Clinical manifestations

Presentation and course

The clinical examination of a patient with severe brain injury is similar to the evaluation of a comatose patient. Following evaluation and stabilization of airway, breathing, and circulation, attention is focused on the neurologic examination. The Glasgow Coma Scale score is a quick and reliable test of neurologic function. The Glasgow Coma Scale scores eye-opening, motor response, and verbal output to obtain a score between 3 and 15. Intubated patients are assigned a verbal score of 3. In addition to the Glasgow Coma Scale, several tests must be performed on each patient. These tests include:

Prognosis and complications

The following section is adapted from 65 (65).

Prognosis. Despite the advances made in understanding the pathophysiology, neuroimaging, and intensive care management of severe head injury, morbidity and mortality remain high. Jennett and colleagues reported in 1979 that approximately half of patients with severe head injuries died (54). By 1991, the results of the Traumatic Coma Data Bank demonstrated a 35% improvement in the mortality rate of patients with severe head injuries, with over 40% of patients having a good recovery (68). Spain and colleagues described 133 patients treated between 1994 and 1996 with severe traumatic brain injury. Overall mortality was 18% (99). In an international trial of an excitatory amino acid inhibitor, 427 patients were enrolled. Of those patients, 409 were included in the final, 6-month data analysis. Overall, 23% of patients died by the 6-month follow-up. Fifty-seven percent of patients had favorable outcomes (72).

Aggressive treatment of the comatose patient remains, with few exceptions, the standard. Tien and colleagues examined patients with a Glasgow Coma Scale score of 3 and compared patients with at least one reactive pupil to those with bilaterally fixed and dilated pupils. They found 100% mortality in the group with bilateral fixed pupils but a 58% survival rate in patients with at least one reactive pupil (103).

Chamoun and associates examined a cohort of 189 patients presenting with a Glasgow Coma Scale score of 3 (18). The overall mortality rate was 49%, with 80% mortality in patients with bilateral fixed and dilated pupils. However, only 13% of the population had a favorable outcome at 6 months. Younger age, lower intracranial pressure, and reactive pupils served as statistically and significantly favorable prognostic factors.

Mauritz and colleagues specifically looked at prognosis in patients with a Glasgow Coma Scale score of 3 on presentation, with bilateral fixed and dilated pupils (69). Eight of 92 (8.7%) individuals in a large trauma cohort who met these criteria had a favorable 12-month outcome. The same group also reported that age over 65, although a poor prognostic factor, does not inevitably lead to an unfavorable outcome (14).

Wang and associates added another prognosis variable in traumatic brain injury. In addition to traditional prognostic factors, such as age, GCS score, use of anticoagulants, and presence of traumatic subarachnoid hemorrhage, they found that the fibrosis-4 score, a marker for baseline hepatic function, was an independent predictor of mortality in a multivariate model (109). Kunapaisal and colleagues examined admission platelet count as a marker of coagulopathy and a potential prognostic factor after severe traumatic brain injury (62). A retrospective study of over 44,000 patients found that patients older than 50 years of age have lower admission platelet counts, which was independently associated with greater mortality after traumatic brain injury.

The MRC CRASH Trial accrued a validated cohort of 8509 patients and attempted to develop a model for outcome prediction in this cohort. Factors influencing outcome included age, Glasgow Coma Scale score, pupil reactivity, and the presence of a major extracranial injury in a basic model; an additional CT model included significant factors such as petechial hemorrhage, subarachnoid hemorrhage, midline shift, nonevacuated hematoma, and obliteration of cisterns. The study was designed to differentiate between injuries in low- and high-income populations (73). Steyerberg and colleagues used criteria obtained from 11 trials but used the same cohort to validate differences in 6-month outcomes based on age, Glasgow Coma Scale motor score, papillary reactivity, hypoxia, hypotension, low hemoglobin, high glucose, and CT findings, including traumatic subarachnoid hemorrhage (100).

Complications. Complications include all sequelae of prolonged coma and recumbency and systemic trauma, including systemic infection, deep venous thrombosis, electrolyte abnormalities, fat embolism syndrome, stress ulceration, and anemia. Meningitis is common in patients with skull base fractures leading to dural tears. All patients must be monitored for signs of rhinorrhea and otorrhea. Postoperative patients, especially those whose original wound was dirty, should also be monitored for signs of infection. Hydrocephalus may develop in patients with subarachnoid or intraventricular hemorrhage.

Clinical vignette

A 14-year-old male was struck by a car while riding a bicycle. He had a Glasgow Coma Scale score of E4M6V3=13 at the scene but then suffered a seizure and was brought to the hospital. In our emergency room, he was found to have equal round and reactive pupils and was Glasgow Coma Scale E3M6V3=12. CT scan showed evidence of subarachnoid hemorrhage, a small parafalcine subdural hemorrhage, and brainstem swelling.

Approximately 18 hours after the injury, he suffered a respiratory arrest and was resuscitated and intubated. His Glasgow Coma Scale was E1M4VT=5T. CT was unchanged, and an intracranial pressure monitor was placed. Initial pressure was 20 Torricelli units. The patient was treated with intermittent doses of mannitol, 25 g propofol every 4 hours, and a paralytic agent. The pressure remained less than 25, with cerebral perfusion pressure greater than 70 for 48 hours. On the morning of postinjury day 3, intracranial pressure was noted to be elevated to the low thirties. CT scan showed worsened brain swelling as well as a left frontal contusion. The pupils were unequal, with the left pupil dilated but sluggishly reactive.

The patient was taken emergently to the operating room, where a bifrontal craniectomy with bilateral subtemporal decompressions was performed. An expansive duraplasty and suture ligation of the sagittal sinus and division of the anterior falx were also performed. Postoperative CT scan showed the extent of decompression. The patient was noted after surgery to be briskly localizing with all extremities, and the pupils became equal and reactive. The patient was extubated on postoperative day 6 and Glasgow Coma Scale 15 on postoperative day 8. The patient required cognitive rehabilitation but had no neurologic deficits other than memory changes.

Biological basis

Etiology and pathogenesis

Severe head injuries are preventable in most cases. Ethanol intoxication is involved in almost half of all severe traumatic brain injuries in the United States. Illegal drug intoxication is also involved in a significant number of traumatic brain injuries.

The following is adapted from Graham (40):

Pathophysiology. The pathophysiology of severe head injury involves well-defined radiographic patterns of injury.

Epidural hematoma. Epidural hematoma is most often caused by injury to the middle meningeal artery after it exits the foramen spinosum and runs beneath the temporal bone and the sphenoid wing. The blood clot forms between the dura and the skull and pushes in towards the brain.

Subdural hematoma. Subdural hematoma usually occurs due to tearing of bridging veins between the dura and the brain. Small arteries on the brain’s surface may be causative factors as well. Blood layers along the surface of the brain, compressing the ipsilateral hemisphere.

Intracerebral hematoma. Hematoma formation occurs with direct trauma to intracranial vessels. These changes often occur at the frontal and temporal poles. Delayed hemorrhage may result from coagulopathy.

Contusion. A hemorrhagic transformation is characterized by venous injury, localized cerebral edema, and possible increased intracranial pressure.

Diffuse axonal injury. A distinct pattern of white matter damage is characterized radiographically by punctate lesions throughout the white matter. The corpus callosum is typically involved. Often, there is associated intraventricular hemorrhage. Brainstem involvement may lead to especially poor prognosis despite relatively normal intracranial pressure measurements. Damage in the brainstem typically involves areas of the dorsolateral quadrant of the rostral brainstem. The pathologic findings include axonal bulbs that appear as axonal swellings on axons running in one direction.

Graham and associates used diffusion tensor imaging to examine the brains of 55 patients who had suffered diffuse axonal injury versus 19 controls (41). They found a strong correlation between diffuse axonal injury and axonal injury and subsequent neurodegeneration. High rates of atrophy were seen specifically in central white matter tracts like the corpus callosum.

Direct vascular injury. Vascular injury after closed head injury usually involves one of three mechanisms: (1) traumatic pseudoaneurysm, (2) carotid-cavernous fistula, or (3) traumatic dissection. Although the former two diagnoses are usually delayed, the latter diagnosis should be considered in any unexplained neurologic deficit after a closed head injury. Angiography is the test of choice for carotid or vertebral artery dissection.

Cerebral blood flow and head injury. After a closed head injury, regional and global cerebral blood flow changes may affect the neurologic outcome. Cerebral blood flow is proportional to the cerebral perfusion pressure times the fourth power of the diameter of the vessel. The two most common methods for assessing cerebral blood flow after trauma are the 133Xenon radiographic technique and jugular venous saturation.

Muizelaar reported a series of 92 patients who underwent early 133Xenon monitoring after traumatic brain injury. Areas of focal ischemia (cerebral blood flow less than 18 mL/100 g per minute) were seen in 41% of patients with subdural hematoma and 50% of individuals with diffuse swelling, compared to no ischemia in patients with epidural hematoma or normal CT scans. Regional ischemia predicted poorer late outcomes. Cerebral blood flow was lowest within 4 to 6 hours of injury (average 22.5 ± 5.2 mL/100 g per minute), gradually reaching a baseline by 12 to 18 hours after injury. Muizelaar reasons that using hyperventilation to reduce intracranial pressure will also reduce cerebral blood flow by almost 60%, thereby worsening both regional and global cerebral ischemia (74).

Jugular venous saturation may also provide an estimate of cerebral blood flow. A fiberoptic catheter is introduced in a retrograde fashion into the jugular bulb. A low jugular venous saturation correlates with a high oxygen extraction and ischemia. Woodman and Robertson examined episodes of jugular venous desaturation (jugular venous saturation less than 50%, more than 10 minutes) in patients after traumatic brain injury (113). Severe disability, vegetative outcome, or death occurred in 91% of individuals with multiple episodes of ischemia versus 55% of individuals with no episodes of desaturation.

Biochemical alterations in traumatic brain injury. A complete discussion of biochemical abnormalities in traumatic brain injury is beyond the scope of this text. Neuronal injury from trauma is similar in most respects to injury from other forms of ischemia. Consequently, the same pharmacological interventions used to combat ischemic disease have been tested in preclinical and clinical trials of traumatic brain injury. In traumatic injury, a degree of primary damage from the injury is refractory to any means of neuroprotection. However, a penumbra of tissue may be responsive to neuroprotective agents (70).

Several groups have attempted to quantify the injury severity or the potential for secondary injury using either microdialysis catheters placed directly into the brain or measuring either serum or spinal fluid markers of tissue injury. Rhodes and colleagues demonstrated higher levels of serum cytokines interleukin-8 and monocyte chemoattractant protein-1 in patients with more severe brain injury (81). However, changes in cytokine concentration were not predictive of clinical deterioration. Marklund and colleagues demonstrated via cranial microdialysis catheters elevation of tau and beta-amyloid after head injury, with different patterns signifying focal versus a diffuse axonal pattern of injury (66). Semple and colleagues examined the role of chemokine CC ligand-2 (CCL2), a monocyte chemoattractant, in both the CSF of patients with brain injury and a mouse knockout model (91). Their study found the presence of CCL2 in patients after traumatic head injury. They found that the knockout mice showed identical injuries to the wild type in the first week but significantly reduced delayed tissue damage over 2 to 4 weeks, indicating a possible role of CCL2 in posttraumatic secondary tissue damage.

Epidemiology

Trauma causes approximately 150,000 deaths per year in the United States. About half of these injuries are the result of brain injuries. Young male adults are preferentially affected. The most common cause of severe traumatic brain injury is motor vehicle accidents. Assaults, falls, penetrating injuries, and self-inflicted injuries comprise the majority of other injuries (60). Holmes and colleagues have presented an extensive analysis of head injury epidemiology. This group accrued 13,728 patients in a prospective multi-institutional study and examined every trauma patient receiving a CT scan in the emergency department of 21 trauma centers. Logistical regression demonstrated an increased risk of head injury with male sex (65%, relative risk 1.27), age over 65 years (relative risk 1.59), and age under 10 years (relative risk 1.44). Overall, 8.7% of the cohort was determined to have significant radiographic abnormalities on CT scan (47).

Sorani and colleagues examined the role of race and ethnicity in an urban trauma population and concluded that weak trends for increased mortality in African-American and Asian patients were present (98). Shandro and colleagues examined ethanol usage in regard to outcome in a large trauma cohort and found no influence of admission ethanol levels on both long- and short-term mortality (92).

Investigators have examined specific preinjury pharmacological treatments as outcome predictors in traumatic brain injury. Smith and Weeks looked at what would seem to be a clear-cut adverse predictor of poor outcome, namely using anticoagulants at the time of head injury (97). Especially in older individuals, the use of warfarin or similar anticoagulants and antiplatelet agents would be assumed to result in poorer outcomes Still, the data are not conclusive, with some data showing up to a five or six times higher likelihood of morbidity and mortality. Other authors suggest such effects are related to the degree of anticoagulation not simply the use, or that, alternatively, the negative effect centered on patient age and use of anticoagulants was not an independent predictor of poor outcome but instead was dependent on age.

Antihypertensive agents have also been evaluated in terms of protective or detrimental effects. Mohseni and associates examined a cohort of 662 patients suffering acute traumatic brain injury, of whom 25% were taking beta blockers before their event (71). Logistic regression analysis demonstrated that the absence of beta blockers was an independent predictor of poor outcomes with a two-fold increased risk of mortality. Conversely, Catapano and colleagues looked at a similar number of patients with severe traumatic brain injury not taking beta blockers who were on angiotensin-converting enzyme inhibitors and found in logistic regression analysis a three-fold increased risk of mortality (17).

Prevention

Legislative attempts to limit the incidence of traumatic brain injury have centered on motor vehicle safety. Specifically, mandatory seat belts and air bags, reasonable speed limits, and violence prevention programs have been instituted to try to reduce the number of individuals affected.

Differential diagnosis

Rarely, a patient will feign severe brain injury. Noxious stimuli such as cold calorics or sternal rub may help expose these individuals. A non-comatose patient will not allow his arm to hit his head if it is lifted in the air and released. Often, traumatic brain injury must be differentiated from ethanol intoxication.

Diagnostic workup

Diagnostic workup of traumatic brain injury includes angiography, laboratory tests for coagulopathy, ethanol levels, and other drug intoxication, and routine blood chemistry. Coagulation studies are particularly important in patients who may require surgical intervention.

Radiographic workup specifically addressing the central nervous system includes plain x-rays (AP, lateral, and open mouth odontoid) of the cervical spine to reveal injury to the cervicothoracic junction; CT scan of the brain to include bone, brain, and blood windows; and after the primary workup, plain films of the thoracic and lumbar spine. In patients with traumatic brain injury and minimal evidence of damage on CT scan, MRI scans may help to reveal focal injuries to the brainstem and supratentorial white matter, suggesting diffuse axonal injury. Angiography is necessary when vessel dissection, traumatic aneurysm formation, or carotid-cavernous fistula is suspected. Many centers routinely perform follow-up CT scans in patients within 24 hours of injury or admission. Velmahos and associates examined this practice and determined that in a 1-year sample of 179 consecutive trauma patients who underwent routine repeat head CT after brain injury, 37 (21%) demonstrated radiographic worsening, in which seven (4%) required intervention. Each of those seven patients also demonstrated clinical deterioration, leading to the conclusion that repeat CT scanning is only needed in the face of clinical deterioration (106).

The physiological monitoring of patients suffering from severe traumatic brain injury (Glasgow Coma Scale less than 8) has traditionally consisted of intracranial pressure, oxygen saturation, and hemodynamic monitoring. Neurologic surgeons and other physicians tending to patients with head injuries employ these values to guide therapy and determine prognosis. Somatosensory evoked potentials have become a tool to assist in prognosticating outcomes in severe traumatic brain injury.

Somatosensory evoked potentials are inexpensive and less susceptible to sedation than electroencephalography (36). Absent bilateral cortical responses on initial examination reliably predict death, and normal responses predict good clinical outcomes in prospective trials. Improvement in serial somatosensory evoked potentials may precede improvements in the clinical exam, with emphasis on the amplitude and cortical conduction time of the evoked potentials (23). Median somatosensory evoked potential values have also been shown to predict the degree of diffuse axonal injury sustained by patients with brain injury (59).

It should be emphasized that somatosensory evoked potentials should be accompanied by monitoring of intracranial pressure, oxygen saturation, and hemodynamic permanents, as outlined in the Joint Section on Neurotrauma and Critical Care guidelines (06).

In the acute phase setting, biomarkers may help predict the presence or severity of traumatic brain injury. Bazarian and associates completed a prospective trial of biomarkers in patients with traumatic brain injury in a multicenter trial at 22 centers (ALERT-TBI) (07; 08). Immediate plasma concentrations of glial fibrillary acidic protein (GFAP) and ubiquitin carboxyl-terminal hydrolase L1 (UCH-L1) were used in individuals with nonpenetrating mild to moderate closed head injury, with levels correlating to the presence of abnormality on CT scan. The negative predictive value for individuals with negative testing was quite high, with only three of 1959 patients exhibiting a positive CAT scan with negative test values. Similarly, positive values correlated well with the presence of pathology on CT, although the primary value was determining which patients did not require imaging.

Another trial looked at different biomarkers and found that GFAP had better predictive value than UCH-L1 and calcium-binding protein beta at 12 to 32 hours post-injury (64). Chihi and colleagues specifically looked at the role of brain natriuretic peptide and found an association between elevated levels and postoperative seizures in patients with acute subdural hematoma (19). The use of biomarkers at presentation may eventually predict the need for additional imaging or intervention as well as prognosis.

Management

Management strategies will be discussed in the format of the American Association of Neurological Surgeons guide for managing severe head injury.

Resuscitation of blood pressure and oxygenation. All traumatic brain injury victims should be managed according to advanced trauma life support protocols. Airway, breathing, and circulation should be addressed in the initial evaluation. Systemic hypotension (systolic blood pressure less than 90 mm Hg) and hypoxemia (PaO2 less than 60 mm Hg) adversely affect the outcomes of patients with traumatic brain injury. In an analysis of patients entered in the Traumatic Coma Data Bank, a single episode of hypotension was found to more than double the mortality rate (60% vs. 27%). Patients with hypoxia and hypotension had a 75% mortality rate, and 94% had unfavorable outcomes (Glasgow Outcome Scale of death, vegetative state, and severe disability) (67). Avoidance of either condition is recommended for managing traumatic brain injury.

Neither the effects of erythropoietin nor maintaining a high transfusion threshold on patients with closed head injury resulted in improved neurologic outcomes (84). Robertson and colleagues found no beneficial influence of raising the hematocrit, either chemically or via transfusion.

Intracranial pressure monitoring. Intracranial pressure monitoring can be performed through the subarachnoid, the ventricular, or, rarely, the epidural space. Monitoring intracranial pressure allows the clinician to monitor and treat elevations in brain pressure associated with traumatic brain injury. Intracranial pressure monitoring is recommended for all severe traumatic brain injury victims with an abnormal CT scan. Patients with a normal CT scan but no evidence of drug or alcohol intoxication also may be considered likely candidates. Results from the Traumatic Coma Data Bank indicate that the proportion of time a patient spends with intracranial pressure elevation over 20 mm Hg is highly predictive of outcome (67). Ventricular pressure monitoring is the gold standard; however, fiberoptic subarachnoid or intraparenchymal catheters may be used if the ventricles are small. Newer technologies will allow multiparameter recording. A publication by Valadka and colleagues described using brain tissue PO2 monitors in head injury (104). Sustained brain PO2 values less than 15 torr or any value less than 6 torr were associated with increased mortality. Similar CO2, pH, and temperature probes are available, although the clinical utility is not yet proven.

Continuous or intermittent drainage has been employed when ventricular monitors are used. Nwachuku and colleagues performed a randomized trial and found more effective intracranial pressure control with continuous drainage (77).

Over the past decades, intracranial pressure monitoring has become a standard in most neurosurgical centers in the developed world. Chesnut and colleagues challenged this approach in an ambitious trial in Bolivia and Ecuador (21). Patients were randomly assigned after severe traumatic brain injury to be managed according to monitoring or solely based on imaging and clinical examination. Surprisingly, the treatment group did not affect ICU days, adverse events, and survival. Whether this finding extrapolates to modern American trauma centers remains to be seen.

Threshold for intracranial pressure and cerebral perfusion pressure. Intracranial pressure above 20 mm Hg has been considered elevated in traumatic brain injury studies. This value has never been substantiated in rigid clinical trials. The American Association of Neurological Surgeons guidelines focus on cerebral perfusion pressure as an option (cerebral perfusion pressure or mean arterial pressure-intracranial pressure). A critical value of cerebral perfusion pressure of 70 mm Hg has been suggested. Again, this value has not been substantiated by a clinical trial. In an analysis of data collected from the Selfotel trial, Juul and colleagues analyzed the effect of threshold values of intracranial pressure and cerebral perfusion pressure on worsening in patients after severe head injury (57). They found that elevation of intracranial pressure of 20 mm Hg or more was far more predictive than cerebral perfusion pressure of unfavorable outcomes.

Hyperventilation. In the past, chronic prolonged hyperventilation was used to enact a reduction in intracranial pressure. Lowering the PaCO2 causes reduced cerebral blood flow. However, subsequent research has suggested that prolonged hyperventilation may be detrimental. Severe traumatic brain injury is associated with a decreased cerebral blood flow independent of hyperventilation. Further reductions in cerebral blood flow may compromise areas of relative ischemia in damaged brain. Muizelaar randomized patients to hypocapnia and normocapnia after severe traumatic brain injury (74). Patients with a low CO2 had poorer outcomes at 3- and 6-month follow-up. Diringer and colleagues studied cerebral blood flow after closed head injury in patients treated with moderate hyperventilation and found no impairment in global cerebral blood flow, although regional blood values were not reported (29).

Mannitol or diuretic treatment. Mannitol can provide control of elevated intracranial pressure. Mannitol is thought to work not only as an osmotic diuretic but also as a plasma expander, reducing blood viscosity and increasing cerebral blood flow. Mannitol is normally bolused in a 1 g/kg dose followed by chronic administration of either a drip or intermittent boluses. All patients receiving mannitol should have a Foley catheter and have their serum osmolarity monitored to avoid hypernatremia and dehydration. Furosemide or other thiazide diuretics or urea are other potential osmotic agents.

Mannitol administration was subjected to a Cochrane Database review in 2005. Mannitol administration in patients with severe closed head injury was superior to placebo or barbiturate but potentially inferior to hypertonic saline (108). Mannitol administration effectively lowered intracranial pressure in 14 instances of intracranial pressure elevation, with maximal effects at 40 minutes and effects continuing up to 100 minutes after administration with a mean reduction from 25 to 17 mm Hg (88).

Hypertonic saline. Hypertonic saline has been employed as an alternative to mannitol or diuretics, providing similar control of elevated intracranial pressure without renal toxicity. Ware and colleagues used a 23.4% hypertonic saline solution in adult patients and compared results to mannitol administration (110). They found that both agents reduced intracranial pressure effectively; however, the hypertonic saline provided longer-lasting results.

Mannitol versus hypertonic saline. Trials have been designed to compare the use of mannitol versus hypertonic saline in severe traumatic brain injury. Rickard and colleagues concluded that although both treatments effectively reduced intracranial pressure, there was a trend favoring hypertonic saline (82). Boone and associates reviewed the literature and found no consistent advantage of either regimen (12). Bulger and colleagues performed a randomized prospective trial of prehospital hypertonic saline administration by emergency medical services personnel. There were no clinical benefits at 6 months of either hypertonic saline or hypertonic saline with dextran as opposed to normal saline (15). Conversely, Li and associates found a significant benefit of hypertonic saline as opposed to mannitol in reducing intracranial pressure (63). Finally, Burgess and colleagues found that hypertonic saline resulted in better intracranial pressure reduction than saline; however, the mortality and outcomes were unaffected (16).

Barbiturates. Barbiturate therapy has been found experimentally to reduce intracranial pressure. Eisenberg and colleagues conducted a prospective, double-blind study using a loading dose of 10 mg/kg of pentobarbital over 30 minutes, followed by 5 mg/kg per hour for three doses and 1 mg/kg per hour subsequently (34). In this study, patients who "responded" to barbiturates had a lower mortality, although no effect on outcome was substantiated. Schwartz and colleagues compared pentobarbital to mannitol and found no change in outcome in patients with mass lesions and worse outcomes with pentobarbital in individuals with diffuse injury (90).

Glucocorticoids. The use of glucocorticoids is not currently indicated in managing head injury. Large prospective studies by Braakman, Cooper, Giannotta, Deaden, and colleagues have failed to show an advantage of steroids over a placebo (25; 13; 39; 28). Subsequent trials of nonglucocorticoid steroids have failed to show a definitive effect.

A 1996 survey of UK neuroscience intensive-care units showed that corticosteroids were used in 14% of units to treat traumatic brain injury (51), and a similar study in the United States demonstrated steroid administration to be a component of traumatic brain injury therapy in nearly two thirds of U.S. trauma centers (37). Results of a systematic review conducted in 1997 suggested that steroid therapy for patients with closed head injuries decreased the absolute risk for death by 1% to 2% compared with no steroid treatment (02). However, an evidence-based review of the available literature, conducted by the Cochrane Group in 2000, failed to demonstrate any conclusive benefit of corticosteroid administration in patients with traumatic brain injury (03).

A large multicenter, randomized clinical trial has been completed to address the question of steroid therapy in closed head injury. A total of 10,008 patients were enrolled at 239 hospitals in 49 countries. Inclusion criteria required a Glasgow Coma Scale score of 14 or less and initiation of treatment within 8 hours of injury. Patients were randomized to receive either methylprednisolone or placebo for 48 hours. Primary outcomes were mortality within 2 weeks and death and disability at 6 months. At 2 weeks, 21.1% of patients treated with corticosteroids and 17.9% of those treated with placebo had died (relative risk, 1.18; P=0.0001). At 6 months, 25.7% of the corticosteroid group had died versus 22.3% of the placebo group (relative risk 1.15, p=0.0001). Subset analysis showed that the mortality difference was not affected by injury severity score, time from injury to initiation of steroids, or degree of extracranial trauma (83; 33).

Based on the class I data now available from the trial and supported by Cochrane Group analysis, it does not appear that the routine administration of corticosteroids has a role in the management of traumatic brain injury. This conclusion should not be generalized to spinal cord injury management at this time (04).

Amantadine therapy. Giacino and associates investigated the effectiveness of amantadine in improving the pace of recovery following traumatic brain injury. Patients were randomized to treatment versus placebo between 4 and 16 weeks after severe traumatic brain injury. Amantadine was postulated to work as a dopamine or N-methyl-D-aspartate antagonist. The amantadine treatment group showed an accelerated rate of neurologic recovery, although after treatment, the placebo group caught up in terms of functional recovery (38).

Progesterone therapy. Numerous preclinical and phase 1 and 2 trials have demonstrated the neuroprotective effects of progesterone, suggesting a potential beneficial effect in human closed-head injury (Wei and Xiao 2013). Skolnick and colleagues conducted a multicenter clinical trial using dosing with progesterone in patients with severe closed head injury (95). Unfortunately, they found no progesterone benefit in severe traumatic brain injury in this randomized, prospective trial.

Seizure prophylaxis. Annegers and colleagues found that 17% of severe traumatic brain injury victims had new unprovoked seizures (05).

The risk of seizures was increased in individuals with brain contusion, subdural hematoma, skull fracture, loss of consciousness, or amnesia of greater than 1 day and age greater than 65 years. Temkin and colleagues demonstrated that seizure prophylaxis with phenytoin was not protective of late seizures after severe traumatic brain injury (102). Phenytoin only is recommended in these patients, starting with a loading dose of 1 g of phenytoin or its more rapid-acting metabolite, fosphenytoin.

More recent trials have called into doubt the need and the safety of seizure prophylaxis after severe head injury. Bhullar and colleagues showed detrimental effects on recovery without influencing seizure frequency, and they advised further randomized studies (10). When seizure prophylaxis is utilized, a shift from phenytoin to levetiracetam has been advocated (61).

Hypothermia. Although a multicenter National Institutes of Health-funded trial failed to show a statistically significant benefit of induced hypothermia after traumatic brain injury, enthusiasm for this approach persists. Jiang and colleagues demonstrated improved mortality rates and outcomes in patients treated with mild (33°C to 35°C) hypothermia for between 3 and 14 days (55). Conversely, no benefit of hypothermia was found in a trial of patients presenting with low intracranial pressure scores (94). A study examining the effect of both hypothermia and progesterone has been reported to show the beneficial effects of hypothermia in an early analysis of 64 of a planned accrual of 250 patients (89). Hypothermia is typically induced with cooling blankets, although more direct methods exist. The patient is typically chemically paralyzed and dosed with chlorpromazine to prevent shivering. Rewarming is performed passively. Hypothermia is used in combination with other therapies for lowering intracranial pressure. New intracranial pressure monitors allow real-time monitoring of brain temperature.

Nonetheless, enthusiasm has been tempered by negative clinical trials, including an examination of hypothermia in pediatric patients that showed no benefit for hypothermia of 32.5 degrees initiated within 8 hours and continuing for 24 hours (50).

Erythropoietin as a neuroprotective agent. Based on preclinical studies showing a possible benefit of erythropoietin in traumatic brain injury through better oxygen-carrying capacity, Nichol and colleagues designed a randomized prospective trial comparing subcutaneous erythropoietin versus placebo within 24 hours of injury (75). There were safety concerns regarding thrombotic events in the trial design. The trial accrued 606 patients in 29 centers and, unfortunately, did not show a statistically significant benefit of erythropoietin in terms of improving neurologic outcome. On the other hand, there was no increase in thrombotic complications seen in the patients who received erythropoietin (76). A post hoc analysis of the trial looked at individuals with more severe injury severity scores indicating significant extracranial injury and found a statistically significant benefit in these patients who received a rather poor score compared to controls in terms of 6-month mortality (96).

Hyperbaric oxygen in traumatic brain injury. Because of the known adverse effect of hypoxemia on brain function, hyperbaric oxygen treatment has been postulated as a potential agent to improve outcomes after traumatic brain injury both in the acute setting and also to potentially alleviate postconcussive symptoms. Rockswald and colleagues reported a phase 2 trial of 42 patients randomized within 24 hours of injury to receive three consecutive treatments of 3 hours of hyperbaric oxygen (85). This trial showed improvements in both secondary markers, such as brain tissue partial pressure of oxygen and improved lactate/pyruvate ratio and also a 26% reduction in mortality compared to controls and a 36% improvement in favorable outcome. A Cochrane review of the subject identified a total of 571 individuals accrued in seven studies. Although two studies demonstrated improvements in the GCS scale over the treatment course, these results could not be extrapolated, and a larger randomized study was suggested to examine potential benefits (09). A systematic review suggested that the safety profile of hyperbaric oxygen after traumatic brain injury has been established, with strong proof of concept evidence for treatment pending a larger trial (27).

Rapid correction of coagulopathy in traumatic brain injury. Coagulopathy in the setting of intracranial hemorrhage can result from concomitant warfarin use, massive volume replacement in multisystem trauma, hepatic disease, or release of tissue thromboplastin in head injury alone. The dire consequences of anticoagulation in the traumatic brain injury population are supported by 50% to 60% mortality for these patients suffering an intracranial hemorrhage (42; 58).

Complicating the care of these patients is the fact that they often harbor an intracranial mass lesion requiring emergent surgical evacuation. Traditional treatment of coagulopathy has been with fresh frozen plasma. Fresh frozen plasma for the correction of coagulopathy is often time-consuming, which can significantly delay movement of the patient to the operating room for definitive therapy. In the late 1980s, a recombinant form of activated factor VII (rFVIIa) (NovoSeven; Novo Nordisk, Copenhagen, Denmark) was developed to treat bleeding in patients with hemophilia. rFVIIa cleaves factor X to Xa, leading to the generation of thrombin (46). It has subsequently been used for hemostasis in cardiac surgery, liver transplantation, and gastrointestinal hemorrhage (112; 44; 43). In the neurosurgical population, rFVIIa has been studied in the aneurysmal subarachnoid hemorrhage population, in pediatric spinal fusions, and in patients with traumatic intracerebral hemorrhage. Veshchev and colleagues reported on a single patient with warfarin-induced coagulopathy and an acute subdural hematoma treated with rFVIIa for rapid correction of an INR of 6.39 (107). This patient then underwent emergent hematoma evacuation. Repeated INR measurement after a single dose of rFVIIa was 1.25 and remained normal (107).

Park and colleagues have also reported on the use of rFVIIa in intracerebral hemorrhage (78). They examined nine patients who received rFVIIa for emergent neurosurgical intervention in the face of coagulopathy. Five of the patients had suffered an intracerebral hemorrhage. All patients’ coagulation profiles normalized following administration of rFVIIa, as quickly as 20 minutes in most cases (78).

Dosing regimens for rFVIIa range from 35 to 90 mcg/kg per dose given IV Push (93). Complications are rare but include acute myocardial infarction and pulmonary embolism (79; 86). A single 1200 mcg vial of rFVIIa costs approximately USD 00.00 (78).

Investigations have centered on antifibrinolytic agents, most specifically tranexamic acid. Tranexamic acid has been used for many years intraoperatively to address excessive bleeding. Examinations of tranexamic acid to prevent aneurysm rebleeding have shown some efficacy offset by medical complications. Yutthakasemsunt and associates performed a randomized trial of tranexamic acid versus placebo in 238 individuals with Glasgow Coma Scale randomized within 8 hours from injury giving just a single 2 gm dose (115). They found a slightly reduced risk of hemorrhage progression, but statistical significance was not met. No increased risk was seen.

In another small, randomized trial, Jokar and colleagues randomized 80 patients to receive a placebo or two doses of tranexamic acid in patients presenting with CT evidence of traumatic intracranial hemorrhage (56). The patients receiving tranexamic acid showed a statistically diminished rate of hematoma expansion. This was a small study, however, and included both intracerebral and extra-axial hemorrhages.

Finally, and most importantly, the CRASH-3 trial was published in 2019 as a randomized placebo-controlled trial in 175 sites in 29 countries (20). This trial was a follow-up to CRASH-2, which was performed in victims of general trauma. This trial included patients with a Glasgow Coma Score of 12 or lower with any intracranial bleeding on CT. Recruitment was initially within 8 hours of injury, with that window narrowed to 3 hours midway through the trial based on interim analysis. An initial bolus of 1 gm was provided, followed by an infusion of another gram over 8 hours. Over 12,000 patients were randomized. The trial showed a reduction in injury-related deaths in patients with moderate closed head injury treated within 3 hours. There was no increase in vascular occlusive events or seizures. Patients with severe head injury did not show the same benefit.

More recent trials have focused on fibrinogen administration after traumatic brain injury. Sabouri and colleagues reported a randomized controlled trial of 137 patients with severe traumatic brain injury and a Glasgow Coma Scale score of less than 9 (87). Patients with low fibrinogen (lower than 200 mg/dl) were randomized, with 71 patients included in the final analysis. This group found that the patients receiving fibrinogen infusion had improvements in short-term Glasgow Coma Scale scores and improved measures of hematoma expansion as well as Glascow Outcome Scale-extended scores. Fujiwara and associates examined two protocols for predicting and treating hypofibrinogenemia but found no significant differences in terms of 30-day mortality (35). They found that coagulation abnormalities on arrival were severe risk factors for developing hypofibrinogenemia and showed that fibrinogen levels were maintained with transfusion.

Diabetes insipidus in closed head injury. Following traumatic brain injury, abnormalities of water balance are among the most commonly recognized metabolic disorders. Damage to the supraoptic and paraventricular nuclei of the hypothalamus, their axons within the pituitary stalk, or the neurohypophysis results in polyuria, an inability to concentrate the urine, and elevated serum osmolality. This constellation of findings, in the presence of low levels of circulating vasopressin, is known as central diabetes insipidus.

In 1986, Edwards and Clark noted diabetes insipidus to occur in 23 of 53 (43%) patients reported with posttraumatic pituitary dysfunction (32). In their review, Yuan and Wade concluded that traumatic diabetes insipidus should be considered relatively rare, arising only after severe brain trauma (114). However, additional work has shown diabetes insipidus to occur acutely in about one quarter of patients following head injury and is unrelated to the severity of the head injury, as assessed by Glasgow Coma Score (01). Diabetes insipidus has also been reported to occur in 13% of patients in the late head injury period (11).

The drug of choice for the treatment of central diabetes insipidus is desmopressin, which is a synthetic analogue of arginine vasopressin. Desmopressin may be administered orally, intranasally, or parenterally. There are wide individual variations in the dosage required to control diuresis. Daily requirements for oral preparations vary from 100 to 1200 µg in three divided doses, for intranasal around 2 to 40 µg, and for parenteral 0.1 to 1 µg. A low dose should be used initially, which can then be increased as needed. Dilutional hyponatremia is the primary side effect if desmopressin is administered in excess over a prolonged period.

Surgical management of diffuse brain injury. Decompressive procedures to address diffuse as opposed to focal mass lesions have remained controversial and are given as no more than an option in managing these conditions. Although some reports have offered a case-control evaluation of decompressive craniectomy showing a benefit in select populations (80), prospective studies have not yet validated this finding. In a population-based study of 147 patients undergoing either unilateral or bilateral decompressive craniectomy over a 4-year period, Honeybul and colleagues found that functional outcome was significantly improved compared to a head injury prediction model for either procedure (48). Chibbaro and associates detailed a prospective multicenter trial of 147 consecutive patients undergoing decompressive craniectomy (22). They found that 67% had a favorable outcome based on a Glasgow Outcome Score of 4 or 5. Ecker and associates reported 33 patients treated with decompressive craniectomy during the conflicts in Iraq and Afghanistan and documented Glasgow Outcome Score of 4 or 5 in 60% of patients and 23% mortality (31).

Cooper and colleagues performed a randomized prospective trial comparing bifrontotemporalparietal decompressive craniectomy to standard medical care in 155 patients (24). In this trial, surgery improved intracranial pressure, but outcomes were actually worse than in the control patients at 6 months. However, delays in randomization and choice of a less aggressive surgical approach may affect the applicability of the results. Conversely, Ho and associates analyzed 168 patients undergoing decompressive craniectomy and documented that unless the predicted unfavorable outcome risk was greater than 80%, decompressive craniectomy was not only effective but also resulted in significant cost savings (45).

Outcomes

Overall, prognosis of severe traumatic brain injury is highly dependent on initial presentation. Individual diagnoses also color recovery. Humble and associates examined 311 victims of severe traumatic brain injury (49). Within this cohort, 47% had clinical diffuse axonal injury, and 56% had radiographic evidence of diffuse axonal imaging on MRI. Clinical but not radiographic evidence of diffuse axonal injury was related to poorer short-term recovery scores and functional independence measurements at hospital discharge. However, neither clinical nor radiographic diffuse axonal injury correlated to survival, long-term Glasgow outcome scale scores, or long-term quality of life scales. Conversely, van Eijck and colleagues performed a systematic review of the same topic and found that the risk of unfavorable outcomes in traumatic brain injury with diffuse axonal injury was three times higher than in patients without it (105). The overall favorable outcome rate with diffuse axonal injury was 62%. Lesions in the corpus callosum were associated with more unfavorable outcomes.

Special considerations

Anesthesia

In general, anesthetic management of patients with severe traumatic brain injury should avoid hypotension and hypoxemia and optimize intracranial pressure and cerebral perfusion pressure.

Premedication should, if possible, be reserved until after a formal neurologic examination because the neurologic examination will be blunted by sedatives and eliminated by paralytics. Benzodiazepines (midazolam 0.01 to 0.05 mg/kg) and narcotics (fentanyl 100 to 200 µg) provide an excellent combination for sedation in patients with head injury. Propofol given as a continuous drip may provide a rapidly reversible alternative that can lower intracranial pressure. Monitoring must include a precordial Doppler probe if venous sinus injury is suspected. Aspiration pneumonia is common in these patients and should be addressed if suspected clinically or radiographically. Anesthetic induction is often initiated with lidocaine (1 to 1.5 mg/kg) to blunt physiological changes during intubation. Thiopental and etomidate both reduce intracranial pressure and cerebral blood flow and are neuroprotective. Paralysis with succinylcholine may increase intracranial pressure. Rocuronium or vecuronium are alternatives.

Anesthetic maintenance should be directed towards lowering the cerebral metabolic rate and intracranial pressure without causing systemic hypotension. A combination of narcotics, benzodiazepines, and barbiturates can achieve this balance. Nitrous oxide is generally to be avoided because of the risk of tension pneumocephalus. Volatile anesthetics may cause cerebral vasodilation and increase intracranial pressure. When possible, blood loss should be carefully monitored and replaced with packed red blood cells or colloid (30).