The treatment of head injury may be divided into the treatment of closed head injury and the treatment of penetrating head injury. While significant overlap exists between the treatments of these 2 types of injury, some important differences are discussed. Closed head injury treatment is divided further into the treatment of mild, moderate, and severe head injuries.
Mild head injury
Most head injuries are mild head injuries. Most people presenting with mild head injuries will not have any progression of their head injury; however, a small percentage of mild head injuries progress to more serious injuries. Mild head injuries may be separated into low-risk and moderate-risk groups. Patients with mild-to-moderate headaches, dizziness, and nausea are considered to have low-risk injuries. Many of these patients require only minimal observation after they are assessed carefully, and many do not require radiographic evaluation. These patients may be discharged if a reliable individual can monitor them. Patients on anticoagulation therapy should have radiographic imaging performed even if they have had minimal head trauma as they can progress from a mild head injury to a catastrophic injury because their ability to coagulate blood has been medically inhibited.
Patients who are discharged after mild head injury should be given an instruction sheet for head injury care. The sheet should explain that the person with the head injury should be awakened every 2 hours and assessed neurologically. Caregivers should be instructed to seek medical attention if patients develop severe headaches, persistent nausea and vomiting, seizures, confusion or unusual behavior, or watery discharge from either the nose or the ear.
Patients with mild head injuries typically have concussions. A concussion is defined as physiologic injury to the brain without any evidence of structural alteration. Concussions are graded on a scale of I-V. A grade I concussion is one in which a person is confused temporarily but does not display any memory changes. In a grade II concussion, brief disorientation and anterograde amnesia of less than 5 minutes' duration are present. In a grade III concussion, retrograde amnesia and loss of consciousness for less than 5 minutes are present, in addition to the 2 criteria for a grade II concussion. Grade IV and grade V concussions are similar to a grade III, except that in a grade IV concussion, the duration of loss of consciousness is 5-10 minutes, and in a grade V concussion, the loss of consciousness is longer than 10 minutes.
As many as 30% of patients who experience a concussion develop postconcussive syndrome (PCS). PCS consists of a persistence of any combination of the following after a head injury: headache, nausea, emesis, memory loss, dizziness, diplopia, blurred vision, emotional lability, or sleep disturbances. Fixed neurologic deficits are not part of PCS, and any patient with a fixed deficit requires careful evaluation. PCS usually lasts 2-4 months. Typically, the symptoms peak 4-6 weeks following the injury. On occasion, the symptoms of PCS last for a year or longer. Approximately 20% of adults with PCS will not have returned to full-time work 1 year after the initial injury, and some are disabled permanently by PCS. PCS tends to be more severe in children than in adults. When PCS is severe or persistent, a multidisciplinary approach to treatment may be necessary. This includes social services, mental health services, occupational therapy, and pharmaceutical therapy.
After a mild head injury, those displaying persistent emesis, severe headache, anterograde amnesia, loss of consciousness, or signs of intoxication by drugs or alcohol are considered to have a moderate-risk head injury. These patients should be evaluated with a head CT scan. Patients with moderate-risk mild head injuries can be discharged if their CT scan findings reveal no pathology, their intoxication is cleared, and they have been observed for at least 8 hours.
Moderate and severe head injury
The treatment of moderate and severe head injuries begins with initial cardiopulmonary stabilization by ATLS guidelines. The initial resuscitation of a patient with a head injury is of critical importance to prevent hypoxia and hypotension. In the Traumatic Coma Data Bank study, patients with head injury who presented to the hospital with hypotension had twice the mortality rate of patients who did not present with hypotension. The combination of hypoxia and hypotension resulted in a mortality rate 2.5 times greater than if neither of these factors was present. Of note, recent studies have demonstrated that hyperoxia in severe head injury patients who have been intubated can also be deleterious. Hyperoxia with a PaO2> 300 mm Hg in ventilated TBI patients was associated with higher in-hospital case fatality. 
Once a patient has been stabilized from the cardiopulmonary standpoint, evaluation of their neurologic status may begin. The initial GCS score provides a classification system for patients with head injuries but does not substitute for a neurologic examination. After assessment of the coma score, a neurologic examination should be performed. If a patient has received muscle relaxants, the only neurologic response that may be evaluated is the pupillary response.
After a thorough neurologic assessment has been performed, a CT scan of the head is obtained. The results of the CT scan help determine the next step. If a surgical lesion is present, arrangements are made for immediate transport to the operating room. Fewer than 10% of patients with TBI have an initial surgical lesion.
Although no strict guidelines exist for defining surgical lesions in persons with head injury, most neurosurgeons consider any of the following to represent indications for surgery in patients with head injuries: extra-axial hematoma with midline shift greater than 5 mm, intra-axial hematoma with volume greater than 30 mL, an open skull fracture, or a depressed skull fracture with more than 1 cm of inward displacement. In addition, any temporal or cerebellar hematoma that is larger than 3 cm in diameter is considered a high-risk hematoma because these regions of the brain are smaller and do not tolerate additional mass as well as the frontal, parietal, and occipital lobes. These high-risk temporal and cerebellar hematomas are usually evacuated immediately
If no surgical lesion is present on the CT scan image, or following surgery if one is present, treatment of the head injury begins. The first phase of treatment is to institute general measures. Once appropriate fluid resuscitation has been completed and the volume status is determined to be normal, intravenous fluids are administered to maintain the patient in a state of euvolemia or mild hypervolemia.
Another supportive measure used to treat patients with head injuries is elevation of the head. When the head of the bed is elevated to 20-30°, the venous outflow from the brain is improved, thus helping to reduce ICP. If a patient is hypovolemic, elevation of the head may cause a drop in cardiac output and CBF; therefore, the head of the bed is not elevated in hypovolemic patients. In addition, the head should not be elevated (1) in patients in whom a spine injury is a possibility or (2) until an unstable spine has been stabilized.
Sedation is often necessary in patients with traumatic injury. Some patients with moderate head injuries have significant agitation and require sedation. In addition, patients with multisystem trauma often have painful systemic injuries that require pain medication, and many intubated patients require sedation. Short-acting sedatives and analgesics should be used to accomplish proper sedation without eliminating the ability to perform periodic neurologic assessments. This requires careful titration of medication doses and periodic weaning or withholding of sedation to allow periodic neurologic assessment. Intravenous lidocaine administered along with rapid sequence induction before endotracheal intubation is not associated with significant hemodynamic changes in traumatic brain injury patients. 
The use of anticonvulsants in patients with TBI is a controversial issue. No evidence exists that the use of anticonvulsants decreases the incidence of late-onset seizures in patients with either closed head injury or TBI. Temkin et al demonstrated that the routine use of Dilantin in the first week following TBI decreases the incidence of early-onset (within 7 d of injury) seizures but does not change the incidence of late-onset seizures.  In addition, the prevention of early posttraumatic seizures does not improve the outcome following TBI. Therefore, the prophylactic use of anticonvulsants is not recommended for more than 7 days following TBI and is considered optional in the first week following TBI.
After instituting general supportive measures, the issue of ICP monitoring is addressed. ICP monitoring has consistently been shown to improve outcome in patients with head injuries. ICP monitoring is indicated for any patient with a GCS score less than 9, any patient with a head injury who requires prolonged deep sedation or pharmacologic relaxants for a systemic condition, or any patient with an acute head injury who is undergoing extended general anesthesia for a nonneurosurgical procedure. Recent data from the American College of Surgeons Trauma Quality Improvement Program involving 10,628 adults with severe TBI revealed that ICP monitoring utilization was associated with lower mortality. 
ICP monitoring involves placement of an invasive probe to measure the ICP. Unfortunately, noninvasive means of monitoring ICP do not exist, although they are under development. ICP may be monitored by means of an intraparenchymal monitor, an intraventricular monitor (ventriculostomy), or an epidural monitor. These devices measure ICP by fluid manometry, strain-gauge technology, or fiberoptic technology.
Intraparenchymal ICP monitors are devices that are placed into the brain parenchyma to measure ICP by means of fiberoptic, strain-gauge, or other technologies. The intraparenchymal monitors are very accurate; however, they do not allow for drainage of CSF. Epidural devices measure ICP via a strain-gauge device placed through the skull into the epidural space. This is an older form of ICP measurement and is rarely used today because the other technologies available are more accurate and more reliable.
A ventriculostomy is a catheter placed through a small twist drill hole into the lateral ventricle. The ICP is measured by transducing the pressure in a fluid column. Ventriculostomies allow for drainage of CSF, which can be effective in decreasing the ICP. A risk of symptomatic hemorrhage exists with ventriculostomy placement, and Bauer et al report from a retrospective study that an international normalized ratio (INR) of 1.2-1.6 is an acceptable range for emergent ventriculostomy placement in patients with TBI. 
Once an ICP monitor has been placed, ICP is monitored continuously. No absolute value of ICP exists for which treatment is implemented automatically. In adults, the reference range of ICP is 0-15 mm Hg. The normal ICP waveform is a triphasic wave, in which the first peak is the largest peak and the second and third peaks are progressively smaller. When intracranial compliance is abnormal, the second and third peaks are usually larger than the first peak. In addition, when intracranial compliance is abnormal and ICP is elevated, pathologic waves may appear.
Lundberg described 3 types of abnormal ICP waves, A, B, and C waves.  Lundberg A waves, known as plateau waves, have a duration of 5-20 minutes and an amplitude of 50 mm Hg over the baseline ICP. After an episode of A waves dissipates, the ICP is reset to a baseline level that is higher than when the waves began. Lundberg A waves are a sign of severely compromised intracranial compliance. The rapid increase in ICP caused by these waves can result in a significant decrease in CPP and may lead to herniation. Recent data from a prospective observational study of severe blunt head injury patients demonstrated that a single episode of sustained increased ICP is an accurate predictor of poor outcome and is associated with an increased in-hospital mortality. 
Lundberg B waves have a duration of less than 2 minutes, and they have an amplitude of 10-20 mm Hg above the baseline ICP. B waves are also related to abnormal intracranial compliance. Because of their smaller amplitude and shorter duration, B waves are not as deleterious as A waves.
C waves, known as Hering-Traube waves, are low-amplitude waves that may be superimposed on other waves. They may be related to increased ICP; however, C waves can also occur in the setting of normal ICP and compliance.
When treating elevated ICP, remember that the goal of treatment is to optimize conditions within the brain to prevent secondary injury and to allow the brain to recover from the initial insult. Maintaining ICP within the reference range is part of an approach designed to optimize both CBF and the metabolic state of the brain. Treatment of elevated ICP is a complex process that should be tailored to each particular patient's situation and should not be approached in a "cookbook" manner. Many potential interventions are used to lower ICP, and each of these is designed to improve intracranial compliance, which results in improved CBF and decreased ICP.
Acute treatment of increased intracranial pressure
The Monro-Kellie doctrine provides the framework for understanding and organizing the various treatments of elevated ICP. In patients with head injuries, the total intracranial volume is composed of the total volume of the brain, the CSF, intravascular blood volume, and any intracranial mass lesions. The volume of one of these components must be reduced to improve intracranial compliance and to decrease ICP. The discussion of the different treatments of elevated ICP is organized according to which component of intracranial volume they affect.
The first component of total intracranial volume to consider is the blood component. This includes all intravascular blood, both venous and arterial, and comprises approximately 10% of total intracranial volume. Elevation of the head increases venous outflow and decreases the volume of venous blood within the brain. This results in a small improvement in intracranial compliance and, therefore, has only a modest effect on ICP.
The second component of intracranial vascular volume is the arterial blood volume. Hypocapnia is capable of reducing cerebral blood flow 4% for each mm Hg change in PaCO2. The control mechanism is probably extravascular pH changes in fluid bathing cerebral resistor vessels, which alter smooth muscle intracellular calcium concentrations. This may be reduced by mild-to-moderate hyperventilation, in which the PCO2 is reduced to 30-35 mm Hg. This decrease in PCO2 causes vasoconstriction at the level of the arteriole, which decreases blood volume enough to reduce ICP. The effects of hyperventilation have a duration of action of approximately 48-72 hours, at which point the brain resets to the reduced level of PCO2. This is an important point because once hyperventilation is used, the PCO2 should not be returned to normal rapidly. This may cause rebound vasodilatation, which can result in increased ICP.
Below a PaCO2 of 25-30 Torr, CBF falls much less rapidly, presumably because of severe enough vasoconstriction to induce hypoxemia in brain tissues, limiting oxygen delivery. PaCO2 tensions less than 25 Torr are sufficient to change brain metabolism into anaerobic, which increases acidosis. Low arterial O2 tensions influence CBF but to a lesser degree than PaCO2. No measurable changes in CBF occur during hypoxemia until the PaO2 drops below 50 Torr, at which time CBF gradually increases. In addition to reducing CBF, the resultant respiratory alkalosis may reverse local tissue acidosis, which develops in cerebral edema, benefiting cellular respiration and restoring autoregulation. Within 48-72 hours, renal mechanisms for handling bicarbonate excretion compensate for altered PaCO2 tensions, thereby normalizing cerebral pH and returning CBF to baseline values.
There are 3 theoretical physiologic paradoxes to hyperventilation therapy for the control of ICP.
Since cerebral vasospasm is a serious concern in subarachnoid hemorrhage (SAH), attempts to create further vessel constriction by hyperventilation in order to decrease concomitant cerebral edema are rarely indicated unless the amount of edema is clinically emergent.
Vessels in the damaged area of the brain have lost their autoregulatory control. While unaffected brain regions would vasoconstrict normally to the stimulus of decreased PaCO 2, damaged areas might vasodilate in response to diminished cerebral blood flow. This can create a “reverse steal” phenomenon, where blood and nutrients are diverted away from “normal” areas of the brain and into “damaged” areas. This diversion would feed the increased metabolic requirement of damaged tissues, but the sum total effect may cause more harm to the rest of the brain. In addition, the increased hydrostatic pressure combined with the capillary permeability damage might, in some cases, paradoxically increase ICP in damaged areas.
Sudden increases in PaCO 2, as a result of ventilator changes, often result in dramatic increases in CBF, and rapid deteriorations in the patient’s condition. During hyperventilation, the cerebral bicarbonate level gradually adjusts to offset the lower level of CO 2, maintaining normal pH. If the pCO 2 is allowed to rise suddenly, the excess CO 2 rapidly crosses the blood-brain barrier, but the bicarbonate level in the brain increases much less rapidly. The result is cerebral acidosis, with attendant cerebral vascular dilatation, increased cerebral blood volume, and elevated ICP, usually resistant to further hyperventilation.
Unfortunately, little objective evidence exists that treatment by hypocapnia has significantly improved mortality or survival. At best, it seems to be a temporary stop-gap measure until some other curative measure, such as surgery, might be attempted. Patients with the most prompt response to hyperventilation generally have the best prognosis for recovery. No evidence exists that hyperventilation therapy produces benefit in hypoxemic-anoxic encephalopathy.
CSF represents the third component of total intracranial volume and accounts for 2-3% of total intracranial volume. In adults, total CSF production is approximately 20 mL/h or 500 mL/d. In many patients with TBI who have elevated ICP, a ventriculostomy may be placed and CSF may be drained. Removal of small amounts of CSF hourly can result in improvements in compliance that result in significant improvements in ICP.
The fourth and largest component of total intracranial volume is the brain or tissue component, which comprises 85-90% of the total intracranial volume. When significant brain edema is present, it causes an increase in the tissue component of the total intracranial volume and results in decreased compliance and increased ICP. Treatments of elevated ICP that reduce total brain volume include diuretics, perfusion augmentation (CPP strategies), metabolic suppression, and decompressive procedures.
Diuresis and brain edema
Diuretics are powerful in their ability to decrease brain volume and, therefore, to decrease ICP. No single diuretic has been established in randomized controlled trials to be superior to another. Mannitol, an osmotic diuretic, is the most common diuretic used. Mannitol is a sugar alcohol that draws water out from the brain into the intravascular compartment. It has a rapid onset of action and a duration of action of 2-8 hours. Mannitol is usually administered as a bolus because it is much more effective when given in intermittent boluses than when used as a continuous infusion. The standard dose ranges from 0.25-1 g/kg, administered every 4-6 hours.
Because mannitol causes significant diuresis, electrolytes and serum osmolality must be monitored carefully during its use. In addition, careful attention must be given to providing sufficient hydration to maintain euvolemia. The limit for mannitol is 4 g/kg/d. At daily doses higher than this, mannitol can cause renal toxicity. Mannitol should not be given if the patient's serum sodium level is greater than 145 or serum osmolality is greater than 315 mOsm.
Other diuretics that sometimes are used in patients with TBI include furosemide, glycerol, and urea. Mannitol is preferred over furosemide because it tends to cause less severe electrolyte imbalances than a loop diuretic. Interestingly, mannitol and furosemide have a synergistic effect when combined; however, this combination tends to cause severe electrolyte disturbances. Urea and glycerol have also been used as osmotic diuretics. Both of these compounds are smaller molecules than mannitol and, as a result, tend to equilibrate within the brain sooner than mannitol; therefore, they have a shorter duration of action than mannitol. Urea has the additional problem that it can cause severe skin sloughing if it infiltrates into the skin.
Boluses of mannitol can generate a dramatic diuresis, resulting in rapid intravascular depletion and potential kidney damage. Mannitol can cause as much as 1500 cc of fluid to diurese in the space of 2 hours, as intravascular fluid depletion occurs, hematocrit can rise,and blood viscosity can increase. This makes the area of brain irritation much more amenable to stroke.
Hypertonic saline (3-23%) has generated some interest in the treatment of intracranial hypertension secondary to brain edema because it is thought to be less disruptive to fluid and electrolyte balance than other diuretic agents. In a recent review of the literature, 23% saline was noted to decrease elevated ICP by nearly 50% an hour after administration in patients with life-threatening elevations in ICP. 
Saline 3% or 7.5% administered in continuous infusion generates a more predictable and gentle osmotic flow of brain intracellular water into the interstitial space. The maximum effect occurs after the end of infusion and is visible over 4 hours. Therapeutically, the limits of serum sodium and osmolality are in the range of 155 and 320 respectively. More research is needed to elucidate the exact method of action of hypertonic saline, the optimal dosing and timing and the contraindications. 
Other supportive treatments
While awaiting possible operative therapy, other supportive treatments are as follows:
Early extraventricular drainage of CSF is sometimes of value in controlling brain edema if there is a suspicion that the ventricles will progressively diminish in size because edema cannot be cannulated from a burr hole.
Coughing and straining increase venous pressure, restricting drainage and backing up blood into the head, thereby increasing ICP. Neuromuscular paralysis may decrease ICP by preventing sudden changes related to coughing or straining and by promoting systemic venous pooling that increases venous drainage from the head. Any other restrictions to jugular blood drainage, such as a kinked neck from positioning in bed, increase ICP by retarding jugular drainage, transmitting pressure back into the brain.
Trying to differentiate a drug-induced coma from an increased ICP–induced coma with a trial of naloxone (Narcan) is contraindicated, as it invariably induces agitation if the stupor is narcotic induced. Agitation increases catecholamine response, increases cardiac output, and increases blood flow to the head, thereby increasing hydrostatic pressure and ICP.
Use of positive end-expiratory pressure (PEEP) for mechanical ventilation is controversial in TBI patients with acute lung injury/acute respiratory distress syndrome. Zhang et al found that PEEP can have a varied impact on blood, intracranial, and cerebral perfusion pressure in patients with cerebral injury. When applying this technique, mean arterial and intracranial pressure monitoring appears beneficial. 
Management of cerebral perfusion pressure
CPP management involves artificially elevating the blood pressure to increase the MAP and the CPP. Because autoregulation is impaired in the injured brain, pressure-passive CBF develops within these injured areas. As a result, these injured areas of the brain often have insufficient blood flow, and tissue acidosis and lactate accumulation occur. This causes vasodilation, which increases cerebral edema and ICP. When the CPP is raised to greater than 65-70 mm Hg, the ICP is often lowered because increased blood flow to injured areas of the brain decreases the tissue acidosis. This often results in a significant decrease in ICP.
Metabolic therapies are designed to decrease the cerebral metabolic rate, which decreases ICP. Metabolic therapies are powerful means of reducing ICP, but they are reserved for situations in which other therapies have failed to control ICP. This is because metabolic therapies have diffuse systemic effects and often result in severe adverse effects, including hypotension, immunosuppression, coagulopathies, arrhythmias, and myocardial suppression. Metabolic suppression may be achieved through drug therapies or induced hypothermia.
Barbiturates are the most common class of drugs used to suppress cerebral metabolism. Barbiturate coma is typically induced with pentobarbital. A loading dose of 10 mg/kg is administered over 30 minutes, and then 5 mg/kg/h is administered for 3 hours. A maintenance infusion of 1-2 mg/kg/h is begun after loading is completed. The infusion is titrated to provide burst suppression on continuous electroencephalogram monitoring and a serum level of 3-4 mg/dL. Typically, the barbiturate infusion is continued for 48 hours, and then the patient is weaned off the barbiturates. If the ICP again escapes control, the patient may be reloaded with pentobarbital and weaned again in several days.
Hypothermia may also be used to suppress cerebral metabolism. The use of mild hypothermia involves decreasing the core temperature to 34-35°C for 24-48 hours and then slowly rewarming the patient over 2-3 days. Patients with hypothermia are also at risk for hypotension and systemic infections. This therapy remains controversial because of the lack of high quality trials that have clearly established neurologic benefits or mortality reduction.  There are no known randomized controlled trials of modest cooling therapies (35-37.5 degrees Celcius). 
Another treatment that may be used in patients with TBI with refractory ICP elevation is decompressive craniectomy. In this surgical procedure, a large section of the skull is removed and the dura is expanded. This increases the total intracranial volume and, therefore, decreases ICP. Which patients benefit from decompressive craniectomy has not been established. Some believe that patients with refractory ICP elevation who have diffuse injury but do not have significant contusions or infarctions will benefit from decompressive craniectomy. Nirula and colleagues investigated whether early decompressive craniotomy, done within 48 hours of TBI was associated with improved survival in patients with refractory intracranial hypertension. Their results did not demonstrate a survival benefit when compared to patients treated with standard medical therapy. This led the authors to recommend that neurosurgeons pause before considering this "resource-demanding form of therapy." 
Management of elevated ICP involves using a combination of treatments. Each patient represents a slightly different set of circumstances, and treatment must be tailored to each patient. Although no rigid protocols have been established for the treatment of head injury, many published algorithms provide treatment schemas.
The American Association of Neurologic Surgeons published a comprehensive evidence-based review of the treatment of TBI, called the Guidelines for the Management of Severe Head Injury. In these guidelines, 3 different categories of treatments, standards, guidelines, and options are outlined. Standards are the accepted principles of management that reflect a high degree of clinical certainty. Guidelines are a particular strategy or a range of management options that reflect a high degree of clinical certainty. Options are strategies for patient management for which clinical certainty is unclear.
The treatment of penetrating brain injuries involves 2 main aspects. The first is the treatment of the TBI caused by a penetrating object. Penetrating brain injuries, especially from high-velocity missiles, frequently result in severe ICP elevations. This aspect of penetrating brain injury treatment is identical to the treatment of closed head injuries.
The second aspect of penetrating head injury treatment involves debridement and removal of the penetrating objects. Penetrating injuries require careful debridement because these wounds are frequently dirty. When objects penetrate the brain, they introduce pathogens into the brain from the scalp surface and from the surface of the penetrating object.
Penetrating injuries may be caused by high-velocity missiles (eg, bullets), penetrating objects (eg, knives, tools), or fragments of bone driven into the brain. Bullet wounds are treated with debridement of as much of the bullet tract as possible, dural closure, and reconstruction of the skull as needed. If the bullet can be removed without significant risk of neurologic injury, it should be removed to decrease the risk of subsequent infection. Penetrating objects, such as knives, require removal to prevent further injury and infection. If the penetrating object either is near or traverses a major vascular structure, an angiogram is necessary to assess for potential vascular injury. When the risk of vascular injury is present, penetrating objects should be removed only after appropriate access has been obtained to ensure that vascular control is easily achieved.
Penetrating brain injuries are associated with a high rate of infection, both early infections and delayed abscesses. Appropriate debridement and irrigation of wounds helps to decrease the infection rate. Some of the risk factors for infection following penetrating brain injury include extensive bony destruction, persistent CSF leak, and an injury pathway that violates an air sinus.
Late-onset epilepsy is a common consequence of penetrating brain injuries and can occur in up to 50% of patients with penetrating brain injuries. No evidence exists that prophylactic anticonvulsants decrease the development of late-onset epilepsy. During the Vietnam War, prophylactic anticonvulsants were used, and the rate of late-onset epilepsy was not different from that of previous wars, when prophylactic anticonvulsants were not used.
Functional deficits resulting from TBI are common and can be divided into 2 categories, as follows: systemic complications and neurologic complications. The systemic complications of TBI are typical of any severe injury and depend on the types of intensive treatments used. Be aware of the complications of intensive care treatment when considering systemic complications of head injury.
One of the most feared systemic complication of TBI is venous thromboembolism (VTE). Valle and colleagues investigated whether polytrauma patients with TBI had higher rates of VTE than polytrauma patients without TBI. While the patients with TBI had worse Injury Severity Scores, longer ICU lengths of stay and more hypercoagulable thomboelastogram values on admission, they did not have higher rates of VTE.  The timing of when to initiate chemical thromboprophylaxis in patients with intracranial hemorrhage has not been fully elucidated. Farooqui and colleagues initiated chemoprophylaxis for TBI patients 24 hours after a stable head CT was obtained. This intervention was associated with a lower DVT rate in patients started on chemoprophylaxis. There was no significant difference in pulmonary embolism rates. Of note, there was no increase in intracranial hemorrhage in patient's following the protocol when compared with patients who did not recieve routine chemoprophylaxis for VTE. 
The neurologic complications of TBI include focal neurologic deficits, global neurologic deficits, seizures, CSF fistulae, hydrocephalus, vascular injuries, infections, and brain death.
Focal neurologic deficits
Focal neurologic deficits are quite common following TBI. Cranial nerves are affected often because of their anatomic location at the base of the brain. When the brain shifts within the skull as it undergoes either acceleration or deceleration forces, significant force is often placed on the entire brain and the cranial nerves. The cranial nerves are tethered at their exit sites from the skull, and, as a result, they may be stretched when the brain shifts as a result of acceleration or deceleration forces. In addition, the cranial nerves are very susceptible to injury as they course through narrow bony canals and grooves. The cranial nerves that are injured most commonly in patients with TBI are cranial nerves I, IV, VII, and VIII.
Anosmia caused by traumatic injury to the first cranial nerve occurs in 2-38% of patients with TBI. It is more common in those with frontal fractures and in those with posttraumatic rhinorrhea. Posttraumatic anosmia improves slowly, and as many as one third of patients do not show any improvement in olfaction.
Injuries to the fourth cranial nerve, the trochlear nerve, are also quite common. This nerve is often injured in patients with head trauma because it has the longest intracranial course of the cranial nerves. Injury to the trochlear nerve causes a positional diplopia, in which those affected experience diplopia when they look down and toward the eye in which the trochlear nerve is injured. As a result, to compensate, the head is tilted up and away from the side of the injury. Trochlear nerve injuries resolve fully in approximately two thirds of those with unilateral injury and in one fourth of those with bilateral injuries.
Facial nerve injuries often occur with head injuries in which the temporal bone is fractured. From 10-30% of persons with longitudinal fractures of the temporal bone and 30-50% of those with transverse fractures of the temporal bone have either acute or delayed facial nerve injury. Immediate facial nerve injury suggests direct injury to the nerve, while delayed injury suggests progressive edema within the nerve. In severely injured patients, a delay in the diagnosis of facial nerve injuries occurs frequently because facial nerve function is difficult to assess in obtunded patients.
Cochlear nerve injury (cranial nerve VIII) is also a common occurrence in patients with head injury, especially in patients with temporal bone fractures. In addition, vestibular disorders, including vertigo, dizziness, and tinnitus, are extremely common in patients with head injuries.
Hydrocephalus is a common late complication of TBI. Posttraumatic hydrocephalus may present as either ventriculomegaly with increased ICP or as normal pressure hydrocephalus. In patients with increased ICP secondary to posttraumatic hydrocephalus, the typical signs of hydrocephalus are often observed and include headaches, visual disturbances, nausea/vomiting, and alterations in the level of consciousness. Normal pressure hydrocephalus usually manifests as memory problems, gait ataxia, and urinary incontinence.
The diagnosis of normal pressure hydrocephalus may be difficult to make in patients with TBI because they often have memory difficulties and gait abnormalities secondary to their head injury. In addition, as many as 86% of patients with TBI demonstrate some degree of ventriculomegaly on follow-up CT scan images. This ventriculomegaly is often secondary to diffuse brain atrophy, and radiographic features rarely help make the distinction between atrophy and normal pressure hydrocephalus. Any patient who develops neurologic deterioration weeks to months following TBI should be evaluated for the possibility of normal pressure hydrocephalus. When CT scan findings cannot help distinguish between normal pressure hydrocephalus and ventriculomegaly secondary to brain atrophy, a high-volume lumbar puncture tap test is performed to ascertain if CSF drainage would improve the patient's neurologic condition.
Posttraumatic seizures are a frequent complication of TBI and are divided into 3 categories. Early seizures occur within 24 hours of the initial injury, intermediate seizures occur 1-7 days following injury, and late seizures occur more than 7 days after the initial injury. Posttraumatic seizures are very common in those with a penetrating cerebral injury, and late seizures occur in as many as half of these patients.
Cerebrospinal fluid fistulae
Cerebrospinal fistulae, either in the form of rhinorrhea or otorrhea, may occur in as many as 5-10% of patients with TBI. They may present either immediately or in a delayed fashion and are more frequent in patients with basilar skull fractures. Approximately 80% of acute cases of CSF rhinorrhea resolve spontaneously within 1 week. A 17% risk of meningitis exists when CSF rhinorrhea is present. Prophylactic antibiotics have not been demonstrated to decrease this meningitis risk, although very few studies have examined this issue. More than 95% of acute episodes of CSF otorrhea resolve spontaneously within 1 week, and CSF otorrhea is complicated by meningitis in fewer than 4% of cases.
When acute CSF fistulae do not resolve spontaneously, a lumbar subarachnoid drain may be placed for several days in an attempt to divert CSF and allow the fistula to close. If this fails, radiographic dye is introduced into the subarachnoid space via lumbar puncture (metrizamide cisternogram), and a high-resolution CT scan is performed in an attempt to identify the origin of the CSF fistula. A craniotomy is performed, and the fistula site is repaired. Delayed CSF fistulae may occur from 1 week after the initial injury to years later. These delayed fistulae are more difficult to treat and frequently require surgical intervention.
Vascular injuries are uncommon sequelae of TBIs. Arterial injuries that may occur following head trauma include arterial transactions, thromboembolic phenomena, posttraumatic aneurysms, dissections, and carotid-cavernous fistulae (CCF).
Arterial occlusions secondary to transactions or thromboembolism following closed head injuries are uncommon occurrences.
Posttraumatic intracranial aneurysms, which are also rare, differ from congenital aneurysms because the posttraumatic aneurysms tend to be located distally, as opposed to the congenital aneurysms, which are typically proximal in location.
Arterial dissections are more common than the aforementioned arterial injuries and should be considered if significant injury has occurred to the petrous portion of the temporal bone, through which the carotid artery passes, or when an unexplained neurologic deficit is present. A cerebral angiogram is often necessary to help exclude arterial injury in these cases.
Posttraumatic CCF occur when the internal carotid artery is injured within the cavernous sinus, resulting in a direct connection between the carotid artery and the veins of the cavernous sinus. This overloads the venous system and results in chemosis and proptosis on the affected side. Other signs of CCF include diplopia, ophthalmoplegia, visual disturbances, and headaches. Some high-risk fistulae may cause intracerebral hemorrhage. CCF are treated with endovascular balloon occlusion of the fistula origin.
Specific intracranial venous injuries are uncommon following TBI if one excludes the injury to the bridging veins, which are the most common source of subdural hematomas. Depressed skull fractures overlying any of the major intracranial venous sinuses may cause injury to the sinus. When these venous sinus injuries require treatment, substantial, and sometimes life-threatening, blood loss can occur.
A second type of venous injury following TBI involves venous sinus thrombosis. Although very rare following head injury, this is a potentially life-threatening injury because the impaired venous drainage often causes severe ICP elevations and venous infarction. The treatment of venous sinus thrombosis is anticoagulation, which presents significant risk in those with acute head injuries. If the thrombosis progresses despite systemic anticoagulation, direct intracranial intravenous thrombolysis is necessary.
Although it is underrecognized, posttraumatic vasospasm (PTV) can cause ischemic damage after severe TBI that predicts poor outcome. A database review of pediatric and adult patients identified fever on admission and small parenchymal contusions as independent risk factors for PTV. It is not known if treatments targeting fever and inflammation are effective in reducing PTV after severe TBI. 
Intracranial infections are another potential complication of TBI. In uncomplicated closed head injury, infection is uncommon. When basilar skull fractures and/or CSF fistulae are present, the risk of infection is increased. In addition, if a patient has had a ventriculostomy for ICP monitoring, the risk of infection is also increased, for either a ventriculitis or meningitis. Other intracranial infections, such as subdural or epidural empyema and intraparenchymal abscesses, are rare following closed head injury. As one would expect, the incidence of infection in penetrating cerebral injuries and open depressed skull fractures increases.
Diagnosis of brain death
Brain death protocols have evolved to become very specific and sensitive. Brain death is a diagnosis of what is, not what might be, and must be proven rather than insinuated.
Initially, for an accurate diagnosis of brain death, there must be clear evidence of an acute, catastrophic, irreversible brain injury, and any reversible conditions that may obfuscate the clinical assessment must be excluded. Subsequently, the physical examination must show complete unresponsiveness, absent motor responses, absent brainstem reflexes, and apnea. Further confirmatory studies, such as EEG or cerebral blood flow studies, may be ordered if there is any ambiguity in the clinical evaluation.
A typical brain death protocol may be summarized as follows:
Confirm that the patient is in a coma.
Evaluate the patient for seizure activity and decerebrate or decorticate movements.
Test for motor response to painful stimulation.
Test for pupillary response to light.
Test for corneal reflex.
Test for oculocephalogyric reflex (doll’s head reflex).
Test for vestibulo-ocular reflex (caloric test).
Test for upper and lower airway stimulation (eg, pharyngeal and endotracheal suction).
Test for gag reflex.
Perform apnea test. This test should be the last test and should be conducted after two clinical examinations (separated by the mandatory observation period) have confirmed the absence of brainstem functions. The patient is disconnected from the ventilator while oxygenation of the lungs is continued passively. On the basis of calculation (whereby PaCO 2 is assumed to rise 4 mm Hg in the first minute and 3 mm Hg every minute thereafter), the patient is allowed to build up to a PaCO 2 of 60 mm Hg or more without becoming hypoxic. If there is no respiratory effort, the apnea test is considered confirmatory.
Consider an EEG. The EEG should show electrocerebral silence for at least 30 minutes and must conform to established criteria for brain death.
If the cause of death cannot be determined with absolute accuracy, consider cerebral angiography. The absence of intracranial arterial circulation, as demonstrated by four-vessel angiography, confirms brain death.
Brain death and life support
In earlier times, it could be said that a person was dead when pulseless and apneic. Today, this view no longer suffices. Death is more a process than an event. Lack of blood flow to the brain leads to loss of consciousness within seconds, but other functions of the brain may persist for much longer. Other somatic organs may take hours to stop functioning, and connective tissues can take days to die.
The evolution of life-support systems capable of prolonging death indefinitely necessitated a more accurate definition of death, which arrived in 1968 with the formulation of the Harvard criteria for the working definition of death. In essence, these criteria considered the irreversible loss of brain function, rather than whole-body metabolic cessation, to be indicative of death. When the Harvard criteria were met, death was inevitable, even with continuing treatment. The Harvard criteria objectified the progression of disease, thereby making it possible for clinicians to predict death accurately even on somatic life support.
In 1981, the President’s Commission established brain death as a criterion for determining death, not simply for predicting the inevitability of death. The Uniform Determination of Death Act (UDDA) made brain death and cardiopulmonary collapse criteria for death in most states. Under the UDDA, death is pronounced at the time the criteria are met, and families may not demand continuing mechanical ventilation or other forms of ICU life support (except in the states of New York and New Jersey, both of which have conscience clauses).
The controversy surrounding death
Providers and ethicists continue to disagree on the question of whether whole-brain death is equivalent to death. These factions do, however, agree asserted that there is only one real definition of death—irreversible cessation of the integrated functioning of the organism as a whole. However, an ambiguous portion of the Dead Donor Rule has muddied the water, stating “an individual who has sustained either irreversible cessation of circulatory and respiratory functions or irreversible cessation of all functions of the entire brain is dead.”
The connections between cardiovascular and neurologic criteria are tenuous, and the criteria mean different things to different people at different times. For example, the criterion of irreversible cessation of all circulatory and respiratory function does not imply the cessation of brain function. Conversely, the criterion of irreversible cessation of all brain function does not imply the cessation of circulation and respiration. It appears, then, that the definition of death does not require the permanent cessation of the functioning of the organism as a whole but, rather, cessation of only certain functions.
Ethicists maintain that the criteria used to fulfill the definition of death should be both necessary and sufficient. This is not an easy standard to meet. For example, loss of consciousness is necessary for death, but it is not sufficient. Loss of heartbeat and breathing is sufficient for death, but it is not necessary if whole-brain death is present. Moreover, the UDDA does not require that every brain cell be dead for brain death to be declared—only those cells that contribute to the integration of the organism as a whole. Brains and bodies can appear dead but can be restored to function (eg, cold water drowning, brief periods of normothermic ischemia). As such, these people are not dead because the essence of death is irreversibility. In fact, “the irreversible loss of life” is the dictionary definition of death.
Outcome and Prognosis
The outcome of TBI is dependent on number of factors in addition to the severity of brain injury. The initial GCS score continues to have utility providing both a description of the initial neurologic condition, and to a lesser extent mortality prediction. A recently analyzed cohort of 2,808 TBI patients across multiple centers found a total GCS of 8 or less to be associated with a mortality of 28.7% by two weeks following injury.  Out of the three components of the GCS score, the motor portion is most predictive of outcome. Compared to patients with a GCS motor score of 6, there was a 2.2 fold increased risk mortality at two weeks with a motor score of 5 ranging to a 17 fold increased risk with a motor score of 1.
Advanced age is also an important negative prognostic factor. An examination of the Crash Injury Research Engineering Network database found a significantly higher mortality rate among elderly motor vehicle accident victims (age >60 years) compared to their younger counterparts.  Even mild TBI has been associated with a greater risk of death with elderly patients having a 1.25 fold greater risk compared to the general population after controlling for age and medical comorbidities. 
In addition to low initial GCS and advanced age other negative prognostic factors have been identified including pupillary size and response, and the development of hypoxia, pyrexia, and high intracranial pressure following initial presentation. These factors have also been associated with poor initial and long-term outcomes.
In a retrospective series of 846 patients with severe TBI (GCS of 8 or less) reported by Jiang et. al. one-third of patients had either died or remained in a persistent vegetative state at one year following their injury. However, 31.56% of individuals still experienced a good recovery defined as the ability to live independently and return to work or school.  These data highlight the difficulty with accurate prediction of individual patient prognosis in TBI.
In an attempt to assist with clinical decision making numerous predictive scoring systems have been developed including the recent International Mission on Prognosis and Analysis of Clinical trials in Traumatic brain injury database (IMPACT models) and the Corticosteroid Randomization After Significant Head Injury trial data (CRASH models). These systems have examined large data sets and undergone external validation with favorable results. However, as with most predictive scoring systems they are generalizable to the populations they are validated in and are not reliably generalizable to most patients.  Therefore, initial aggressive treatment of severe TBI will continue to be prudent until an accurate, fully generalizable means of prediction is available.
Future and Controversies
The most significant controversy today in the treatment of TBI is the minimum desirable CPP to achieve in the patient with a head injury. Previously, a CPP of 79 mm Hg was considered the minimum; however, many now believe that a CPP of 60 mm Hg is sufficient. Further controversy also exists as to whether elevated ICP or decreased CPP is a more important prognostic factor. Karamanos and colleagues recently shed some light on this issue in a prospective study that involved 216 patients who met Brain Trauma Foundation criteria for ICP monitoring. They concluded that decreased CPP did not affect survival, while a single episode of sustained ICP elevation was associated with increased mortality.  This is an important distinction because it directs the main goals of therapy in severely injured patients. If ICP elevations are considered a more important factor, then efforts may be directed at lowering ICP as a primary goal and improving CPP as a secondary goal.
Another controversy surrounds transfusion thresholds and the role of erythropoeitin. There is a trend toward better outcomes with a conservative transfusion strategy; meaning holding on transfusion for patients with a hemoglobin above 7 g/dl. In a retrospective review of data, Elterman and collegues discovered that in patients with TBI and no evidence of shock, transfusing patients with a hemoglobin over 10g/dl was associated with decreased 28-day survival and an increase in ARDS.  Robertson and colleagues recently published a randomized clinical trial evaluating transfusion thresholds of 7 vs. 10 g/dl as well as the impact of erythropoetin on recovery 6 months post injury. Their data demonstrated that erythropoeitin had no impact on favorable outcomes. There was no difference in rates of favorable outcome in either transfusion cut off arm but the rate of thromboembolic events was higher in those with the 10 g/dl transfusion threshold. 
There are on-going investigations into multiple therapies. Yutthakasemsunt and colleagues randomized TBI patients with GCS scores from 4-12 to recieve a single 2 gram dose of tranexamic acid in addition to standard therapy or placebo and standard therapy. Patients who recieved tranexamic acid had a trend toward reduction of progression of intracranial hemorrhage. Larger trials will be needed to see if this inexpensive and widely available therapy will have an impact on death or disability in TBI patients.  Other therapies for TBI under investigation include the use of dexamethasone, progesterone and vasopressin.