Criteria for Selection of Cases

A computer search of the medical records of all horses admitted to the University of California Veterinary Medical Teaching Hospital between January 1994 and December 2004 with acute cranial injury was performed. Records were included in the study only if head trauma was associated with neurologic signs consistent with acute brain injury and had occurred ≤ 72 hours prior to initial evaluation at the hospital. Horses with bony fractures of the skull and soft tissue or ocular injuries were not included if neurologic dysfunction was not reported in the records.

Medical records were excluded if acute neurologic signs were determined to be a result of spinal cord trauma only. Neonates were excluded if neurologic signs or history was consistent with perinatal asphyxia. Cases were also excluded if neurologic signs were caused by factors other than acute trauma (eg, infectious diseases, temporohyoid osteoarthropathy, intracranial abscess, neoplasia, idiopathic neuritis, bacterial meningitis, or other organ failure).

Procedures

Data extracted from the medical records included age; sex; breed; interval from trauma to initial evaluation; history and type of injury; signs detected by owners and referring veterinarians and physical examination findings at initial evaluation; specific neurologic examination findings; results of diagnostic imaging procedures, CBC, serum biochemical analyses, venous blood gas analysis, and plasma osmolarity assessment; treatment (provided by referring veterinarian or at the hospital); duration of hospitalization; survival (whether horses survived to discharge from the hospital); and long-term follow-up information obtained via telephone contact with owners.

Statistical analysis—Data are presented as median values, mean ± SD, and range. A Mann-Whitney U test was used to compare clinicopathologic variables in horses in the survivor and nonsurvivor groups. These included results of hematologic and serum biochemical analyses, and venous blood gas analysis and lactate concentration assessment. Risk factors for nonsurvival were identified by use of a Fisher exact test and calculation of ORs with 95% confidence intervals. Significance was set at a value of P < 0.05.

Results

Thirty-four horses met the inclusion criteria. The computer search of the medical records initially identified 221 horses with head trauma and neurologic signs. One hundred eighty-seven horses were subsequently excluded from the study for the reasons previously stated (eg, excluded if neurologic signs were determined to be a result of spinal cord trauma only). Neonates were excluded if neurologic signs or history was consistent with perinatal asphyxia. Cases were also excluded if neurologic signs were caused by factors other than acute trauma (eg, infectious diseases, temporohyoid osteoarthropathy, intracranial abscess, neoplasia, idiopathic neuritis, bacterial meningitis, or other organ failure). Median age of horses was 12 months (mean ± SD age, 4.4 ± 5.6 years; range, 3 weeks to 21 years). There were 16 females, 10 geldings, and 8 sexually intact males of various breeds (12 Thoroughbreds, 6 Quarter Horses, 5 Arabians, and 11 horses of other breeds). Estimated interval from injury to initial evaluation at the hospital was 12 hours (mean interval, 21 ± 22 hours; range, 1 to 72 hours).

Fifteen of the 34 (44%) horses had a history of sustaining injury to the poll subsequent to rearing and falling over backwards onto the dorsum during halter training or restraint. Nine of those 15 horses were ≤ 12 months old. Four horses sustained head injury while struggling from entrapment in a fence or subsequent to becoming prone in a stall, 1 horse was injured during recovery from anesthesia, 1 horse ran into a tree, and 1 horse was kicked by another horse. The cause of trauma was unknown for the remaining 12 (35%) horses.

The type of neurologic abnormalities detected in each of the 34 horses was assessed (Table 1). In horses for which the traumatic incident was witnessed, neurologic signs were immediately evident. In horses for which the cause of the traumatic injury was unknown, the interval between occurrence of the traumatic incident and the onset of neurologic signs could not be determined retrospectively from the medical record. Many horses had multiple neurologic deficits such as ataxia, abnormal mentation, and nystagmus, which were the most frequently identified abnormalities (affecting 10/34 [29%] horses). Two horses that were comatose also had mydriatic, nonresponsive pupils; flaccid paralysis; and an irregular breathing pattern, all consistent with tentorial herniation. Vision, pupillary size and symmetry, and pupillary light response were reported to be abnormal in 16 of 34 (47%) horses. Abnormal physical examination findings at initial evaluation included tachycardia (median heart rate, 60 beats/min; mean, 69 ± 25 beats/min; range, 36 to 130 beats/min) in 26 of 32 horses in which heart rate was recorded.

At initial evaluation, clinicopathologic variables were also assessed (Table 2). Abnormalities included moderate neutrophilia and mild lymphopenia in most horses. Band neutrophilia was identified in 7 horses. Moderate hyperglycemia (median, 169 mg/dL; mean, 165 ± 36 mg/dL; range, 106 to 259 mg/dL) and mildly high serum creatine kinase activity (median, 1,302 U/L; mean, 2,859 ± 3,602 U/L; range, 335 to 12,617 U/L) were also detected. Venous blood gas analysis in 12 horses revealed a mildly high blood lactate concentration (median, 2.6 mmol/L; mean, 3.8 ± 4.3 mmol/L; range, 1.0 to 13.4 mmol/L).

Plasma osmolarity of venous blood was reported for 13 horses. One to 17 plasma osmolarity measurements were recorded during hospitalization for each horse. Among those horses in which osmolarity was measured serially and specified as being before (n = 58 measurements) or after (n = 7 measurements) hyperosmolar therapy, median plasma osmolarity was 289 mOsm/L (mean, 291 ± 13 mOsm/L; range, 272 to 332 mOsm/L) and 313 mOsm/L (mean, 318 ± 12 mOsm/L; range, 298 to 331 mOsm/L), respectively.

Comparison of clinicopathologic variables for horses in the survivor (n = 21) and nonsurvivor (13) groups revealed significantly higher PCV at the time of hospital admission among nonsurvivors (mean, 40%) than among survivors (mean, 33%). All other variables were not significantly different between those 2 groups.

The most frequent diagnostic imaging techniques performed, either alone or in combination with other techniques, were survey radiography of the head (31 [91%] horses), followed by upper airway endoscopy (9 [26%]) and CT (4 [12%]). A postmortem diagnosis was determined in 10 of 13 horses that were euthanized.

In 15 of the 34 (44%) horses, fractures of the basilar (basisphenoid and basioccipital) and temporal bones associated with poll impact were identified as either suspected or confirmed by use of radiography or CT or on the basis of findings of postmortem examination. Of these 15 horses, survey radiography revealed a suspected fracture in 9, but none of the fractures was confirmed by this technique (radiography was not performed in 2 horses that were euthanized soon after admission). Six horses that had a clinical diagnosis of basilar bone fracture (on the basis of radiographic evidence of a suspected fracture and supporting clinical signs) survived to be discharged from the hospital.

Bony fractures of the calvarium, including basilar bone fractures, were identified or suspected on the basis of results of radiography (n = 11), endoscopy (1), CT (2), or postmortem examination (8) in 22 of the 34 (65%) horses. Among those 22 horses, a suspected fracture was identified radiographically in only 11 (50%). Twelve of the 22 (55%) horses survived to time of discharge from the hospital. Bacterial meningoencephalitis secondary to fracture and external perforation may have developed in some horses, but was not definitively determined via CSF analysis; therefore, the effect of this complication on survival rate could not be determined.

Among the 34 horses, 11 (32%) did not have a fracture of the cranium identified via any technique (not including 1 horse that was euthanized against recommendations; in this horse, postmortem examination revealed the presence of hemorrhage into the middle ear without a fracture and confirmed the radiographic interpretation). Ten of these 11 horses survived to be discharged from the hospital.

Information regarding treatment, either by the referring veterinarian or at the hospital, was available for all horses. Twenty-five of the 34 (74%) horses were administered dimethyl sulfoxide (1 g/kg [0.45 g/lb] as a 10% solution, IV). Of the 31 (91%) horses that received nonsteroidal anti-inflammatory drugs, 29 were treated with flunixin meglumine (0.25 to 1 mg/kg [0.11 to 0.45 mg/lb], IV or PO, q 8 to 12 h), 6 horses were treated with phenylbutazone (2.2 to 5.0 mg/kg [1 to 2.27 mg/ lb], IV or PO, q 12 h) in addition to flunixin meglumine, and 2 horses received phenylbutazone only (3.0 to 7.5 mg/kg [1.36 to 3.41 mg/lb], IV or PO, q 12 h).

Of the 30 (88%) horses that received glucocorticoids IV, dexamethasone was most frequently administered (n = 19 horses), followed by dexamethasone sodium phosphate (6), prednisolone sodium succinate (7), and methylprednisolone (1). Six horses received more than 1 different formulation of glucocorticoid during treatment. Horses received 1 to 4 doses of corticosteroids during the treatment period. Of the 19 horses that were initially administered dexamethasone by the referring veterinarian, the dose was determined from the medical record for only 14 horses (0.03 to 0.8 mg/kg [0.01 to 0.36 mg/lb], IV).

Antimicrobials were administered to 27 (79%) horses. Treatments included aminoglycoside-penicillin combination (n = 11 horses), ceftiofur (11), and trimethoprim-sulfonamide (12). Nine (26%) horses received treatment with an anticonvulsant (diazepam [n = 6] or phenobarbital [3]). Osmotic or diuretic agents were administered to most horses; 27 (79%) received 20% mannitol, 2 (6%) received hypertonic saline (7% NaCl) solution, and 2 (6%) received furosemide. Supportive treatments consisted of IV administration of crystalloid fluids (29 [85%] horses), vitamin B complex or thiamine (7 [21%]), and an antioxidant (vitamin E; 17 [50%]).

Twenty-one of the 34 (62%) horses survived to be discharged from the hospital. Median duration of hospitalization for survivors was 9 days (mean, 11 ± 7 days; range, 3 to 28 days). Of the 13 horses that did not survive, 11 were euthanized by the attending clinician on the basis of poor prognosis (lack of response to treatment or deteriorating neurologic status), 1 horse died following owners' request not to perform recommended euthanasia, and 1 horse was euthanized against recommendations at the owner's request (this horse was excluded from further statistical analyses in terms of nonsurvival). Nineteen of the 21 (90%) horses had persisting neurologic deficits at the time of discharge.

Long-term follow-up was available for 6 of the 34 (18%) horses. Owners reported that these horses were performing at their level of intended use and believed that the horses had not been compromised by the neurologic injury. One Thoroughbred in which a basilar bone fracture had been identified via CT had returned to racing. One horse had equivocal, residual unilateral visual deficits.

Of the neurologic abnormalities at initial evaluation and clinical diagnoses, risk factors identified to be associated with nonsurvival among horses with TBI included prolonged recumbency (duration extending beyond the initial 4 hours after admission; OR = 18; 95% confidence interval, 3.0 to 107.8; P < 0.001) and fracture of the basilar bones (identified before or after death; OR = 7.5; 95% confidence interval, 1.5 to 37.7; P = 0.01).

Discussion

Poll injury, subsequent to rearing and falling over backwards onto the dorsum during halter training or restraint, was the most common type of head trauma among horses in the present study. The biomechanical aspects of such a maneuver and associated injury to anatomic structures are fairly unique to horses and have been well described.^4,5,7-10 The high proportion of young (≤ 12 months) horses that sustained this type of injury in the present study is consistent with other retrospective reports.^5,6,8,11 Proposed explanations for this age-related increase in susceptibility of young horses to poll injury focus on the reaction or response of untrained horses to restraint applied to the head during training.⁸ Young horses may also be more susceptible to fracture of the basilar bones because the suture between the basisphenoid and basioccipital bones remains open until 2 to 5 years of age.^12–14 In addition, this is the site of insertion of the largest flexor muscle of the neck (rectus capitis ventralis major) that exerts considerable traction forces during head and neck hyperextension at the time of impact.¹⁰ In prior reports,^6,8 male horses were overrepresented among equine patients with head trauma. However, a sex predilection was not identified in the present study.

Neurologic examination findings in horses with TBI in the study of this report were variable. Ataxia was most frequently reported. Proprioceptive deficits are most often associated with spinal cord injury, which was concurrent in 3 of the study horses. Although cerebral injury should not affect gait, damage to the brainstem may result in limb ataxia and weakness. Additional signs of cranial nerve dysfunction are useful in localizing the site of injury to the brainstem.

Damage to the vestibular apparatus was a common finding in the horses with poll injury in the present study. Abnormal mentation, either hyperexcitability or lethargy and obtundation, was reported frequently in these horses. The effect of sedative drugs administered to an unknown number of horses prior to examination may have contributed to the number of horses with abnormal mentation.

Abnormalities of pupillary size, symmetry, and response to light were common findings in this group of horses. Bilateral mydriasis with nonresponsive pupils is suggestive of optic nerve damage, as well as severe midbrain injury with risk of impending tentorial herniation, and is regarded as a poor prognostic sign.⁹ Additional clinical signs included epistaxis, hemorrhage or CSF leakage from the ear, and tachycardia.

Recumbency of > 4 hours' duration following admission to the hospital was reported for 12 (35%) horses and was associated with nonsurvival (OR, 18). To our knowledge, a statistical association between recumbency and nonsurvival has not been previously reported for horses with TBI, although anecdotally, it is considered a poor prognostic indicator in horses.^3,8 The inherent limitations and complications of the management of recumbent adult horses include ongoing selftrauma from struggling; decubital ulceration; reduced ability to eat, drink, urinate, and defecate; development of pneumonia, cystitis, gastrointestinal tract dysfunction, compartmental or ischemic myopathy, and neuropathy; and requirement of considerable labor and financial investment. These factors, as well as ethical considerations, strongly influence the decision for euthanasia of recumbent horses, much more so than for brain-injured small animals and humans. The ability to manage some of the recumbent adult horses in the present study for a long period of time may have resulted in a better outcome, although this is speculative. The development of techniques for lifting¹⁵ and sling support provides additional options for management of recumbent equids; the recent increased availability and use of these techniques may result in improved survival rate in the future.

In the present study, the development of seizures (detected in 7 [20%] horses) was not associated with nonsurvival. It may be difficult to distinguish horses that are recumbent and having continuous episodes of struggling and thrashing from those that have developed true convulsions, particularly in horses with vestibular disease in which incoordination and nystagmus are also present. Adult large animals appear to have a relatively high seizure threshold, suggesting the presence of considerable brain injury when seizures develop.¹⁶ In 6 of the 7 horses with seizures in the present study (the exception was a foal), the seizure event reported by the owner or attending veterinarian was a single episode that occurred prior to admission of the horse to the hospital. A true grand mal seizure as a result of TBI would involve loss of consciousness, in addition to nystagmus or convulsions. Whether a true seizure occurred in any of those horses could not be determined with certainty because of the historical nature of the information; thus, in terms of prognosis, the presence or absence of seizures in horses with TBI should be interpreted with caution.

The abnormal hematologic findings of mild leukocytosis, neutrophilia, and lymphopenia among the horses with TBI were considered to be nonspecific responses to trauma-induced sympathetic stimulation, specific effects of glucocorticoid administration, or a combination of both. Although within the laboratory reference range, PCV was the only clinicopathologic variable that was significantly different between surviving (33%) and nonsurviving (40%) horses. In horses, high PCV can be attributed to both reduced intravascular volume and splenic contraction. Hypotension, even for brief periods, is a well-established independent predictor of death in humans with TBI.^17–20 Although mean arterial pressure was not recorded for horses in the present study, hypotension resulting from hypovolemia may have contributed to nonsurvival in certain horses and might be useful as an important monitoring tool for management of horses with head trauma. The relatively higher PCV in horses in the nonsurvivor group may be evidence for greater splenic contraction or cardiovascular compromise associated with severity of injury, a longer interval between injury and initial evaluation, comparatively less IV fluid administration prior to hospitalization, or simply breed-associated differences. In the present study, breed did not appear to influence PCV in horses in the nonsurvivor group; Thoroughbreds were not an overrepresented breed in this group (30%), compared with the survivor group (38%).

Mild hyperglycemia was reported for the study horses and can be explained by an expected trauma- and stress-induced sympathetic response, the effect of corticosteroid administration, or a combination of both. Plasma glucose concentration was not significantly different between surviving and nonsurviving horses and therefore was not a useful predictor of survival in horses with TBI. However, in humans, there is an association between hyperglycemia and poor outcome in critically ill trauma patients.^21–23 The specific association between hyperglycemia and poor neurologic outcome in humans with TBI is less well understood and has become a topic of recent and ongoing debate in relation to the use of high-dose corticosteroid administration in the treatment of TBI.^24,25,26 Presently, hyperglycemia is believed to contribute to poor outcome in TBI because of its role in secondary brain injury as a result of its damaging effects on vascular endothelium; osmotic effect; and acceleration of anaerobic glycolysis, cellular acidosis, and cell death.²⁷ Whether hyperglycemia actually contributes to secondary brain injury and worsens outcome or is simply a reflection of an appropriate sympathoadrenal response remains to be determined from ongoing randomized clinical trials in humans.²⁸

Venous blood lactate concentration at initial evaluation was recorded for 7 horses and was mildly high, suggestive of hypoperfusion or high muscle anaerobic metabolism, inflammation, myonecrosis, seizures, or surges of circulating catecholamines. Liver failure as a cause of high blood lactate concentration was not detected in the horses in which lactate concentration was measured. Measurements obtained from a much larger number of horses are needed to determine the prognostic value of blood lactate concentration in horses with TBI.

Conventional survey radiography of the head was the most commonly used diagnostic method in the study of this report. However, subtle nondisplaced fractures and soft tissue damage may not be easily detected radiographically.^7,10 Findings from the present study highlight the limitations of radiography for evaluation of intracranial trauma. Among the 22 horses with fractures of the calvarium, only 50% were identified via survey radiography; the remainder required either CT imaging or postmortem examination for detection of such fractures. Basilar bone fractures, which were identified via radiography in only 60% of horses in which they were ultimately identified via CT or during postmortem examination, are a particular diagnostic challenge. The neurologic status (loss of balance, compulsive circling, and head tilt) of horses with basilar bone fractures may preclude diagnostic radiography. Interpretation of radiographic findings may be difficult because of the incompletely fused spheno-occipital suture in young horses. Considerable displacement of these bones has to occur before a definitive diagnosis of fracture can be made radiographically.¹⁰ In the 9 horses in which interpretation of radiographic views of the skull was highly suggestive of a (suspected) basilar bone fracture, separation of or a gap between these bones was present; however, minimal to no obvious displacement was evident.

Horses in which basilar bone fractures were diagnosed were 7.5 times as likely not to survive as horses without this type of fracture, which is consistent with findings of other investigations.^4,6,8 This association emphasizes the importance of making a definitive diagnosis by use of CT. In horses for which CT is impractical or unavailable, a presumptive diagnosis can often be made on the basis of other supporting radiographic evidence (soft tissue obliteration of the auditory tube diverticula, ventral deviation of the dorsal pharyngeal wall, or irregular bony fragment ventral to the base of the skull) in conjunction with supporting historical and neurologic examination findings.¹⁰ Of interest is the lack of association between the diagnosis of calvarium fractures (inclusive of all types of fractures) and nonsurvival of horses with TBI in the present study. Furthermore, hemorrhage or CSF drainage from the external portion of the ear (a finding consistent with skull base fractures) was also not associated with nonsurvival. As stated previously, these findings should be interpreted with caution because of the small number of horses in the present study and limited long-term follow-up information.

The treatment of head trauma in horses has been reviewed,^{7–9,16,29,30} but to our knowledge, there are no published retrospective or prospective clinical studies to evaluate or support the currently used treatments, which have essentially remained unchanged over the past few decades. Thus, recommendations for treatment of head trauma in horses are based on experiences in small animal and human medicine and have common, goal-directed strategies to prevent or minimize secondary brain injury. Management of horses in the present study included a combination of supportive care; osmotic-diuretic, anti-inflammatory, analgesic, antimicrobial, and antioxidant treatments; and seizure control. Associations between treatments and nonsurvival were not performed because of the lack of controlled evidence and the retrospective nature of the study.

Some recent investigations^31–34 in the management of TBI in humans may have clinical application to horses. Intravenous fluid therapy with isotonic crystalloid solutions was administered to most (85%) of the horses in the present study. Historically, fluid restriction has been used in the management of brain injury to minimize cerebral edema formation. However, current understanding supports treatment and prevention of hypotension because of the negative effect of hypotension on cerebral perfusion pressure and its contribution to worsened outcome in brain-injured humans. Therefore, early, effective treatment of hypotension and maintenance of normal fluid balance should be the goals of fluid therapy in horses with TBI to improve outcome.⁹ Intracranial pressure monitoring, which is performed in humans with severe brain injury, is not presently used in horses on a routine basis. However, direct and indirect blood pressure monitoring (the former used in neonatal foals), in addition to assessment of blood lactate concentration, central venous pressure, urine output, and physical examination variables, are useful monitoring tools to avoid development of hypotension and optimize fluid balance in horses following head trauma.

Hyperosmolar therapy was administered to 29 of the 34 (85%) horses in the present study. Osmotic therapy with mannitol or hypertonic saline solution is effective in the treatment of high ICP in humans. Hyperosmolar solutions decrease ICP by decreasing Hct and blood viscosity through plasma expansion and by reducing cerebral edema through osmotic effects.³⁵ Because ICP is not monitored, hyperosmolar therapy is indicated in equids with TBI in the presence of altered mentation (obtundation, stupor, or coma), papillary edema or retinal detachment visible during fundic examination, or progressive neurologic deterioration and only after fluid resuscitation and euvolemia have been achieved.⁷ Among the 29 horses receiving hyperosmolar therapy in the present study, mannitol was administered to 27 (93%), whereas the remaining 2 (7%) horses received hypertonic saline solution. Mannitol has become a cornerstone in the treatment of humans with severe head injury and increased ICP.³⁵ Mannitol has antioxidant effects as well as plasma-expanding and osmotic effects. Potential adverse effects include hypovolemia as a result of diuresis, acute renal failure, rebound increases in ICP, and blood-brain barrier disruption with chronic or high-dose administration.^35–37 Hypertonic saline solution is comparable to mannitol in terms of its rheologic and osmotic effects for reducing ICP and cerebral water content.^38,39 However, hypertonic saline solution has been suggested to be superior to mannitol in humans because it augments intravascular volume with administration of smaller volumes; has less risk of causing hypovolemia; and may have immunomodulatory, neurochemical, and cell-membrane electrochemical stabilizing effects.^35,36 In addition, with repeated administration, the efficacy of mannitol wanes, whereas hypertonic saline solution continues to be effective—hence, the recommendation for continuous rate infusion in humans.⁴⁰ Potential adverse effects include rebound increases in ICP, central pontine myelinolysis, coagulopathy, and electrolyte and acid-base derangements. Another theoretical justification for administration of hypertonic saline solution is that an intact blood-brain barrier is less permeable to saline solution than to mannitol,^35,41 thereby potentially reducing the risk of rebound increases in ICP. Results of studies^42,43 have also indicated that hypertonic saline solution is effective in reducing high ICP that is refractory to mannitol administration. When combined with a colloid (6% dextran), 7.5% saline solution reduced ICP more effectively than mannitol in humans with TBI.⁴⁴

Measurement of plasma osmolarity may aid in application of hyperosmolar therapy. In the present study, plasma osmolarity was measured in 13 horses that had received prior administration of mannitol or hypertonic saline solution; the purpose of monitoring osmolarity was to ensure that plasma osmolarity was < 320 mOsm/L prior to subsequent administration of a hyperosmotic agent, as recommended in human medicine guidelines.³⁶ To be effective, establishment of an osmotic gradient of at least 10 mOsmol/L across the blood-brain barrier is believed to be optimal.⁴⁵ Therefore, immediate posttreatment measurement of plasma osmolarity is useful to assess whether response to treatment is appropriate and detect excess or persistent increases in osmolarity that should be avoided. Although only 7 posttreatment osmolarity measurements were recorded in the medical records, results indicated that osmolarity did increase and approached values > 320 mOsm/L in some horses.

Administration of high doses of glucocorticoids to treat acute neurologic trauma in horses is widely used, and glucocorticoids were administered to 30 (88%) of the horses in the present study. The theoretical benefits of glucocorticoids for treatment of nervous system trauma include anti-inflammatory and oxygen free-radical scavenging properties and improved neurologic outcome in spinal cord injury.^46,47 However, clinical trials to assess glucocorticoid administration in humans with severe TBI have failed to identify a significant beneficial effect^48,49 and, recently, have revealed no improvement in long-term outcome and a small increased risk of death within the 2 weeks following injury.^50,51 Proposed detrimental effects of corticosteroid treatment are related to the development of hyperglycemia and its contribution to secondary brain injury during ischemic and hypoxic conditions.²⁶ Immunosuppression may also play a role in the negative outcomes. Consequently, routine administration of corticosteroids in humans with TBI is not currently recommended. Given the data that are available from human medicine, the use of corticosteroids in horses with TBI should be judicious until further studied.

Results of the present study have indicated that the overall short-term survival rate for horses with TBI is 62%. With the exception of 2 horses, all study horses that did not survive were humanely euthanized on the basis of the attending clinician's opinion of poor prognosis. Because of the retrospective nature of the study, it is difficult to determine how much, if any, financial constraints or other factors influenced the decision for euthanasia in these horses. Findings from the present study have suggested that horses with acute TBI for which supportive care and specific treatment can be provided have a reasonable prognosis for survival. On the basis of the limited number of cases available for long-term follow-up assessment, return of horses with TBI to prior activity levels is achievable.