To test inter-rater reliability 50 images, randomly 82 
selected from the photographs of stationary group 2 horses, ,the TLangle, depth and area were 83 
measured once by three different measurers.   84 
 4 
 
 85 
 86 
 87 
Markers shown on an equine subject, stood with forefeet and hind feet aligned with equal left/right weight 88 
bearing and head aligned vertically, chin level with shoulder joint.  Symbols added to represent locations used 89 
for thoracolumbar posture measurements:  A/ = Highest point of withers; B/ = Highest point of lumbar or 90 
pelvis region; C/ = Lowest point of thoracolumbar region;  D/ = point on line AB where vertical line from C 91 
bisects. 92 
 93 
Figure 1. Marker location for measurement of sagittal alignment.   94 
 95 
A-priori sample size calculations identified the minimum sample size required to reduce the potential 96 
for type I (alpha: 5%) and type II (beta: 20%) errors [12].  A minimum of 80 horses and more than 2 97 
experimenters were identified to meet these criteria to give a valid representation of inter and intra 98 
reliability with an effect size of 0.5. 99 
Statistical Analysis 100 
Data did not meet the requirements for parametric statistical analysis, therefore non-parametric 101 
analyses were undertaken using SPSS statistical package version 23 (SPSS Inc, Chicago, IL).  Repeated 102 
measures of each image were tested with related samples Friedman’s two way analysis of variance.  103 
 5 
 
Intraclass correlation estimates (ICC) and their 95% confident intervals (CI) were calculated based on 104 
a mean-rating (k = 3), absolute-agreement, 2-way mixed-effects model [13].  To assess for differences 105 
between thoracolumbar posture in static and dynamic images, and between neutral and relaxed 106 
postures, a series of Wilcoxon signed-rank tests were used. For all tests alpha was set at 0.05. 107 
Results 108 
A total of 380 images were obtained and analysed (table 1). 109 
Table 1: Equine thoracolumbar posture measurements 110 
Median and interquartile range (IQR) measurement for each group of images for thoracolumbar angle 111 
(TLangle), depth (TLdepth) and surface area (TLsa). 112 
Measurement 
Method 
Group 1 
Group 1  
total 
sample 
size 
Group 2 
Group 2 
total 
sample 
size 
Static 
Neutral 
Dynamic  Static 
Neutral 
Static 
Relaxed 
 
TLangle (o) Median: 
150.5 
IQR = 4.4 
n=90 
Median: 
155.8 
IQR = 4.8 
n=90 
n = 180 Median: 
149.1  
IQR = 4.6 
n=100 
Median: 
148.6 
IQR = 5.4 
n=100 
n =200 
TLdepth (cm) Median: 9.1 
IQR = 1.9 
n=33 
Median: 11.1 
IQR = 3.2 
n=33 
n=66 Median: 9.8 
IQR = 1.7 
n=72 
Median: 9.6 
IQR = 1.7 
n=72 
n =144 
TLsa (cm2) Median: 
331.5 
IQR = 66.2 
n=90 
Median: 
600.8 
IQR = 254.8 
n=33 
n = 66 Median: 
462.4 
IQR = 113.2 
n=72 
Median: 
457.1 
IQR = 105.9 
n=72 
n =144 
Intra-rater reliability 113 
Intra-rater reliability was excellent between each of the three measures for TLangle with no significant 114 
differences found within experimenters (Group 1: χ2(2)=1.872, p=0.392; ICC: 0.989; CIs: 0.986-0.991; 115 
Group 2: χ2(2)=2.790 p=0.248; ICC: 0.987; CIs: 0.983-0.990) [13]. In contrast, reliability differed 116 
between the two groups assessed for TLdepth and TLsa. Measurement of TLdepth in Group 1 images 117 
was not consistent (χ2(2)=7.826 p=0.020; ICC: 0.994; CIs: 0.991-0.996) however intra-rater reliability 118 
was excellent for Group 2 (χ2(2)=3.737 p=0.154; ICC: 0.989; CIs: 0.981-0.989). For TLsa measurement, 119 
Group 1 demonstrated excellent intra-rater reliability (χ2(2)=2.279 p=0.320; ICC: 0.997; CIs: 0.996-120 
 6 
 
0.998) however this was not repeated for Group 2 data, were significant differences between 121 
measurements were found (χ2(2)=11.014 p=0.004; ICC: 0.995; CIs: 0.994-0.996). 122 
As significant differences between measurements occurred for TLdepth and TLsa, only TLangle data 123 
were analysed for differences by position. Significant differences were found for TLangle between the 124 
group 1 static and dynamic postures as well as group 2 static neutral and static relaxed images (Group 125 
1: z=8.201, p=0.000, r=.86; Group 2: z=-3.661, p=0.000, r=-.37).   126 
Inter-rater repeatability 127 
TLangle, depth and surface area measurements were all significantly different (TLangle χ2(2)=72.840 128 
p=0.000; ICC: 0.765, CIs: 0.182-0.910; TLdepth χ2(2)=34.120 p=0.000; ICC: 0.915; CIs: 0.818-0.956; TLsa 129 
χ2(2)=14.68 p=0.001, ICC: 0.933; CIs: 0.887-0.961) demonstrating poor inter-rater reliability between 130 
the measurements assessed.  131 
Discussion 132 
Three methods were used to obtain measurements of posture from sagittal images of horses standing 133 
square, in a relaxed position and images from walking gait.  TLangle method showed no significant 134 
differences between measurements taken by one observer supporting the results obtained previously 135 
[1] suggesting that TLangle method could be used to measure of thoracolumbar posture. Differences 136 
were noted between the stance positions assessed for TLangle, therefore it would be essential that 137 
the horse is positioned with the same feet and head alignment to ensure reliability when assessing 138 
posture.  However the current study was limited to only measuring each horse at one moment in time, 139 
with no repeated photographs taken to see if posture altered over time.  Tabor and Randle [1] took 140 
three images of each horse at 30-minute intervals and found no differences in repeated measures but 141 
concluded that this may be due to the absence of extraneous factors that could influence posture 142 
occurring in that time frame.   143 
 7 
 
The depth and surface area method were not found to be consistently reliable; this may be due to the 144 
increased accuracy required to measure the distances, AB and CD (figure 1) and trace the dorsal profile 145 
with the curser in Image J™.  Therefore, these methods of recording posture of the equine 146 
thoracolumbar spine should not be considered reliable.  There were also significant differences when 147 
comparing TLangle from photographs of a horse stood square and the same horse in walk. This 148 
indicates that the spinal posture was different in these two conditions, supporting the need for 149 
consistency in the position used and subsequent repeatability of positioning, for taking measures at 150 
different time periods. 151 
Inter-rater reliability was poor for all measurements across the three observers, suggesting that none 152 
of these methods are suitable for use when more than one assessor is assessing spinal posture. 153 
Conclusions 154 
Analysis of equine posture can be made by a single observer measuring the angle of the thoracolumbar 155 
spine from a single image of a horse stood in a neutral position.  However measurement of TL depth 156 
or surface area was not repeatable by a single observer and measurement of TL posture during gait 157 
did not relate to posture in stance.  None of the methods tested were repeatable with more than one 158 
observer.  The TLangle method of recording thoracolumbar posture in horses, used by the same 159 
observer, has the potential to be used to gain an objective measure of posture when standardised 160 
positioning is applied by the experimenter or clinician.  Therefore, this measure is useful and clinically 161 
applicable if the same person assesses posture throughout of treatment, for example.  The validity of 162 
this objective measure could now be tested to assess changes in TL posture and potentially used to 163 
document the outcome of intervention effects on spinal posture.  164 
This research did not receive any specific grant from funding agencies in the public, commercial, or not-165 
for-profit sectors. 166 
Ethics: Ethical approval gained from the University Centre Hartpury Ethics Committee 167 
 8 
 
Conflicts of interest: none 168 
 169 
References 170 
[1] Tabor, G.F. and Randle, H. (2013) Validation of a simple method to quantify equine thoracolumbar 171 
posture. Proceedings of the ninth International Society of Equitation Science Conference. USA 172 
[2] Van Weeren, P.R., McGowan, C., Haussler, K.K., 2010. Science overview: Development of a 173 
structural and functional understanding of the equine back. Equine Veterinary Journal, 42(s38), 174 
pp.393-400. 175 
[3] McGowan, C.M., Stubbs, N.C., Jull, G.A., 2007. Equine physiotherapy: a comparative view of the 176 
science underlying the profession. Equine Veterinary Journal, 39(1), pp.90-94. 177 
[4] Cocq, P.D., Weeren, P.V., Back, W., 2004. Effects of girth, saddle and weight on movements of the 178 
horse. Equine Veterinary Journal, 36(8), pp.758-763. 179 
[5] Lesimple, C., Fureix, C., Menguy, H. and Hausberger, M., 2010. Human direct actions may alter 180 
animal welfare, a study on horses (Equus caballus). PloS one, 5(4), p.e10257. 181 
[6] Fureix, C., Hausberger, M., Seneque, E., Morisset, S., Baylac, M., Cornette, R., Biquand, V. and 182 
Deleporte, P., 2011. Geometric morphometrics as a tool for improving the comparative study of 183 
behavioural postures. Naturwissenschaften, 98(7), p.583. 184 
[7] Lesimple, C., Fureix, C., De Margerie, E., Sénèque, E., Menguy, H. and Hausberger, M., 2012. 185 
Towards a postural indicator of back pain in horses (Equus caballus). PloS one, 7(9), p.e44604. 186 
[8] Mitchell, T., O'Sullivan, P.B., Burnett, A.F., Straker, L., Smith, A., 2008. Regional differences in 187 
lumbar spinal posture and the influence of low back pain. BMC Musculoskeletal Disorders, 9(1), p.152. 188 
[9] Smith, A., O'sullivan, P., Straker, L., 2008. Classification of sagittal thoraco-lumbo-pelvic alignment 189 
of the adolescent spine in standing and its relationship to low back pain. Spine, 33(19), pp.2101-2107. 190 
 9 
 
[10] Dyson, S.J., Tranquille, C.A., Collins, S.N., Parkin, T.D.H., Murray, R.C. 2011. An investigation of the 191 
relationships between angles and shapes of the hoof capsule and the distal phalanx. Equine Veterinary 192 
Journal. 43, 295–301. 193 
[11] Faber, M., Schamhardt, H., Weeren, R.V., Johnston, C., Roepstorff, L. and Barneveld, A.B., 2000. 194 
Basic three-dimensional kinematics of the vertebral column of horses walking on a treadmill. American 195 
Journal of Veterinary Research, 61(4), pp.399-406. 196 
[12] Jones, S.R., Carley, S. and Harrison, M., 2003. An introduction to power and sample size 197 
estimation. Emergency Medicine Journal, 20(5), pp.453-458. 198 
[13] Koo, T.K. and Li, M.Y., 2016. A guideline of selecting and reporting intraclass correlation 199 
coefficients for reliability research. Journal of Chiropractic Medicine, 15(2), pp.155-163. 200 


Equine Posture Analysis: Development of a simple tool to record equine thoracolumbar posture

Taylor, Sienna

Go Guinea Pigs! A Guide to Training Guinea Pigs

Sumner, Rebecca

Tomlinson, Matthew

Craigon, Jim

England, Gary C.W.

Lea, Richard G.

Independent and combined effects of diethylhexyl phthalate and polychlorinated biphenyl 153 on sperm quality in the human and dog

Moran, Jason

Blagrove, Richard C.

Drury, Ben

Fernandes, John F. T.

Paxton, Kevin

Chaabene, Helmi

Ramirez-Campillo, Rodrigo

Effects of Small-Sided Games vs. Conventional Endurance Training on Endurance Performance in Male Youth Soccer Players: A Meta-Analytical Comparison

Wilson, M.

Williams, Jessica

Montrose, V. Tamara

Variance in stallion semen quality among equestrian sporting disciplines and competition levels

Gaining insights into factors associated with rehoming of horses from equine charities

Birke, Lynda

Hosey, Geoff

Melfi, Vicky

“You Can’t Really Hug a Tiger”: Zookeepers and Their Bonds with Animals

Not a member yet