A Case-controlled Study Investigating 
Health, Management and Behavioural Features
of Horses Commonly Described as Headshakers
K Taylor,¹ S Cook,¹ and  DS Mills^2
1 Department of Medical Statistics, Faculty of Computing Sciences and Engineering,
De Montfort University, Leicester, LE1 9BH, UK
Voice 44 (0)116 2078510     Fax 44 (0)116 2506114      katyt@dmu.ac.uk
² Animal Behaviour, Welfare and Cognition Group, University of Lincoln, Caythorpe Lincs NG32 3EP, UK

Introduction

Headshaking is a poorly understood condition affecting the ridden horse, involving the sudden, uncontrolled tossing of the head.  Many cases remain idiopathic, but it has been suggested that headshakers are more likely to have immune-related problems such as chronic obstructive pulmonary disease (COPD), respiratory disease and allergies (Mair and Lane, 1993).  It has also been suggested that management practices and stabling, diet (in particular over-feeding and under-exercise) and type of use (dressage and jumping) may play a role in exacerbating the condition (Cook, 1980, 1992).

In order to establish whether any of these putative risk factors appear to be significant in headshaking, a case-controlled study was carried out comparing the reported prevalence of these factors in headshakers and a similar non-headshaking subject from the same area. In addition we aimed to assess how prevalent the typical behaviours found in headshakers were in horses described as non-headshakers, i.e. the uniqueness of the signs to the general description of a horse as a “headshaker”.

Methods

Owners of apparent headshakers who had already participated in the National Equine Headshaking Survey (Mills et al, 2002) were approached by post with a request to fill in an adapted version of the survey for a horse that was not thought to be a headshaker.  Owners were requested to choose a control horse on the basis of its similar physical attributes and its geographical proximity to the problem horse. Data relating to the breed, sex, age, height, professional versus leisure use of the horse, locality and size of yard were recorded for both populations for the purposes of assessing the level of match between subjects.

Specific questions for comparison addressed the horse’s workload, management, diet, medical history and headshaking behaviour.  The prevalence of 12 behaviours which are commonly referred to in the description of headshakers was used for this latter purpose. These were headshaking at rest, at exercise and when excited, vertical and horizontal headshaking, ‘flipping’ the nose, acting like a bee just flew up the nose, snorting, rubbing the nose on the ground when stationary and when in motion, rubbing the nose on objects and striking at the nose with the foreleg. 

Questions with two response options were evaluated using a McNemar’s test, Q_M, for paired samples, which approximates to a chi-square distribution with one degree of freedom (SAS v6.12 SAS Institute Inc, 1996).  Where one or more variables had an expected count of five or less, a Fisher’s exact test was performed.  Bowker’s test for symmetry, Q_B, was used for questions with more than 2 response categories.  A critical probability of 0.05 was used for data evaluation in all cases.

Results

Eighty-three headshakers were successfully paired with an animal not considered to be a headshaker.  The horses were well matched with regards to sex, breed, age, height, professionalism of use, locality and size of yard, with no significant differences between the two groups in any of these categories (p>0.05).  There was no significant difference between the pairs with regards to amount of turnout, type of bedding, regularity of teeth checking, worming or removal of whiskers (which was practiced by roughly a quarter of all owners). 

Whilst horses were matched for professional status, they were not presumptively matched for specific uses and owners were requested to specify all the pursuits for which their horses were used. Amongst other pursuits, 64% of horses were used for hacking out.  The next most popular uses were dressage (27%), jumping (25%) and eventing (21%).  Other pursuits were reported with a prevalence of less than 16%.  There was a significant difference between the two groups with respect to hacking out.  Sixty control horses (72%) were used for hacking compared to 47 (57%) of the headshakers (Q_M = 6.259, p = 0.012).  There were no significant differences between the horse pairs with regard to the likelihood of being used for dressage or jumping. There was also no difference between the groups for the total number of uses for each horse.  Thus neither group was used in a more varied manner, with horses tending to be used for two pursuits.  The majority of horses (78%) were worked at least 3 days a week and no significant difference was found between the pairs in the amount of work undertaken.

Herbal supplements were more likely to be given to the headshaking population (37% of the headshakers, 18% of the controls; Q_M = 9.85, p = 0.002).  There were no other significant differences between the two groups for any of the other food types listed.  Ninety-five percent of subjects were provided with fresh grass and 89% hay.  Sixty-two percent were fed chop (chopped straw), 61% concentrated feed supplements, 55% sugar beet, 39% vitamin supplements and 37% nuts or cubes (prepared cereal based feed supplement).  The feeding of haylage (ensiled hay) (20%), straights (cereals) (16%), fruit (10%), probiotics (live microbial feed supplement) (9%) and other types of feed (12%) was less common.  There were no significant differences between the pairs with regard to the total number of foods given to each horse.  On average both groups were given between 5 and 6 of these food types in their diet.

On average, 5% of the horses were reported to have COPD (5 controls, 4 headshakers), 9% other respiratory problems (5 controls, 10 headshakers) and 9% other allergies (6 controls,13 headshakers).  The proportion of horse pairs that disagreed on this was not significant for COPD or other respiratory problems, but there was a suggestion that headshakers may be reported to suffer from non- respiratory allergies more commonly than controls (Q_M = 3.267, p = 0.071). 

The headshakers were significantly more likely to be reported with each of the 12 characteristics often associated with headshaking (Fig 1. p<0.001 in all cases).  The prevalence of the behaviours seen in the headshaking group ranged from 25% for horizontal headshaking to 100% for headshaking at exercise.  Most of these characteristics were rare in control horses (between 0% and 4%) with the exceptions of shaking the head when excited and vertical headshaking (13% and 10% respectively).  However, the mean number of symptoms per horse was 0.4 (95% CL 0.20, 0.64) for the controls and 8.7 (95% CL 8.01, 9.20) for the headshakers.

The majority of horses in each group were reported to have been treated by a veterinarian at some time in their lives (Fig 2).  There were significant differences between the two groups, with the headshaking group being more likely to have sought veterinary therapy, back specialists, homeopathy and other alternatives. However, for conditions other than headshaking, owners of headshakers were only significantly more likely to have sought the assistance of back specialists (Q_M= 6.26, p = 0.012) and homeopaths (Q_M= 5.56, p = 0.018).

Discussion

Although epidemiological tools, such as case-control surveys, do not distinguish between cause and effect, they can highlight potential risk factors worthy of further investigation. Currently suggested risk factors are based on their apparent frequency in case series or individual expert opinion.  For example, Cook (1992) suggested that sports such as dressage and jumping may lead to headshaking as an avoidance of excessive poll flexion or cervical pain.  This case control study suggests that, in general, there is no association between headshaking and the practice of dressage or jumping, which is in agreement with the reports of Lane and Mair (1987) and Newton et al (2000).  However, the population sampled was mainly that of privately owned, general-purpose horses, which may preclude assessment of the significance of individual specific competitive uses such as dressage, show jumping or racing. The association found between headshaking and not being hacked out may be a causal one, since owners of headshakers may be wary of this activity given the unpredictability of the behaviour in many cases. Alternatively, it may reflect an association between uses other than hacking (i.e. amateur competition work) and headshaking. 

This study does not support the hypothesis that headshaking is due to over-feeding of certain foods or under-exercise, resulting in a hyper-excitable horse (Cook, 1980).  The lack of any significant differences between the pairs may reflect a change in the horse’s diet and management in response to the onset of the condition.  However, a proximate association between management and behaviour (e.g. a reaction to excessive energy in the diet) as is often implied does not appear to exist.

Contrary to the speculation of Mair and Lane (1993), the headshaking group was not more likely to be reported to have COPD or other respiratory problems.  If reports of COPD and other respiratory problems are combined as possible causes of clinical airway inflammation, the results in both headshakers and controls are comparable to that of Lane and Mair (1987) who found clinical signs of COPD in 18% of their headshakers.  The suggestion of a possible association with other allergies is worthy of further investigation. 

The increased prevalence in use of alternative therapies amongst these owners  probably reflects frustration with the lack of success of conventional treatments for the treatment of headshaking which is commonly reported (Lane and Mair, 1987). This trend not only represents a source of lost revenue to the veterinary profession, but more importantly may pose a risk to the well-being of subjects in future disease, since the efficacy of such treatments remains to be established scientifically.

Comparison of the symptoms of the two groups suggested that many of the behavioural signs associated with headshaking are reported to occur only rarely if at all in horses not described as headshakers by their owners. These results might suggest that any horses showing a number of these signs is labeled as a headshaker and/or that headshaking is a distinctive, medical syndrome in which many management factors are not highly significant. A recent survey of horses described as “headshakers” (Mills et al, 2002) identified a potential for the term to be applied also to horses which simply nodded repetitively.  Nodding in the stable can be treated successfully with increased social contact (Cooper et al, 2000) or with a mirror in the stable (Mills and Davenport, 2002). Therefore it must not be assumed that when an owner describes their horse as a headshaker that it is necessarily suffering from some organic pathological process. However, the type of headshaking described in this study is accompanied by a variety of other symptoms suggestive of pain or irritation.  It is likely therefore that the horses in this survey tended to suffer from some form of head pain rather than behavioural disorder. This would support theories regarding the condition by Madigan et al (1995) and Newton et al (2000) and would explain the relative lack of significant management or health factors associated with the condition identified in this study. These results also reinforce the contention that attention to the signs of the presenting complaint and response to tests based on hypotheses arising from this are likely to be more productive than investigation of management and diet factors.

Acknowledgements

We would like to thank Professor Byron Jones for his initial suggestions regarding statistical tests and all the owners and veterinary surgeons without whose support this work would not have been possible.

References

Cook WR (1980) Headshaking in horses: Part 4. Special diagnostic procedures. Equine Pract 2: 7-15.

Cook WR (1992) Headshaking in horses: An afterword. Compend Contin Ed Pract Vet 14: 1369-1371.

Cooper JJ, McDonald L, Mills DS (2000) The effect of increasing visual horizons on stereotypic weaving; implications for the social housing of stabled horses. Appl Anim Behav Sci 69:  67-83.

Lane JG, Mair TS (1987) Observations on Headshaking in the horse. Eq Vet J 19: 331-336.

Madigan JE, Kortz G, Murphy C, Rodger L (1995) Photic headshaking in the horse: 7 cases. Eq Vet J 27: 306-311.

Mair TS, Lane JG (1993) Headshaking in Horses (Chapter 6) In Boden E (Ed), Eq Practice 2. Balliere Tindall, London, pp 109-119.

Mills DS, Cook S, Taylor K, Jones B (2002) Analysis of the variations in clinical signs shown by 254 cases of equine headshaking. Vet Rec 150: 236-240.

Mills DS, Davenport K (2002) The effect of a neighbouring conspecific versus the use of a mirror for the control of stereotypic weaving behaviour in the stabled horse. Anim Sci 74: 95-101.

Newton SA, Knottenbelt DC, Eldridge PR (2000) Headshaking in horses: possible aetiopathogenesis suggested by the results of diagnostic tests and several treatment regimes used in 20 cases. Eq Vet J 32: 208-216.