Headshaking in horses is defined as a syndrome of involuntary or paroxysmal shaking of the head with no noticeable external stimulus, suspected to be mediated by trigeminal neuralgia either through an idiopathic or photic pathway.[1-5] Despite extensive investigation, no specific etiology has been identified and postmortem examinations of headshakers have found no physical abnormalities in the trigeminal nerves. Current medical and surgical treatments for headshaking have high failure and complication rates.[4-11]
Practitioners frequently prescribe a trial of nonsteroidal anti-inflammatory medications or corticosteroids during the initial diagnosis and treatment of headshakers. In laboratory animal models, there are limited data to suggest that dexamethasone can attenuate some signs of neuropathic pain.[12, 13] Although a single 30 mg intramuscular dose of dexamethasone did not reduce clinical signs of headshaking in 9 horses, pulsatile dexamethasone treatment has been used with anecdotal success in multiple horses. We hypothesized that high-dose intermittent corticosteroid treatment would reduce clinical signs in horses with idiopathic headshaking syndrome.
- Top of page
Before initiation, approval was obtained from the University of Pennsylvania's Privately Owned Animal Protocol Review Board. The study design was a blinded, placebo-controlled, randomized clinical trial using a convenience sample of healthy, adult, client-owned horses that had received a diagnosis of equine idiopathic headshaking syndrome from their referring veterinarian based on clinical examination and signs consistent with those described in the veterinary literature.[1, 6, 7] Horses remained on their farms throughout the study period. Each horse was randomly assigned via paper lottery to a treatment group that received compounded dexamethasone paste1 (60 mg PO daily for 4 days every 21 days for 4 pulses) or a placebo1 group (vehicle paste only, same regimen). The dexamethasone content of the paste was validated by a commercial laboratory2 and found to match the reported concentration to within 10%.
Owners were asked to keep a diary ≥3 times per week and record a headshaking severity score (HS; Table 1) as well as pertinent data on type of work performed, the weather, concurrent medications, or use of devices such as a fly mask or nose net. Owners were asked to evaluate their horses under the same conditions every time to the extent possible. This meant that they were expected to continue with any therapeutic interventions that they currently employed (medications or devices) and were supposed to have the horse perform the same level of exercise in the same location each time. Owners were asked to have a minimum of 4 weeks of evaluation recorded before the first dose of medication.
Table 1. Headshaking scoring system.
|1||<5 shakes per 30 minutes|
|2||5–10 shakes per 30 minutes|
|3||10–30 shakes per 30 minutes|
|4||Almost constant; unrideable, unworkable|
All baseline (pretreatment) scores were used in analysis. Only HS from the 14 days after and including the first dosage of each 4-day pulse were included. The 14-day period was chosen as the most likely to give a reasonable number of data points while avoiding most effects of season change. The percent reduction in median HS between the baseline and each treatment pulse was determined as follows: [(baseline − treatment)/baseline]*100. The proportion of horses with >0, >50, and >70% improvement were compared between groups by Fisher's exact test.
Proportional odds ordinal logistic regression was used to evaluate the effect of dexamethasone treatment on HS.3 Group (treatment versus placebo), period (baseline, pulses 1–4), sun (cloudy, mixed, sunny), weather (no wind/rain, wind, rain), device (no device, face mask, nose net), and exercise (none/walk in hand, lunge, ride at walk, ride at trot/canter) were included as covariates, with HS as the ordinal outcome variable. The baseline or reference points were set as placebo, baseline period, cloudy, no wind/rain, no device, and none/walk in hand for exercise. All data were clustered on horse.
Proportional odds logistic regression yields an odds ratio as an output. In this case, the odds ratio describes the relative odds of being in a higher outcome level than a lower level and applies for any intensity region. For example, if the odds ratio for the treated group versus the placebo group is 2, then horses in the treated group have twice the odds of having a HS of 1–4 versus 0 as horses in the placebo group. This same model applies across all outcome levels, so horses in the treated group also have twice the odds of having a HS of 2–4 versus 0–1, or 3–4 versus 0–2, etc.
- Top of page
Twenty horses were entered in the study. Seven horses had no diary returned or incomplete diaries and were excluded from analysis. The most common reason for exclusion was that owners did not include a baseline diary, 1 owner did not record any scores, and another never returned the diaries. One additional horse was discontinued from the study after it developed high fevers and laminitis during the first pulse of dexamethasone and was diagnosed with Potomac horse fever via PCR by the attending veterinarian. Twelve horses completed the study; serendipitously there were equal numbers of placebo horses (n = 6) and treated horses (n = 6) with complete diaries that covered baseline and at least 3 treatment periods.
After each pulse, the percent change in median HS in the placebo and treated groups were Pulse 1 placebo −50 to 50%, treated −33 to 100%; Pulse 2 placebo −50 to 88%, treated −17 to 100%; Pulse 3 placebo −50 to 100%, treated 0 to 100%. Both groups had a significant (P = .042) decline in HS over time (Fig 1). At each pulse, the proportion of horses with percent improvement from baseline >0, >50, or >70% was not different between placebo and treated groups (P = .180–1.00; Table 2).
Table 2. Improvement in median headshaking score after pulsatile treatment with 60 mg PO dexamethasone or placebo for 4 days. The number of horses in the placebo:treated groups with improvements in headshaking scores >0, 50, or 70% from baseline at each pulse period. There were 6 horses total in each group. Reported P-values are based on Fisher's exact test.
|Pulse Period||>0% Improvement||>50% Improvement||>70% Improvement|
|1||3: 4 (P = 1.0)||1: 3 (P = .55)||0: 3 (P = .18)|
|2||2: 4 (P = .57)||2: 3 (P = 1.0)||1: 3 (P = .55)|
|3||3: 4 (P = 1.0)||3: 2 (P = 1.0)||1: 2 (P = 1.0)|
Figure 1. Average owner-recorded headshaking scores (0–4, with 4 being the most severe) ± 95% CI for horses with idiopathic headshaking syndrome treated with pulses of dexamethasone (60 mg PO QD × 4 days, q3 weeks for 4 months) or placebo (inert paste). Once trigger factors (weather, exercise) or devices (face mask) were accounted for, there was no significant effect of treatment.
Download figure to PowerPoint
The proportional odds logistic regression model had good fit to the data (P < .001). There was no effect of treatment with dexamethasone on HS (OR, 1.01; 95% CI, 0.34–3.03; P = .987). In both groups, HS in period 1 were not different from baseline; however HS in periods 2–4 were significantly lower than the baseline scores (P ≤ .020). Sunny days were associated with significantly higher HS than cloudy days (OR, 2.59; 95% CI, 1.54–4.36; P < .001), and days with mixed sun and clouds had a tendency to higher HS than completely cloudy days (OR, 1.74; 95% CI, 0.91–3.32; P = .095). Wind but not rain was associated with significantly higher HS (OR, 1.8; 95% CI, 1.07–3.05; P = .028). Horses that wore face masks had significantly higher headshaking scores than horses that did not wear any device (OR, 2.95; 95% CI, 1.15–7.50; P = .023). Few horses wore nose nets, and therefore these were excluded from analysis. All levels of work were associated with significantly higher HS than no exercise or hand walking. The odds of having a higher score was highest associated with lunging (OR, 9.4; 95% CI, −3.0–29.5; P < .001), followed by riding at the walk (OR, 3.2; 95% CI, 1.03–10.0; P = .045), then riding at trot or canter (OR, 2.66; 95% CI, 1.37–5.15; P = .004).
- Top of page
This study found no significant reduction in idiopathic headshaking after treatment with dexamethasone, confirming previous findings of limited response to dexamethasone treatment in headshaking horses.[7, 8] In a rodent model of trigeminal neuralgia, rats responded favorably to treatment only if dexamethasone treatment was initiated on the day after nerve trauma, whereas in horses identified as idiopathic headshakers the duration of signs is generally weeks to months before treatment. The dose of dexamethasone used in the rat model was 25 mg/kg body weight, a level considered unsafe in horses because of the perceived risk of laminitis. As a result, a much lower dose of approximately 0.1 mg/kg was used in this study. The effect of time and dose might have contributed to treatment failure, as well as that the scoring system might have failed to evaluate some less typical manifestations of this condition such as rotary head movements, face rubbing or nostril clamping. There is also ongoing debate in the literature regarding the efficacy of glucocorticoids for the treatment or prevention of neuropathic pain as there is some evidence that glucocorticoid agonists might actually potentiate neuropathic pain by promoting upregulation of N-methyl-D-aspartate receptors in the spinal cord. Additionally, corticosteroids have multiple significant adverse effects in mammals in general, and horses specifically. Their use has been associated with decreased immune function and thus an increased risk of infection, gastrointestinal and renal adverse effects, and possibly also laminitis, especially at the higher doses used in this study. No clinically evident adverse effects were seen in any treated horse. Although 1 horse receiving the active compound did develop laminitis, the causative relationship was confounded by infection with Potomac horse fever. Whether the corticosteroid regimen made this horse more susceptible to infection remains speculative.
Despite the lack of effect of dexamethasone, this study confirmed the effect of previously reported trigger factors. Owners and veterinarians have commonly reported that horses with idiopathic headshaking syndrome appear to be worse in spring or summer season, bright sunlight, wind or rain, and under increased exercise levels, perceptions borne out by this investigation. The reduction in HS observed in both groups with each period is most likely a reflection of the seasonality of the disease as recording began in the late spring and continued into the fall. Not all horses have the same season of headshaking, however, and a single horse that was reported to have a winter season by the owner did show increasing HS throughout the study period. Interpretation of the increased HS for horses wearing face masks was confounded by the fact that this intervention was not randomized and was likely applied by owners in cases where the horse was showing worse signs or on days when the headshaking behavior was worse. The increased HS therefore more likely represents a failure of the intervention rather than an exacerbation of signs by the intervention, but further study would be required to elucidate the true relationship. Work level was the trigger factor most strongly associated with HS. As with the interpretation of the face mask, the precise influence of work level on HS is complicated by the fact that the work levels were not standardized or randomized, but were chosen by the owner. For instance, on days when horses had high HS on the lunge line, they were not ridden by owners, thus falsely lowering the assessments of HS under riding conditions. Regardless of the precise effects, this study confirms that exercise exacerbates signs of headshaking in horses. The mechanism for this is not known, but might be influenced by changes in autonomic nervous tone or local hemodynamics affecting the trigeminal nerve.
In this study, no evidence was obtained to support the use of dexamethasone treatment in cases of idiopathic headshaking syndrome in horses. Prior anecdotal evidence that suggested efficacy might have one or several explanations. First, seasonal reduction in headshaking might have been falsely attributed to dexamethasone; second, there might have been a degree of placebo effect regarding owner interpretation of signs; or third, there might be a subset of horses that exhibit headshaking behaviors that have an etiology responsive to corticosteroids. Nonetheless, this study was limited by small sample size, the inconsistencies inherent with field trials and by the difficulties of evaluating a disease with many trigger factors. A more controlled study with greater case numbers, closer oversight of participating owners to minimize variations in trigger factor exposure, or direct veterinary evaluation would strengthen results in future investigations and perhaps resolve the questionable efficacy of corticosteroids against headshaking in horses.