Reasons for performing study: Head and neck positions (HNP) in sport horses are under debate in the equine community, as they could interfere with equine welfare. HNPs have not been quantified objectively and no information is available on their head and neck loading.
Objectives: To quantify in vivo HNPs in sport horses and develop o a model to estimate loading on the cervical vertebrae in these positions.
Methods: Videos were taken of 7 Warmbloods at walk on a straight line in 5 positions, representing all HNPs during Warmblood training and competition. Markers were glued at 5 anatomical landmarks. Two-dimensional angles and distances were determined from video frames for the 5 HNPs and statistically compared (P<0.05). A new simulation model was developed to estimate nuchal ligament cervical loading at these HNPs.
Results: The mean angles were significantly different between the 5 HNPs for the line between C1 and T6 with the horizontal and for the line connecting the facial crest (CF) and C1 with the vertical, while the vertical distance from CF to the lateral styloid process of the radius (PS) was significantly different between all 5 positions (P<0.05). The estimated nuchal ligament loading appeared to be largest at the origin of C2 for all HNPs, except for the ‘hyperextended’ HNP5; the ‘hyperflexed’ HNP4 showed the largest loading values on the nuchal ligament origins at all locations.
Conclusions: HNPs can be accurately quantified in the sagittal plane from angles and distances based on standard anatomical landmarks and home-video captured images. Nuchal ligament loading showed the largest estimated values at its origin on C2 in hyperflexion (HNP4).
Potential relevance: Modelling opens further perspectives to eventually estimate loading for individual horses and thus ergonomically optimise their HNP, which may improve the welfare of the sport horse during training and competition.
Since the publication of Ricardi and Dyson (1993) there has been a growing interest in the head and neck region as a potential source of pathological locomotor disturbances and the availability of EMG techniques (Tokuriki et al. 1999; Wijnberg et al. 2004) has proven to be a great asset in investigating this region. These studies proved that a (peri-)neuritis in combination with myelopathy and facet joint arthrosis can, as in man, also result in pain and disability in horses (Moore et al. 1992; Dunbar et al. 2008; Levine et al. 2008). Moreover, different head and neck positions (HNP) have become under debate in the equine community, especially in the disciplines of dressage and showjumping, as they might also have an effect on normal, physiological function (Jeffcott et al. 2006).
The use of specific HNPs in training is not uncontested. Recent discussion in the international dressage world focuses on the extremely flexed HNP ‘hyperflexion’, ‘rollkur’ or ‘low, deep and round’ (Jeffcott et al. 2006; Heuschmann 2007; von Borstel et al. 2009), which is believed by some trainers to be a useful tool to improve the gymnastic ability that is asked from today's high performance dressage horses and is rejected by others on the basis of presumed negative effects on equine welfare.
Some recent studies on locomotion demonstrated that, although a flexed head-neck position induced an increase in range of motion in the lumbar back in the unridden horse and would imply an activated use of the hindlimbs, a better step under the horse's body and a more equal divided weight load between fore- and hindlimbs (Gómez-Álvarez et al. 2006), this effect could not be reproduced while horses were ridden (Rhodin et al. 2009). An extremely elevated neck, however, caused an increase in extension of the thoracic and lumbar back in the unridden as well as in the ridden horse (Gómez-Álvarez et al. 2006; Rhodin et al. 2009). Furthermore, an extremely elevated neck was shown to affect the functionality of the locomotor apparatus much more than an extremely low neck by an increase in peak vertical force in the forelimbs, which is, among other factors, a potential risk factor for the development of injuries (Biau et al. 2002; Weishaupt et al. 2006; Waldern et al. 2009).
Therefore, the first objective of this study was to test whether 5 in vivo HNPs commonly used during training and competition in sport horses, could be objectively quantified in the sagittal plane from straightforward angles and distances using a home-video camera and standard anatomical landmarks. The second objective was to test whether the loading on the cervical vertebrae could be calculated from an anatomical model, developed from measured cervical vertebral dimensions (Part 1), calculated centres of rotations (Part 2), determined intervertebral angle limitations (Part 3), earlier reported nuchal ligament properties (Part 4), and, subsequently, whether the model could be used to estimate the relative differences in loading between the objectively quantified HNPs.
Materials and methods
Seven healthy, base-level trained Dutch Warmblood Sport horses (sex: 5 mares, 2 geldings, age: 10.3 ± 3.6 years [mean ± s.d.], height at the withers: 161.2 ± 1.4 cm, weight: 531 ± 47.3 kg), with no history of respiratory disease, cardiovascular disease, musculoskeletal or neuromuscular disorders participated in the study. Radiographic and ultrasound examination of their spinal column showed no abnormalities.
Head and neck positions
Videos were taken from the total group of 7 horses at the walk on a straight line in 5 different HNPs, which were accomplished using side rains (Fig 1). The 5 HNPs, of which HNP1, HNP2, HNP4 and HNP5 were identical to those used in an earlier experiment (Weishaupt et al. 2006; Gómez-Álvarez et al. 2006; Rhodin et al. 2009; Waldern et al. 2009), were defined as followed (Fig 2): HNP1: Free, unrestrained, neutral position; HNP2: Neck raised; bridge of the nose around the vertical; HNP4: Neck lowered and considerably flexed; nose pointing toward the chest; HNP5: Neck raised and considerably extended; bridge of the nose in front of the vertical; and HNP7: Neck lowered and flexed; nose pointing towards the carpus.
In addition, they were evaluated by an international dressage team to check for realistic and correct interpretation, and thus considered as being representative and commonly used to date in training and competition. HNP7 was included because 2 interpretations of the hyperflexion, ‘rollkur’ or ‘low, deep and round’ training position were found to exist among riders (HNP4 and HNP7). To accustom the horses to the experimental set-up, they were trained in the different head and neck positions on the lunge for at least 3 weeks. The Animal Experimental Committee of Utrecht University (DEC) had approved the study.
Spherical reflective markers (Proflex)1 were glued to the skin over the dorsal spinous process of T6, the wing of the atlas (C1), the rostral part of the facial crest (CF), the suprascapular tuberculum (TS) and the lateral styloid process of the radius (PS). Two additional markers were placed on the horses bridle, while the bridge of the nose (BN, BN') could be identified in every video frame. The known distance between these 2 bridle markers was used as a calibration reference for the video-frame analysis. A home-video camera (Canon digital camcorder MV600i)2 was set on the line A–C perpendicular to the long side of the riding arena, where the horses were evaluated.
Video analysis was performed using a custom-made software programme (ImageJ 1.42)3. The coordinates of the reflective markers were determined using the software and HNPs were described by use of 4 angles and 2 distances. For each HNP the following angles and distances were measured (Fig 1):
Angle 1[C1-T6-Hor]: the angle between the wing of the atlas (C1), the dorsal spinous process (T6) with the horizontal. This angle has a negative value when the atlas is distal in relation to T6.
Angle 2[CF-C1-T6]: the angle between the rostral part of the facial crest (CF), the wing of the atlas (C1) and the dorsal spinous process of (T6).
Angle 3[C1-CF-vertical]: the angle between the wing of atlas (C1), the rostral part of the facial crest (CF) with the vertical. This angle has a negative value when the facial crest is more caudal in relation to the atlas.
Angle 4[BN'-BN-vertical]: bridge of nose with the vertical. This angle has a negative value when the bridge of the nose is behind the vertical.
Distance A[CF–TS]: the horizontal distance between the rostral part of the facial crest (CF) and the supraglenoid tubercle (TS).
Distance B[CF–PS]: the vertical distance between the rostral part of the facial crest (CF) and the lateral styloid process (PS).
For each HNP, 5 video captures were taken per horse with the axes of the head and the neck in a straight line in the sagittal plane to avoid parallax errors. Measurements were done on the video frames captured at the walk on a straight line with the camera perpendicular to and on the left side of the horse and the wall of the riding arena on the right side of the horse at the moment, when the left front limb (LF) was at midstance with the left metacarpal bone in a vertical position (Figs 1, 2). For Angles 1–4 and the Distances A and B, the means and s.e. for the group of 7 horses were calculated.
Radiographical analysis and CT analysis
For every HNP digital L/M radiographs were made of the cervical vertebrae C2–C7 of one of the horses of the group (Horse 1 from the group of 7 horses), while it was standing square with the head and neck in the sagittal plane. Therefore, the in vivo longitudinal angles between the cervical vertebrae (C2–C7) could be subsequently measured. Cervical angles were given a negative value when the corpus of the cranial vertebra sloped in regards to the corpus of its successive caudal vertebra.
From an additional Warmblood horse (Horse 8), which was subjected to euthanasia for reasons other than musculoskeletal or neuromuscular disorders, the ex vivo longitudinal angles between the cervical vertebrae (C2–C7) were measured in 5 different positions (neutral, extension 20 and 40°, flexion 20 and 40°), which according to Sleutjens et al. (2010: ex vivo HNPs) would represent, respectively, the in vivo angles HNP 1, HNP 2, HNP5, HNP 4 and HNP7.
The (ex vivo) CT images (Horse 8) were overlayed on top of the (in vivo) digital radiographical images (Horse 1) using imaging processing software (Adobe Photoshop 7)4 to facilitate accurate and representative cervical intervertebral angle measurements to feed the model.
The first part in building the model was to measure the cervical vertebral dimensions from the ex vivo specimen (Horse 8). For this, digital images were produced of the vertebrae C1–T2 using a CT scan (Sleutjens et al. 2010) and a rectangle was projected over the vertebral corpus with the cranial head and the caudal tail as longitudinal, sagittal reference length. Therefore, using free software (Delphi 7) the dimensions of the vertebrae were determined. The following values were found C1: 4.4 cm, C2: 13.4 cm, C3: 11.0 cm, C4: 11.1 cm, C5: 10.6 cm, C6: 9.8 cm, C7: 8.3 cm, T1: 5.9 cm, T2: 5.9 cm, T3: 5.6 cm, T4: 5.3 cm, T5: 5.1 cm and T6: 5.1 cm.
The second part was to calculate the centres of rotations from the same ex vivo specimen (Horse 8). For this, the position of the markers objectively determined and the joint centres of rotation determined according to the method of Reuleaux (1875) and from visual inspection. For this, videos were made of the ex vivo movement of the head and C1, while these anatomical structures were covered with a set of markers. Using commercially available software (Delphi 7) the coordinates of the markers in the video frames were calculated and thus the joint centre of rotation of C1 was determined. The other joint centres of rotation were determined in a similar way and supported by visual inspection: the most likely position of the joint centre of rotation was deduced from their intervertebral movement.
The third part was to determine angle limitations and these were determined again from the ex vivo specimen (Horse 8). The maximal flexion of C1 against the caudal head was 96° and maximal extension 173° leaving a total range of motion of 77°. This value falls within the range of Clayton and Townsend (1989) for head C1 movement (69.7–103.1°). In the model, angle limitations are set relative to a 90° angle: for flexion this was set to +6° (range 90–96°) and for extension to +83° (range 90–173°). The other intervertebral angles were determined in a similar way and set within the limits provided by Clayton and Townsend (1989) and Sleutjens et al. (2010).
The last, fourth part is to use earlier reported nuchal ligament properties (Gellman and Bertram 2002a,b). For this, the rest length of the nuchal ligament, being 64.7 cm length measured from the origin on the head, was used as 100% reference (Gellman and Bertram 2002a,b). The model was then designed to estimate in vivo angular values (Table 3) for the 5 HNPs using the earlier measured intervertebral cervical angles (Table 2). From this model the in vivo loading on the different origins of the nuchal ligament (head, C2, C3, C4, C5: Fig 3) could be quantitatively estimated. For this, the cross section values of respectively 4.2 cm2 for origin occiput, 6.0 cm2 for origin C2, 1.1 cm2 for origin C3, 2.0 cm2 for origin C4 and 1.1 cm2 for origin C5 of Gellman and Bertram (2002a,b) were applied, which were also verified ex vivo in Horse 8. Furthermore, the curves of the relative length of the nuchal ligament (as % from rest length) and the force on its 5 different origins (in N, according to Gellman and Bertram 2002a,b) were used to estimate the forces at the origins occiput, C2, C3, C4, C5 for the 5 HNPs (Table 4).
Table 3. Estimated angular values from an anatomical model based upon cervical intervertebral angles C2–C7 in HNP 1, 2, 4, 5 and 7 from Table 2, n = 2)
Table 2. Ex vivo objective quantification of different head and neck positions (HNP) through concomitant cervical vertebral body angles (°) determined from in vivo digital radiographic (Horse 1) and ex vivo CT images (Horse 8) in the 5 different HNPs
In vivo HNP of n = 1
Ex vivo HNP of n = 1
Table 4. Estimated loading values of the nuchal ligament and its branches in different head and neck positions (HNP) based upon in vivo and ex vivo radiographic and CT data of 2 horses (n = 2)
In vivo HNP
1) bold: highest L, S, F value between HNPs. 2) italic: highest F value within a HNP.
F occiput (N)
F origin C2 (N)
F origin C3 (N)
F origin C4 (N)
F origin C5 (N)
Ex vivo HNP
Statistical analysis of the angle and distance data was performed in a linear mixed model with post hoc Bonferroni correction (P<0.05) using free software (‘R’ version 2.8.1)6.
The measured Angles 1–4 (Table 1) were statistically compared with those estimated from the model (Table 3) in a paired t test; the correlation between measured and estimated angular values was tested in a Pearson linear regression test using commercially available software (SPSS version 16.0)7 at a significance level of P<0.05.
Table 1. Mean ± s.e. in vivo measured angles (°) and distances (cm) defining different head and neck positions (HNP) commonly used in training and competition evaluated in a group of horses (n = 7)
P<0.05; pairs a, b and c are not significantly different.
For each HNP, Table 1 shows both mean and s.e. of the measured in vivo Angles 1–4 and Distances A and B of the evaluated group of horses (n = 7). Mean angles were significantly different between the 5 HNPs for C1–T6 with the horizontal (Angle 1) and for CF–C1 (Angle 3) and bridge nose (Angle 4) with the vertical (P<0.05). For Angle 2 all HNPs, except for HNP2 and HNP4, were significantly different (P<0.05). The horizontal distance from CF to TS (Distance A) was different only between HNP1 and all others, while the vertical distance from CF to PS (Distance B) was significantly different between all 5 HNPs (P<0.05).
Table 2 shows the ex vivo measured intervertebral angles in the 5 different ex vivo HNPs (neutral, extension 20 and 40°, flexion 20 and 40°; Sleutjens et al. 2010), being comparable to each of the 5 in vivo HNPs (respectively HNP 1: -16,2°, HNP2: 3.6°, HNP5: 34.8°, HNP4: -33.3 and HNP7: -41.5°). In vivo HNP1 and HNP2 were approximately 15° underestimated by the applied ex vivo HNPs.
The model was then further used to estimate angular values (Angles 1–4: Table 3), which did not appear to be significantly different from the angular values measured (Angles 1–4: Table 1). Moreover, there was a rather high and significant correlation between measured and estimated angular values in the 5 HNPs.
The loading on the 5 different origins of the nuchal ligament was estimated (Table 4) and appeared to be relatively largest at HNP4 and at the origin at C2 in nearly all positions, except at HNP5; there the relatively largest load was nearly similar to the relatively smallest load at HNP4.
The aim of this study was to objectively quantify 5 in vivo HNPs commonly used during training and competition in sport horses, based upon those used in earlier studies (HNP1–6). Position HNP7 was included because 2 interpretations of the hyperflexion, ‘Rollkur’ or ‘low, deep and round’ training position were found to exist among riders (HNP4 and HNP7).
Head and neck positions could be accurately quantified in the sagittal plane and appeared significantly different in their 2D angles and distances. It was therefore possible to discriminate objectively between the HNPs that are representative for what is used nowadays in training and competition, using the mean angles of C1–T6 with the horizontal (Angle 1) and of CF–C1 with the vertical (Angle 3) in combination with the horizontal distance from CF to TS (Distance A) and the vertical distance from CF to PS (Distance B).
The in vivo‘hyperextended’ HNP5 (ex vivo 40° extension) was the HNP with the largest, most positive Angles 1, 3 and 4 and the longest vertical Distance B; in HNP2 (ex vivo 20° extension) the Angle 3 was close to zero, the in vivo‘neutral’ HNP1 had the largest Angle 2 and the longest horizontal Distance A; HNP4 (ex vivo 20° flexion) the smallest Angles 3 and 4; the in vivo‘hyperflexed’ HNP7 (ex vivo 40° flexion) was the HNP with the smallest, most negative Angle 1 and the shortest Distance B (P<0.05).
The anatomical model used in this study was developed as a pilot to estimate the loading on the cervical vertebrae in the 5 different HNPs. To date, the HNPs of sport horses are under debate in the equine community. Therefore, it makes sense to compare relative loading values between the different HNPs used in practice. The estimated angular values (Table 3) in the 5 different HNPs were not significantly different from those measured (Table 1), which can be considered as an illustration of the accuracy of the anatomical model used.
The estimated load at the different origins of the nuchal ligament all were largest at HNP 4 and at the origin of C2 in nearly all positions, except in HNP5, where loading was least and nearly similar to the smallest measured load in HNP4. These output data of the model, however, are estimated values and based on measurements from a single horse in vivo (Horse 1) and one ex vivo (Horse 8). Data are therefore preliminary and there is a need to more extensively test the repeatability of the output from the model used in this study.
Nevertheless, it is known that mechanical nociceptive threshold (MNT) values vary in horses with differing subject status and ridden exercise level (Haussler and Erb 2006). Therefore, pressure algometers could clinically provide an objective, noninvasive and repeatable tool to measure mechanical nociception in horses in the different HNPs and thus could be used to monitor (over-)loading of the origins of the nuchal ligament, eventually preventing pain and dysfunction of the neck area and objectively evaluate the effect of manual therapy treatments (De Heus et al. 2010).
In conclusion, the first hypothesis of this study appears to be supported. From the angles of C1–T6 with the horizontal and of CF–C1 with the vertical in combination with the vertical distance from CF–PS, we were able to objectively quantify the HNP in a particular horse during training and competition from home-video images. This enables routine (standard video) monitoring, including timing, of the applied HNPs during training and competition in the practical field situation.
Furthermore, the second hypothesis also seems to be supported. The outcome of the anatomical model showed possible loading difference between different HNPs and between different cervical regions. Although it is realised that the model presented in this study is in its infancy and needs further elaboration and validation, it is thought that the outcome contributes to our understanding of the ergonomic impact (van Weeren 2005) of HNPs that are requested from today's sport horse. Further testing in a larger group of horses would show whether these (over-)loading differences indeed are structural for particular HNPs and regions. This would be of potential help in better understanding the occurrence and eventually in improving treatment and prevention of locomotor disturbances in sport horses, induced by suboptimal HNPs during training and competition. In this way, this approach may contribute to equine welfare and indirectly to promotion of the equine industry.