Variation in frontal plane joint angles in horses




Reasons for performing study: Conformation in horses is often considered an indicator of athletic ability, performance and resistance to orthopaedic disease. Evaluation is performed in the standing horse and repeatability influenced by stance. Ground reaction forces increase in the moving horse as speed increases.

Objectives: To determine the effect of locomotion on equine carpus, tarsus, metacarpophalangeal (MCP) and metatarsophalangeal (MTP) joints in the frontal plane.

Hypothesis: Valgus/varus angulation will change when moving.

Materials and methods: Kinematic data were collected standing, walking and trotting for 2 groups of horses. The change in angle for carpus, tarsus, MCP and MTP joints was calculated standing and midstance for each stride. Comparison of joint angles between left and right limbs, standing, walking and trotting were made. Inter- and intrahorse variations were investigated.

Results: Significant differences were observed between groups of horses and left and right forelimbs. Between walk and trot, the MCP joint changed from valgus to varus, and the tarsus and MTP joints increased in valgus deformity. Between standing and walk the carpus increased in valgus deformity. Interhorse variation was significantly different, intrahorse variation was not. Variation in measurements between gaits was minimal.

Conclusions: This study validates the measurement of joint angles from the front in the walking and trotting horse using kinematic data, interhorse variation in joint angle measurements exceeding intrahorse variation. The increases in joint angles between standing and walking, and walking and trotting warrant further investigation.


It is widely accepted that a substantial proportion of wastage within the equestrian population can be attributed to lameness (Jeffcott et al. 1982; Rossdale et al. 1985; Dyson et al. 2008); and this has wide implications both economically and ethically. Joint disease, particularly with regard to the metacarpophalangeal and carpal joints (Bailey et al. 1999), has been implicated as one of the greatest causes of lameness in racehorses (Wilsher et al. 2006). The industry has anecdotally made associations between conformation and musculoskeletal injury, although few studies have investigated this (Anderson et al. 2004; Gnagey et al. 2006; van Weeren and Crevier-Denoix 2006; Weller et al. 2006a). It is therefore important to understand any factor that may predispose and/or result in joint injury and wastage, including conformation.

Conformation in the horse is often used as an indicator of an individual's athletic ability, future performance and their resistance to orthopaedic disease (Holmström 2001). The term conformation refers to the structural arrangement of body segments relative to each other. This includes symmetry, segment lengths and shapes, joint angles and the deviations of these segments. Conformation may also refer to the aesthetic appearance of an animal according to the observer's perception (Blood and Studdert 2000). Certain conformational characteristics may be considered beneficial or detrimental, depending on the discipline and/or breed (Marks 2000).

The term angular limb deformity (ALD) is synonymous with a deviation in the limb, when viewing the horse in the frontal plane, or from the front (Mitten and Bertone 1994). Several deviations within the limbs in the frontal plane are observed, including valgus and varus deviations of the carpus, metacarpophalangeal (MCP) joint, tarsus and metatarsophalangeal (MTP) joint. A valgus deformity refers to a deviation of the limb laterally about a reference point, or the joint; and a varus deformity pertains to a medial deviation of the limb. Valgus and varus ALDs may occur in isolation or in combination with compensatory deformities, for example, MCP joint varus could be offset by a carpal valgus (Auer and Stick 2006).

Studies have previously investigated the changes in conformational characteristics associated with growth (Anderson and McIlwraith 2004; Santschi et al. 2006) and with increasing risk or incidence of disease. In man, knee malalignment is a risk factor for the progression of knee osteoarthritis (Cerejo et al. 2002), and increases in knee valgus of 1° have been associated with the risk of lateral compartment cartilage defects in subjects with radiographic changes of osteoarthritis (Janakiramanan et al. 2008). In a National Hunt Thoroughbred population, carpal valgus was associated with superficial digital flexor tendonitis and tarsal valgus with pelvic fractures (Weller et al. 2006b). However, in a different Thoroughbred population, it was demonstrated that a degree of carpal valgus was protective against carpal effusions and carpal fracture injuries (Anderson et al. 2004). The differing results are at least partly due to the different techniques used to measure conformation in these studies; it has been shown that measuring conformation reliably is a technical challenge (Weller et al. 2006a).

Early conformation assessment was performed with the ‘naked eye’ and based on subjective evaluation with the horse positioned standing so that all 4 feet were ‘square’. Quantitative methods of conformation and gait assessment have since been developed, that allow evidence-based assessment of conformation. Initially, these involved tape measures and goniometers, then photography (Kronacher and Ogrizek 1931) and videography (Hunt et al. 1999), later superseded by motion analysis systems (Weller et al. 2006c). The use of motion analysis over photography and videography resolves geometric errors associated with the acquisition of 2-dimensional (2D) images from 3D structures, and is independent of camera position relative to the horse (Weller et al. 2006c). Conformation measurements vary with the stance of the horse (Holmström et al. 1990; Weller et al. 2006c) and it has been demonstrated that positioning of horses during standing is one of the sources of error in determining a horse's conformational parameters (Holmström et al. 1990; Weller et al. 2006c). The use of a 3D method for measurement of conformational parameters has previously demonstrated that the variability in data obtained was smaller standing than walking (Pourcelot et al. 2002) and this method has been applied to the measurement of international level sports horses (Crevier-Denoix et al. 2004).

Ground reaction force (GRF) describes the force experienced by the limb, as a result of supporting the bodyweight against gravity. In the standing horse, the limbs act to support the body and maintain stability. In contrast, during locomotion the limbs must support the body during the shorter time period of stance, in addition to withstanding the associated positive and negative horizontal accelerations. As speed increases, stance period will decrease and an increase in the peak GRF experienced by the limb ensues (Biewener 2003). Greater peak GRFs result from faster locomotion and the cyclic loading experienced by the musculoskeletal system over time has important roles in bone remodelling and adaptations of the system with regards to injury (Pool and Meagher 1990).

Injuries occur as a result of the cyclic high forces experienced during locomotion. It is, therefore, important to assess precursors to injury, such as conformation, under the conditions at which they are sustained. Currently, measurements of conformation are made in the standing horse. It is, therefore, the purpose of this study to investigate the effects of locomotion on the frontal angle of joints in the distal limbs.

The aim of this study was to determine the effect of locomotion on the carpus, metacarpophalangeal (MCP), tarsus and metatarsophalangeal (MTP) joint angles in the frontal plane in the horse. Investigations compared joint angles in the frontal plane during standing, walking and trotting in horses. The variability of measurements obtained for repeat stances (standing) or strides (moving), between individual horses (interhorse) and within horses (intrahorse) was explored.

We hypothesised that: 1) the interhorse variation is greater than the intrahorse variation between strides; 2) the inter-stride variation within a given gait is smaller than the inter-stance variation in the repositioned standing horse; and 3) joint valgus or varus angulation will become more pronounced in the moving horse compared to the standing horse; joint angulation will also increase from walk to trot.

Materials and methods

Two different groups of horses were used in this study. Group 1 consisted of 6 Royal Veterinary College owned pony mares, aged 3–11 years; and Group 2 contained 29 Thoroughbreds (TBs) aged 2–3 years in flat race training.

Kinematic data for both groups were collected using an 8 camera infrared motion analysis system (Qualisys)1 and spherical 25 and 19 mm diameter retroreflective (Scotchlite 8850)2 markers attached to the skin over palpable bony landmarks (Weller et al. 2006c). A standard wand-based 3D calibration procedure was performed prior to data collection. The linear calibration value of the 3D motion capture system was then used to determine the theoretical maximum error associated with the measurements of joint angles in the frontal plane. Given that linear errors will have a greater effect on joint angles based on smaller segment lengths, the effect on joint angle measurements were estimated for the MCP/MTP angle. In particular, the distance between the virtual mid MCP/MTP marker and the coronet marker were used for the smallest segment length. This was then combined with the linear calibration error to calculate the largest angle error using trigonometry.

Kinematic data of all 4 limbs were collected for Group 1 whilst walking and trotting at a steady state through the calibrated motion capture volume. A minimum of 32 complete strides in walk and 17 strides of trot were successfully acquired for each pony.

Data were collected for Group 2 by setting up the motion analysis equipment at the training yard. The forelimbs of each horse had markers placed over bony landmarks, as shown in Figure 1. Horses were walked into the calibrated area and positioned standing squarely and equally weightbearing. Data were collected for 3 s; the horse was led away and repositioned each time to acquire 3 repeated measurements. The horse was then walked at a steady state 4 times through the area to obtain mid-stance walking information.

Figure 1.

Cranial view of distal forelimbs of subject from Group 2 with retroreflective motion analysis markers attached.

The acquired 2D data were then converted to 3D coordinates using Qualisys Track Manager Software (QTM)1 and the marker trajectories labelled appropriately. Custom written programs within the software MATLAB3 were used for subsequent data processing. Firstly, a horse-based reference system was defined based on 2 virtual markers (mid-point between the left and right proximal scapula spine and between the left and right tuber sacrale) (Fig 2). The x and y coordinates of these virtual markers were used to define the sagittal plane of the horse and this defined the x-axis of the local (horse based) reference frame (cranial positive). The z-axis of the local reference frame remained vertical (positive upwards), the local y-axis orthogonal to the local x and z-axes to form a right handed Cartesian coordinate system (positive to the left). This local reference frame was applied to the data for each forelimb mid-stance phase (in the walking and trotting horse) and during quiet standing. Midstance was calculated as the frame half way between ‘foot on’ and ‘foot off’ times, extracted by visual inspection, using the QTM software. Joint angles in the 2D frontal plane, spanned by the y and z-axes within the local horse-based reference frame, were calculated for the carpus, MCP, tarsus and MTP joint from the 3D coordinates (Fig 1).

Figure 2.

Illustration of global (‘Qualisys’) and local (‘horse based’) reference frames used for calculation of frontal plane joint angles at mid stance. The x-axis of the local reference frame is defined using 2 virtual markers (stars), the z-axis remains vertical and the y-axis is chosen perpendicular to the x and z-axis forming a right handed Cartesian coordinate system. Horses were walked in both directions (here in approximately negative Qualisys ‘x’ direction) through the calibrated area, the local reference frame, however, was always chosen with the positive x-axis pointing forwards with respect to the horse's direction of movement. The Qualisys (global) axis system is defined below by the axes of the graph.

Standing joint angles in the frontal plane were determined for Group 2, as were midstance angles for the carpus and MCP joints. For Group 1, walking and trotting midstance joint angles in the frontal plane were determined for the carpus, MCP, tarsus and MTP joints. Angles were recorded from the lateral aspect; that is an angle of <180° represented valgus angulation about that joint and >180° represented varus deviation.

To account for interhorse variation and investigate the variability of measurements, the joint angles were subsequently corrected for their angular limb deformity by determining the deviation of the joint angles measured from their means. Deviations of the carpus, MCP, tarsus and MTP joints were dealt with separately for each limb (left and right) and individual horse during standing, walking or trotting.

Statistical analyses were performed using SPSS (version 17)4. Following the satisfaction of assumptions testing (normality and homogeneity of variance), parametric tests were conducted for data analysis. An independent t test was used to compare the calculated joint angles between: 1) groups during walking; 2) left and right limbs; and (3) gaits during walking and trotting for Group 1 and standing and walking for Group 2. A one-way ANOVA was used to determine inter- and intra-horse variations. All statistical results were deemed significant at the 5% level (P<0.05). Descriptive statistics were used to explore the variability of joint angle measurements in the frontal plane between conditions.


Assessment of measurement errors

The average MCP/MTP to coronet segment length over all stance phases of all horses in Group 1 (ponies) was 100 mm. The largest camera error obtained was 0.74 mm across all calibrations. Trigonometry was applied to calculate the maximum theoretical measurement error for calculation of joint angles in the frontal plane of 1.28°.

Variation between groups

Carpal and MCP joint angles in the frontal plane at midstance were compared for the 2 groups at walking. A statistically significant difference between groups was observed in the carpus (P<0.0001), Group 1 having a greater valgus deviation than Group 2 (Table 1). However, there was no significant difference between groups walking in the MCP joint (P = 0.165).

Table 1. Mean ± s.d. joint angle for the carpus, metacarpophalangeal, tarsus and metatarsophalangeal joints measured in the frontal plane during standing and at midstance for walking and trotting
Standing (TBs)174.2 ± 2.0°176.7 ± 3.87°  
Walking midstance (TBs)174.7 ± 2.0°177.0 ± 4.3°  
Walking midstance (ponies)172.4 ± 1.8°178.1 ± 7.45°172.9 ± 2.7°176.4 ± 5.1°
Trotting midstance (ponies)172.45 ± 2.2°182.4 ± 10.8°169.4 ± 3.17°169.4 ± 6.3°

Left and right limb variation

There was no statistically significant difference between the joint angles in the left and right tarsal (P = 0.100) and MTP joints (P = 0.687) in Group 1, although there was a significant difference between the left and right carpus in Group 2 (left: 174.16 ± 2.03, right: 174.96 ± 1.96, P<0.0001) and the MCP joint in both groups (Group 1, left: 183.09 ± 10.50, right: 177.48 ± 7.42, P<0.0001; Group 2, left: 177.76 ± 3.92, right 175.96 ± 4.13, P = 0.002),

Comparison of standing and walking, and walking and trotting

In Group 1 the MCP joint angle showed a significant difference (P = 0.001) between walk and trot, changing from a valgus angulation to a varus angulation, the calculated angle was (mean ± s.d.) 178.1 ± 7.45° at walk, changing to 182.4 ± 10.8° when trotting (Table 1). In the tarsus and MTP joints a significant difference (P<0.0001) towards an increasing valgus deformity was observed: 172.9 ± 2.7° to 169.4 ± 3.17° in the tarsus and 176.4 ± 5.1° to 169.4 ± 6.3° in the MTP (Table 1). There was no significant difference between walk and trot in the carpus (P = 0.782).

Group 2 showed a significant difference at the carpus (P = 0.045), the valgus deformity decreased from a standing angle of 174.2 ± 2.0° to 174.7 ± 2.0° when walking. There was no significant difference between standing and walking in the MCP joint (P = 0.634) (Table 1).

Interhorse variation

There was a significant difference between all individuals for each joint in every condition (standing, walking and trotting) (P<0.0001); except in the standing MCP joint where no significant difference was observed (P = 0.691).

Intrahorse variation

No statistically significant difference was recorded between stances, or strides, within each horse (P>0.05).

Measurement variation

The effect of variation between individuals (i.e. the ALD of an individual) was removed by subtraction of the mean value (for each joint of each horse) to enable evaluation of the variation in repeated measurements within stances, or strides. Figure 3 shows the variation in frontal plane joint angles between conditions (standing, walking and trotting) within the 2 groups for each joint. The variation of joint measurements between conditions: standing and walking (Fig 3e,f) and walking and trotting (Fig 3a–d) for each group are shown in Figure 3. The MCP and MTP joints (Fig 3b,d,f) demonstrate greater variation than the carpus (Fig 3a) and tarsus (Fig 3c).

Figure 3.

Box plots to compare the deviation of the joint angles calculated in the frontal plane from the mean (individual horse mean for each joint). Comparisons between walking and trotting in the pony group (red); A: carpus, B: metacarpophalangeal, C: tarsus and D: metatarsophalangeal joints. A comparison of standing and walking within the Thoroughbred group (blue) is shown in E and F. Values given are corrected for the interhorse variation and therefore reflect overall intrahorse variation between gaits. The midline represents the median, the box the interquartile range (IQR), whiskers 1.5 IQR, outliers are depicted as circles <3 IQRs and stars >3 IQRs.


Conformational measurements

The mean standing angle measurements and standard deviations recorded in this study for the flat racing Thoroughbred group are in agreement with those previously reported in a study of National Hunt Thoroughbreds (Weller et al. 2006b). The larger standard deviation observed in the MCP and MTP joint angles are all consistent with previous work (Weller et al. 2006b) and may be the result of greater measurement errors at these joints. Segments are shorter distally: radius and metacarpus, or tibia and metatarsus, are longer bones compared to the phalanges. Therefore, error associated with marker placement will have an exaggerated effect at the more distal joints, where the lengths are smaller.

Variation between groups:  Comparison between groups was only possible for forelimb walk data. A significant difference was observed between pony (172.4 ± 1.8°) and Thoroughbred (174.7 ± 2.0°) groups for the frontal plane carpal angle, suggesting that the Thoroughbred population in this study had straighter limbs at their carpi in comparison to the ponies. It is not possible to determine whether this is an effect of the small genetic pool within the Thoroughbred breed (Love et al. 2006), the training regime, cyclic loading (Pool and Meagher 1990) or other environmental factors. No difference was observed at the MCP joint between pony and Thoroughbred groups, the larger standard deviation at this joint may be a result of greater variation amongst both groups, or due to measurement error as discussed. A larger study population that would include hindlimb joint angles may be required to draw further conclusions.

Variation between left and right limbs

No difference was observed in the hindlimb joints between left and right limbs; however a significant difference was observed in the carpus and MCP joints. In horses, 60% of the weight is supported by the forelimbs and the hindlimbs provide power for locomotion (Hoyt et al. 2000). It is possible that the difference observed between left and right forelimbs and the nonsignificant observations between left and right hindlimbs may be a result of the greater support provided by the forelimbs to the GRFs.

It is accepted that human individuals may be left or right handed. Commonly, the ‘favoured’ side will be stronger as a result of induced muscle hypertrophy. It has been demonstrated that other species, including chimpanzees (Hopkins and Nir 2009) and dogs (Colborne 2008), may also exhibit such handedness. It is possible that differences in left and right limb conformation in horses could be associated with an equivalent favouring of one side; either as a natural characteristic or a result of training or the rider.

Comparison of standing and walking and walking and trotting

Conformational measurements obtained previously during walking demonstrate less variation than those obtained during standing (Pourcelot et al. 2002). However, the findings of this study contradict previous work, in that variation appears independent of gait. For each horse the number of strides (n = 3–23, mean carpus and MCP: 6, tarsus: 13, MTP: 12) evaluated for each joint during this study exceeds the 3–5 strides previously suggested as a representative number (Drevemo et al. 1980). Variation in the distal joints (MCP and MTP) was observed to be larger than for the proximal joints (carpus and tarsus). The greater variation reflects the larger compression of these distal joints, which will increase with force (McGuigan and Wilson 2003) and, therefore, as a function of speed. As a consequence of simple geometry, variability in the compression of the joint (e.g. as a function of speed or unequal weightbearing during standing) will result in a variation to the varus or valgus angulation observed in the frontal plane. Therefore the observed increasing varus in the MCP joints and valgus in the tarsus and MTP joints in the ponies is supportive of the hypothesis that joint angulation will become more pronounced with increasing speed. As stance time decreases with faster gaits the GRF increases (Alexander 1979; Witte et al. 2006). The increase in valgus and varus deformities between walk and trot, and standing and walking are likely to be a result of the increased GRF and cause a larger torque at that joint.

Inter- and intrahorse variation

Individual variation was significant in all joints, except the standing MCP. The similarity in the angle measurements for the standing MCP joint may be explained by the artificial positioning of the horse, in a ‘square’ position for the standing conformation assessment. When positioning squarely, the observer will align the feet and most distal limbs, including the MCP, as ‘squarely’ as possible, according to the individual's spatial perception of the subject. As data were only collected for the forelimbs in the Thoroughbred group it is not possible to comment if this result may also have occurred for the MTP joint.

In agreement with the hypothesis, no significant difference was observed between strides within an individual horse.


Passive motion analysis systems requiring the use of retro-reflective markers are susceptible to several sources of error. Firstly, the average measurement error, associated with the 3D multi-camera motion analysis systems in determining the marker coordinates within the calibrated volume (here approximately 6 × 3 m) were 0.74 mm. When evaluating the effect of this linear error on the calculation of frontal plane joint angles most extreme error computed (for the MCP/MTP angle, the angle based on the shortest segments) was 1.3°, a comparatively small value compared to the variation seen within the groups of horses (Table 1). In addition to calibration error, intra- and interoperator variation when placing cutaneous markers over palpable bony landmarks have been reported to be approximately 3° and generally higher for measurements involving more proximal anatomical landmarks (Weller et al. 2006b). Finally, relative movement of the skin over the bony landmarks must be considered a potential source of error when collecting kinematic data in moving animals, and has been shown to have greater effects in more proximal parts of the limbs (van Weeren et al. 1990; Sha et al. 2004).


This study validates the measurement of joint angles from the front in the walking and trotting horse using kinematic data; the interhorse variation in joint angle measurements exceeding that of the intrahorse variation. Variations in measurements obtained in the frontal plane joint angles appear to be small between gaits. The increase in joint angulation observed here between standing and walking, and walking and trotting, is significant and warrants further investigation to determine the effect of faster gaits and speeds on joints. Future work will consider how an individual's conformation may affect the changes in angulation during movement, allowing predictions on the influence of conformation with regards to performance, career and injury.


We would like to thank the Horserace Betting Levy Board (HBLB) for funding this work.

Conflicts of interest

The authors declare no potential conflicts.

Manufacturers' addresses

1 Qualisys Medical AB, Gothenburg, Sweden.

2 3M Personal Safety Products, Manchester, UK.

3 The MathWorks, Inc., Natick, Massachusetts, USA.

4 SPSS Inc., Chicago, Illinois, USA.