Chapter 93. Head Injury and Intracranial Hypertension

• Secondary brain injury from hypoxia and ischemia should be prevented by appropriate airway ventilatory and fluid management in all comatose patients.
• In appropriately selected patients aggressive monitoring and treatment of intracranial pressure is warranted.
• Early identification and evacuation of operable intracranial hematomas can be life-saving.
• Change in the level of consciousness is one of the hallmarks of brain injury. Its recognition through careful history, physical examination, and close monitoring is therefore essential.
• Prophylaxis and aggressive treatment of seizures following head injury must be instituted.
• Maintenance of normal fluid and electrolyte balance and early provision of enteral nutrition is necessary.
• Risk of central nervous system infection is reduced by surgical débridement and restoration of dural integrity in cases of open cranial injury.
• Systemic sepsis often complicates recovery and should be diagnosed and treated aggressively.

The incidence of head injury in the United States is approximately 200 to 400 per 100,000 population per year. Similar incidence rates have been documented for other countries. Male-to-female incidence ratios vary between 2:1 and 3:1, and incidence peaks in the second and third decades of life.1 Patients with severe head injuries—i.e., those admitted in coma, with a Glasgow Coma Scale score (GCS) of 8 (Table 93-1)2,3—constitute a minority of head-injured patients admitted to hospital, but account for most of the morbidity and mortality (Fig. 93-1). Hospitals with a large primary care population see relatively fewer severely head-injured patients than do those functioning primarily as tertiary referral centers.

The most common cause of closed head injury is road traffic collisions. These include injuries to vehicle occupants, pedestrians, motorcyclists, and bicyclists. Falls are the next most common cause of injury. Gunshot injuries are a major cause of penetrating head injury in the United States and account for up to 44% of head injuries in some series.1 Etiology varies considerably with local patient demographics and proximity to major highways, among other factors, and the resulting case mix will vary from center to center in terms of intracranial hematoma incidence, mean patient age, and consequently outcome from injury. Younger adults are more often involved in vehicular trauma, and there is a lower incidence of intracranial mass lesions in this patient population. Older patients are more often injured as a result of falls and have a higher incidence of mass lesions. At an equivalent depth of coma, both increasing age and presence of intracranial hematomas predispose patients to a poorer outcome.

Intoxication with alcohol or other agents is a significant factor in all causes of injury and across virtually all age groups except the very young and very old. Depending on etiology, 10% to 50% of our patients were obviously intoxicated at admission (Fig. 93-2).

The social and economic costs of head injury are enormous. Severe injury carries a high mortality, and patients who survive severe and moderate injuries may remain substantially disabled for years after the injury. The persistent effects of head injury on personality and mentation may be devastating for both patients and their families.

Acceleration-Deceleration Injury

Injury to the brain is produced by energy transfer to the cranium and its contained structures. In blunt injury, angular acceleration-deceleration forces cause shear strains within the cerebral parenchyma, which in turn are responsible for the characteristic diffuse axonal injury. The force may be applied directly to the head (i.e., impact injury) or indirectly via the body (i.e., impulse injury). The physiologic hallmark of diffuse injury to the brain is loss of consciousness. Concussion is a brief loss of consciousness without obvious sequelae. Coma is defined as a state of unconsciousness in which the patient does not open the eyes, follow commands, or utter any recognizable words. The severity of diffuse brain injury may be gauged by the depth and duration of coma or the duration of posttraumatic amnesia. The duration of both coma and posttraumatic amnesia can be judged only in retrospect, so the depth of coma as quantified by the GCS has been widely adopted as the most convenient measure of injury severity. With severe and prolonged coma, the patient may reach a state characterized by observable periods of eye opening and sleep without appreciation of the environment or of people or objects in it. This has been called the persistent vegetative state.

Acceleration-deceleration injury occurring as a result of impact or impulse mechanisms may produce gross disruption of brain tissue (laceration and/or contusion), diffuse axonal injury, or both. The characteristic histologic findings of diffuse injury consist of axon retraction balls (the pathologic marker of axonal separation) and microglial stars. Experimental evidence, and more recently human postmortem material, have shown that axon retraction balls develop over hours from initially intact axons.4 Hemorrhagic lesions of the anterior corpus callosum may be seen on careful postmortem examination or may be apparent on computed tomography (CT) scans, as may small punctate hemorrhages deep in the hemispheres or small amounts of blood in the ventricular system (Fig. 93-3). For a given force, the greatest degree of injury occurs in the white matter of the hemispheres, and the brain stem is less severely injured.

Laceration/contusion of the brain surface may underlie the site of a blow (coup contusion), but more typically is found on the orbitofrontal surface of the frontal lobes and/or the anterior portion of the temporal lobes, remote from the site of the blow (contrecoup injury; Figs. 93-4 and 93-5). To-and-fro motion of the brain over the uneven surfaces of the anterior and middle fossae has been postulated as the underlying mechanism of contrecoup contusion. Cavitation of the brain produced by the rapid acceleration of the skull away from the brain at contusion sites is an alternate hypothesis. Both cerebral parenchyma and blood vessels are disrupted at the sites of contusion, and the bleeding may be sufficient to produce a subdural hematoma, a large confluent intracerebral hematoma, or both.

In addition to the disruption of cerebral parenchyma, direct impact injury may disrupt the skull and meninges or their blood vessels. Skull fractures are of importance primarily because they may trigger epidural bleeding, provide a portal of entry for infective organisms if compound (either directly or indirectly via the middle ear or paranasal sinuses), or lacerate the meninges and underlying brain if depressed fragments are present. Cranial nerves may be injured as they exit the skull base, and fractures involving the petrous temporal bone may injure the auditory and/or vestibular end organs, or disrupt the ossicles of the middle ear.

Penetrating Injury

Penetrating injury produces direct disruption and laceration of brain tissue. In low-velocity injuries (stab wounds or impalements), damage is confined to the directly disrupted tissue. There is often no loss of consciousness. In missile injuries, cavitation occurs along the track of the missile, and depending on the size and velocity of the missile, disruption of surrounding cerebral tissue is often widespread and severe. Both high- and low-velocity penetrating injuries disrupt the overlying skin, skull, and meninges of the brain, thereby allowing contamination of cerebrospinal fluid (CSF) or brain with infective microorganisms.

Secondary Injury to the Brain

Direct and immediate disruption of the brain tissue by blunt or penetrating trauma constitutes primary injury. Unfortunately, despite encouraging results in the elucidation of the mechanism of primary impact injury and the development of potential neuroprotective therapies in the experimental laboratory over the last decade and a half, effective treatment for primary impact injury still eludes us. Large multicenter trials of pharmacologic agents (including antioxidant drugs and excitatory amino acid antagonists) and hypothermia have failed to yield positive clinical results.5 Clinical management of patients with traumatic brain injury continues to be grounded in the successful prevention, detection, and treatment of secondary insults to the injured brain. These secondary processes may be initiated at the time of injury or at any subsequent time, and these may aggravate existing injury to the brain. Secondary injuries include intracranial bleeding, hypoxia, ischemia, raised intracranial pressure (ICP), infection, and electrolyte and metabolic disturbances.6 The incidence of secondary injury generally increases with the severity of the primary injury, although the relationship is not completely congruent. Patients with devastating primary injuries may initially have very little secondary injury. Conversely, patients with little primary injury may die or be crippled as a result of an enlarging intracranial hematoma. Clinically, the neurologic disturbance produced by primary injury is maximal at onset and then lessens or remains stable. By contrast, secondary injuries produce worsening of the patient's clinical neurologic status as their effects are added to those of the primary injury. Of considerable importance is the fact that secondary injuries are treatable and/or preventable.

Intracranial hematomas occur between the skull and dura mater (epidural; Fig. 93-6), between the dura mater and brain (subdural; Fig. 93-7), or within the substance of the brain (intracerebral; Fig. 93-8). As intracranial hematomas enlarge, they produce compression and shifting of the adjacent brain. This enlargement produces rapid and large increases in ICP as the volume-buffering capacity of the cranial contents is exhausted (see below). Hypoxia may occur as the result of central depression of respiration consequent to loss of consciousness, chest injury, aspiration of gastric contents, neurogenic pulmonary edema, or any combination of the above.

Evidence of ischemia is frequently found at autopsy following blunt head injury.7 Focal ischemia may occur at the site of contusions or may be the result of vasospasm induced by traumatic subarachnoid hemorrhage. Global ischemia may be induced by systemic hypotension or reduction in cerebral perfusion pressure secondary to increased ICP.8 Systemic hypotension at the time of admission to hospital has been shown to increase the mortality from severe head injury by approximately 150%.9

Infection may be local (e.g., meningitis or brain abscess) or systemic. The former occurs when the integrity of the meninges is compromised as a result of penetrating injuries or compound fracture of the vault or skull base. Meningitis or ventriculitis may be iatrogenic, as a result of insertion of ICP-monitoring apparatus. Systemic infection typically involves the lungs or genitourinary tract. Systemic sepsis often causes neurologic deterioration that may improve with resolution of the infection. The accompanying hyperthermia often aggravates existing ICP problems.

Electrolyte disorders are common in patients with moderate and severe head injuries. The most frequent is the syndrome of inappropriate antidiuretic hormone secretion (SIADH). Serum sodium may fall significantly, and deterioration in patients' neurologic status usually occurs at levels below 120 to 125 mEq/L. Diabetes insipidus usually occurs in more severe injuries, often as a preterminal event in patients with progressive rostrocaudal neurologic deterioration caused by uncontrollable ICP elevation. In a much smaller number of patients, diabetes insipidus may be the result of focal injury to the hypothalamic-pituitary axis.

Severe head injury induces a state of catabolism with net excretion of nitrogen similar to that seen in other injuries. Catecholamine-induced tachycardia, increased blood pressure, and increased cardiac output are common. Myocardial ischemia may occur as a result of closed head injury.10

Abnormalities of blood coagulation are known to occur as a result of brain injury and are predictive of a poor outcome.11 The reported incidence of coagulopathy depends to a certain extent on the diligence with which this entity is sought; its presence is often obscured by that of other potential causes of coagulation disturbance, such as massive transfusion in the multiply injured patient. Delayed and recurrent hematomas, including those related to placement of ventricular catheters, have been described in patients with disseminated intravascular coagulation.12 In patients in whom a bleeding diathesis is known or suspected, it is preferable to use a subarachnoid bolt or epidural transducer rather than a ventricular catheter for ICP monitoring when there is a high likelihood of increased ICP (e.g., in the presence of a mass lesion). In patients who are known to have a low risk of raised ICP (i.e., those with normal CT scans), monitoring can be deferred until the coagulation abnormality is reversed.

Intracranial Pressure

Over the past 15 years, monitoring and control of ICP have been hallmarks of aggressive management of traumatic brain injury, and their use has been increasingly common in management of other cerebral disorders such as subarachnoid hemorrhage (SAH), intracerebral hematoma (ICH), brain tumors and liver disease. ICP monitoring and control are routine features of intensive care in most major neurosurgical centers and are done commonly in community hospitals. Although level one evidence of their efficacy is lacking, there is a substantial body of data to support their use and growing application.13–16 Consensus guidelines derived by a panel of experts in the care of neurotrauma patients recommend the use of ICP monitoring in severely head-injured patients.17 In this chapter, ICP monitoring will be discussed in the context of traumatic brain injury, but the techniques of monitoring and treatment are equally applicable to other conditions.

Since Lundberg18 described ICP monitoring in 1960, its use in adults has been mainly in patients with severe head injury, both because of the large number of such patients and because intracranial hypertension is common when the injury is severe. We cannot tell whether a bad outcome in an individual patient results directly from the head injury, with brain swelling and increased ICP being epiphenomena, or whether it stems from secondary brain injury caused by cranial hypertension. In the first case, no remedy is possible. Most neurosurgeons assume and hope that the second case is actually present—i.e., that intracranial hypertension is adding to brain injury—and believe that ICP elevations should be detected and controlled. Any experienced neurosurgeon can relate anecdotes of cases in which detection and treatment of elevated ICP reversed neurologic deterioration and contributed to a good outcome. For example, detection of a delayed expanding intracranial hematoma by observation of rising ICP can lead to more prompt, and possibly life-saving, clot removal.

ICP is the pressure measured from some intracranial site and is generally recorded in millimeters of mercury. The site of measurement is usually in the cerebral ventricles or in the subdural or epidural space, or rarely in the brain parenchyma. While the brain is reasonably plastic, and the ICP measured at one place within the skull reflects fairly well the pressure at any other place, pressure gradients may develop across semirigid dural partitions in the skull when pathology is localized in one compartment (e.g., supratentorial vs. infratentorial). The combination of brain shift from focal masses, high intracranial pressure, and pressure gradients across dural and bony partitions (i.e., falx cerebri, tentorium, and foramen magnum) causes brain tissue to herniate from compartments with higher pressure to compartments with lower pressure. The most important of these cases is transtentorial herniation, signaled by the presence of an ipsilateral dilated and fixed or sluggishly reactive pupil. Intracranial hypertension is clearly potentially injurious. While the precise point at which ICP should be considered abnormal is not agreed upon and undoubtedly varies among different pathologies, ICP levels above 20 or 25 mm Hg that have been present for 10 to 15 minutes generally are considered pathologic and treatment is recommended.17

Cerebral perfusion pressure (CPP) represents the difference between mean arterial pressure (MAP) and ICP (CPP = MAP − ICP). In normal brain, cerebral blood flow (CBF) remains fairly constant at around 50 mL/100 g brain tissue per minute as long as cerebral autoregulation is intact—i.e., cerebral vessels change their diameter and resistance with pressure changes, so that a constant CBF is maintained as long as CPP is about 50 to 150 mm Hg. If MAP falls substantially or ICP rises substantially, then CPP falls too low to be corrected by autoregulation, and cerebral ischemia results. In cases of uncontrolled ICP, the terminal event is usually an increase of the ICP to a level at or slightly above the MAP. In severe head injury, there is some evidence of an autoregulatory breakpoint at a perfusion pressure of about 70 mm Hg.19 In the case of marginal CPPs of around 50 to 60 mm Hg, one must remember that the focal CPP can be lower than the overall CPP if there are focal brain lesions, and a more generous CPP of 70 to 80 mm Hg may be necessary to prevent focal ischemia. A treatment protocol has been proposed (supported by uncontrolled data) wherein the emphasis on treatment resides primarily on elevating and maintaining the CPP above 70 mm Hg, often with systemic vasopressors with less emphasis on managing ICP in isolation.20 A more recent review of ICP and CPP data from a large clinical trial in head injury suggests that it is ICP alone, rather than CPP, that is the more critical variable in determining outcome.21 A randomized prospective trial of ICP- vs. CPP-targeted treatment showed a reduction in episodes of jugular venous oxygen desaturation (a surrogate measure for cerebral ischemia), but failed to show any beneficial effect on overall neurologic outcome. Furthermore, there was a fivefold increase in the incidence of acute respiratory distress syndrome in the CPP-targeted treatment group.22

One particularly dangerous situation for patients with severe head injury is when the hemisphere subjacent to a large subdural or epidural hematoma undergoes massive swelling after the hematoma is removed. The mass effect of the hematoma undoubtedly causes severe unilateral ischemia, but this effect is dramatically worsened by the hypoxia or hypotension that are present in about half the patients who develop this particularly severe postoperative complication. Even if ICP is normal shortly after the initial operative procedure, intracranial hypertension occurs in virtually all of these patients and may be uncontrollable and fatal in a high percentage.23

A small percentage of patients with intracranial hematomas develop delayed traumatic hematomas not seen on the initial CT scan, some after previous surgery for extracerebral hematomas and sometimes more than a day after injury.24 ICP monitoring and recognition of intracranial hypertension can lead to earlier repeat CT and earlier identification and removal of clots than can reliance on the neurologic examination alone,25 although earlier clot removal has an uncertain effect on outcome.26

As may be inferred from the discussion of pathology and pathophysiology presented earlier, the management of head injury during the acute period is based primarily on the timely diagnosis and treatment of secondary insults to the injured brain. In the continued absence of effective drugs or other treatment strategies to reverse the pathophysiologic changes of diffuse axonal injury, management of the primary injury consists of providing the best possible physiologic milieu to permit unimpeded recovery of sublethally injured neurons.

Diagnosis

Diagnosis in closed head injury is based on history, physical examination, and radiologic investigation. The relevant facts to be ascertained from the history are the mechanism of injury, whether there was any initial loss of consciousness, and whether the level of consciousness has improved or deteriorated since the injury. This last item is of paramount importance, since worsening of the patient's neurologic state subsequent to injury always implies the presence of secondary injury to the brain. Patients who have talked at some point after injury and who subsequently lapse into unconsciousness almost invariably are harboring an intracranial hematoma. One should determine whether or not there has been any witnessed seizure activity, since seizures may produce transient profound deterioration in the level of consciousness that may mimic that induced by an expanding intracranial hematoma.

Neurologic examination consists of two components: determination of the level of consciousness and identification of any focal deficits to aid in lesion localization and the measurement of severity. Determination of the level of consciousness is the single most important measure to be taken following traumatic brain injury and is best accomplished by use of the Glasgow Coma Scale (GCS). Although the scores in the three response categories of the GCS are frequently summed for convenience, more information is conveyed by reporting them separately. The motor score from the patient's “best” side is used if there is a discrepancy in response between sides. Intubation may be designated by a “T” for the verbal response score. In patients with a disturbed level of consciousness, the scope of the traditional neurologic examination is severely limited because of its considerable dependence on patient cooperation. In unconscious patients, crude localization may be achieved by observing discrepancies between sides in the patient's motor response to pain. Evidence of brain stem dysfunction should be sought; the common measures are pupillary reaction to light, oculocephalic responses (doll's-eye movements) or oculocaloric responses, and corneal reflexes. Doll's-eye testing should be deferred until the cervical spine is known to be intact, and oculocaloric testing should not be carried out in the presence of a ruptured tympanic membrane. Unless there has been local ocular trauma, the presence of unilateral pupillary dilation and unreactivity implies the presence of a mass lesion, and urgent diagnosis and evacuation of the mass are essential. This is particularly important in cases of unwitnessed unconsciousness, and investigation of metabolic causes of coma should be deferred until a CT scan has been obtained.

Another relevant aspect of the physical examination is examination of the head for signs of trauma. These include bruising or laceration of the face and scalp; open skull fractures; hemotympanum and bruising over the mastoid process (Battle sign), indicating fracture of the petrous temporal bone; and periorbital hematoma (raccoon eye), indicating fracture of the floor of the anterior fossa. One should also look for signs of CSF leakage from the nose or ears. CSF may be mixed with blood; if the draining fluid forms a target shape on a piece of filter paper, the presence of CSF is indicated. Biochemical determination of the presence of CSF is usually not possible because of the difficulty in collecting a sufficient volume of the draining fluid.

CT is essential to the proper management of unconscious head-injured patients. It makes it possible to diagnose all types of intracranial hematoma and to determine their location and the extent of the mass effect they produce (measured by the displacement of the septum pellucidum from the midline). The scan should be obtained as soon as the patient is stable and any more urgent management priorities have been addressed. It is occasionally necessary to forego CT to deal with life-threatening injuries to other systems or because there would be an unacceptably long wait for a scan in a patient with a high likelihood of having an intracranial surface clot. Unconscious patients should have radiologic assessment of the cervical spine to the level of T1 to rule out cervical fracture. In the absence of x-ray evidence of an intact cervical spine, one must presume there is a fracture and continue to immobilize the head and spine until appropriate x-rays are obtained and the necessary treatment instituted.

Magnetic resonance imaging (MRI) has been used on a limited basis with head-injured patients. MRI is better than CT at detecting some types of small extra-axial fluid collections and parenchymal injuries. However, it offers no advantage in detecting hematomas requiring surgical treatment, and the difficulties inherent in having to exclude metallic objects from the scan environment render the technique somewhat impractical in the acute posttrauma setting. A potential application is the improved identification of parenchymal changes following mild and moderate head injury that have been shown to correlate with the neuropsychologic sequelae of head injury.27,28

Preoperative Management

In the initial phase of head injury management, the usual diagnostic and treatment priorities for the management of trauma patients are followed. Establishment of an adequate airway is of paramount importance in unconscious patients. All patients with a GCS ≤8 should have endotracheal or nasotracheal intubation and mechanical ventilation, regardless of whether other indications for intubation are present or not. Direct laryngoscopy can raise ICP dramatically, potentially worsening cerebral ischemia. Intubation should be performed by experienced personnel and after appropriate premedication (see Chap. 35).29 Nasotracheal intubation should be avoided in patients with signs of basal skull fracture. Patients should be hyperventilated to a modest degree (partial pressure of carbon dioxide [PCO2] 30 to 35 mm Hg) to reduce raised ICP. Patients with a clear-cut deterioration in neurologic status should be given mannitol in a dose of approximately 1 g/kg body weight. Contraindications to the use of mannitol are hypotension, anuria, and severe congestive heart failure. Corticosteroids are of no benefit in reducing ICP in head injury, and their use has been abandoned in most centers. As soon as other management priorities have been dealt with, the patient should undergo CT.

Mild therapeutic hypothermia was first proposed as a treatment for brain trauma in the late 1950s. Hypothermia can lower ICP, alter chemical pathways that could contribute to injury, and modulate apoptosis. In 2001 the results of a large multicenter trial were published, demonstrating that hypothermia (target = 33°C) was not effective in improving functional outcome at 6 months, even though ICP was better controlled.30 A more recent study raises again the possibility that hypothermia may confer benefit. Patients with severe head injury (GCS score ≤8) who continued to have intracranial hypertension (ICP >20 mm Hg) despite standard resuscitation, attention to the adequacy of CPP, surgery (when indicated), sedation, mannitol, therapeutic paralysis, and pentobarbital infusion, were cooled to 32°C.31 Hypothermia was maintained for at least 24 hours or until slow rewarming did not cause the ICP to rise (mean duration of cooling = 4.8 days). Hypothermic subjects were less likely to die (62% vs. 72%), but only a small subgroup of patients with a GCS of 5 to 6 accounted for the majority of the response difference (12/25 hypothermia subjects survived vs. 6/25 control subjects). Given these conflicting data, as well as the fact that hypothermia is potentially harmful, leading to hypovolemia, hypotension, and electrolyte disturbances, this intervention should be considered investigational.

Operative Management

Intracranial Hematoma

The cornerstone of acute head injury management is the rapid diagnosis and prompt surgical evacuation of intracranial hematomas. Twenty-five to fifty percent of patients rendered comatose as a result of acute head injury have an operable intracranial hematoma.3,6,14,32,33 Approximately 20% of patients with intracranial hematomas have an extradural hematoma, and the rest have hemispheric acute subdural hematomas, significant brain contusions, or both.6,32,33 Traumatic hematomas in the posterior fossa are unusual. The diagnosis is ordinarily made by CT. Surgical evacuation is then accomplished via craniotomy, with wide exposure and evacuation of the hematoma and coagulation of sources of hemorrhage. Rarely, when there is clear-cut evidence of neurologic deterioration with progression to signs of uncal herniation (i.e., pupillary dilation), it is necessary to proceed directly to operation, bypassing CT. The other circumstance in which exploratory burr holes may be necessary is in comatose patients with signs of brain stem compromise who must be taken to the operating room for life-threatening systemic injuries without benefit of a CT scan. An alternate strategy in these patients is twist-drill air ventriculography in the operating room. The standard landmarks are used for cannulation of the frontal horn of the lateral ventricle. Once the ventricle is cannulated, 4 to 8 mL of air is introduced into the ventricles, and a brow-up anteroposterior skull film is obtained. The presence of a mass lesion is inferred from shift of the ventricular system (Fig. 93-9).

The frequent coexistence of acute subdural hematomas and temporal and/or frontal cortical laceration/contusion dictates the necessity of a wide operative exposure, allowing access to both the frontal and temporal lobes, as well as the medial convexity adjacent to the sagittal sinus, where there may be torn bridging veins (Fig. 93-10). The operative approach to an isolated extradural hematoma may be more circumscribed. Contusions of the frontal and temporal lobes, or confluent blood clots producing a midline shift greater than 5 mm, should be evacuated operatively through the standard large flap. Often, removal of the contused brain must be accompanied by formal frontal or temporal lobectomy to provide adequate internal decompression in the face of a traumatized, swollen brain. In this circumstance, the question of whether or not to leave out the craniotomy bone flap (or remove it later) to provide more room for the swollen brain has engendered some controversy. The prevailing view is that this measure does not improve the functional outcome. It is probable, however, that poor outcomes with this procedure are usually due to the fact that the patient had overwhelming primary injury, and that decompressive craniectomy may be useful in patients in whom the degree of primary injury is compatible with a reasonable neurologic outcome.

At the conclusion of the evacuation of a hematoma, an ICP monitoring device is placed in almost all cases. The exceptions are patients whose level of consciousness was only mildly depressed preoperatively (GCS >10), and in whom the brain is not significantly contused or swollen at the time of operative exposure. The techniques and indications for ICP monitoring will be discussed.

Penetrating Injury

The same principles of prompt evacuation of intracranial hematomas and débridement of contused brain apply to penetrating injuries. The other indication for operating on compound lesions is the restoration of the integrity of the dura and overlying tissues to prevent bacterial infection of CSF (i.e., meningitis) or cerebral parenchyma (brain abscess). Accessible indriven bone fragments should all be removed, as well as underlying hematoma and devitalized brain. It is not necessary to attempt to remove inaccessible missile or bone fragments from the depths of the missile tract. Prophylactic antibiotics are used to reduce the incidence of delayed infection in cases of penetrating injury. Broad-spectrum coverage with antibiotics capable of crossing the blood-brain barrier should be used.34

Basal Skull Fracture

A special case of compound injury is the basal skull fracture with leakage of CSF through the paranasal sinuses or into the mastoid air cells, middle ear, and external ear or pharynx. CSF otorrhea almost always stops spontaneously, and surgical repair of the leak is rarely necessary. Rhinorrhea is more likely to continue, and the continued presence of a CSF fistula will eventually lead to meningitis. Surgical repair of an anterior fossa fistula may be undertaken early when there is an obvious bony disruption of the anterior fossa floor, often coincident with complex facial fractures. In this case, repair of the CSF leak is undertaken at the time of facial reconstructive surgery. In cases of persistent CSF leak without obvious bony disruption, the site of the fistula may sometimes be localized with coronal CT cuts through the anterior fossa following the instillation of water-soluble radiographic contrast material into the CSF. Investigation of persistent CSF leakage should be deferred until the patient is stable with respect to both ICP and coexisting systemic injuries. Occasionally, bilateral anterior fossa exploration is necessary in the face of persistent CSF rhinorrhea for which the site cannot be identified radiologically.

There is considerable debate about the merits of prophylactic antibiotics for CSF rhinorrhea and otorrhea. Unfortunately, there are no good clinical studies to support either their use or their abandonment, and this remains largely a matter of personal preference. We do not use prophylactic antibiotics for CSF leakage in our unit for fear of selecting for multiply resistant organisms. When there is established infection, the diagnosis and treatment are the same as for meningitis of any cause. Operative repair of the dural fistula should not be undertaken until the infection has been cleared with appropriate antibiotic therapy.

Critical Care Monitoring

Monitoring of the head-injured patient is designed to detect and treat secondary injuries before they further compromise the injured brain. Ideally one would like to monitor neurologic function, cerebral blood flow and cerebral metabolism, intracranial pressure, and those systemic parameters required for the maintenance of adequate perfusion and oxygenation of the brain. Measurement of all these parameters is possible, but cerebral blood flow and cerebral metabolism are difficult to measure, and the ideal situation described here is rarely practiced.

Clinical Examination

Monitoring of neurologic function is most commonly and simply carried out by means of hourly repetition of the abbreviated version of the neurologic examination described earlier, in the section on diagnosis. Any deterioration in the patient's neurologic status should trigger an investigation for treatable causes. Neuromuscular blocking agents used to treat raised ICP or respiratory problems make it impossible to conduct a meaningful neurologic examination. In addition, these drugs are associated with numerous complications and may be associated with a worse outcome in head-injured patients.35 When neuromuscular blocking agents are given, only the pupillary responses are available as a means of assessment. Because significant deterioration in the patient's neurologic function may occur before pupillary changes are evident, ICP monitoring becomes essential to provide some information about the status of the patient. Although ICP is not a measure of neurologic function per se, many of the pathologic processes that endanger the head-injured patient in the early phase after injury manifest themselves with raised ICP in addition to neurologic deterioration. In a patient who is pharmacologically paralyzed for ICP control, paralysis should be allowed to wear off at intervals of 12 to 24 hours so that a neurologic assessment may be carried out. Occasionally the effects of reversal of ICP are intolerable, and reversal is not possible. A potential alternative to clinical neurologic monitoring is electrophysiologic measurement of neurologic function. The common techniques are electroencephalography (EEG) and evoked potentials (EPs).

Electroencephalography

EEG recordings in comatose patients typically show slowing of the background frequencies; in most cases there is a correlation between the degree of slowing and the depth of coma. Conventional strip-chart EEG is impractical because of the amount of data generated. Various methods of data compression have evolved, the most common being fast Fourier transformation of the EEG. Fourier transformation yields quantitative frequency and amplitude data from the EEG, and is well suited to detecting long-term trends in the background frequencies and to performing quantitative analysis of such trends. Recognition of specific patterns (e.g., seizure spikes) is sacrificed. Survival from head injury has been correlated with a return of higher-frequency activity in the first week after injury.36

Evoked Potentials

EPs have been demonstrated to be particularly useful in predicting outcome from head injury. The single most useful measure is the somatosensory evoked response,37 although combinations of EPs (somatosensory, auditory, and visual) yield the highest prognostic accuracy.38 Deterioration in serially measured EPs, occurring as a result of secondary injury to the brain, have been shown to correlate with poor patient outcome.39 In the past, EPs have been used primarily in an intermittent manner for determination of prognosis. With further technologic improvements and increased automation of data collection, these techniques may see more widespread use for continuous monitoring in an ICU setting.

ICP Monitoring

Numerous reports suggest a beneficial effect of ICP monitoring and treatment of intracranial hypertension, although there are no controlled trials to support this. In a group of patients with GCS scores of 3, 4, or 5, the mortality rate was 39% when ICP was monitored and 67% when it was not; however, the patients were not randomly allocated to the two groups.40 In another series, using patients evaluated in sequential time periods with slightly different treatment protocols, the mortality rate fell from 46% to 33% when ICP was rigorously controlled below 15 mm Hg.14 In as many as 75% of patients who die after head injury, there is pathologic evidence of intracranial hypertension as measured by medial temporal lobe necrosis, secondary brain stem damage, or cerebral ischemic changes.41 These and other reports have convinced most neurosurgeons that ICP monitoring is appropriate in patients with traumatic coma.

Elevated ICP clearly correlates with a poor outcome in head-injured patients. In one series of patients with severe head injury, 74% of the patients whose ICP remained below 20 mm Hg had a satisfactory outcome, while only 55% of the patients who required therapy to control ICP to this level had a similar outcome. Of the patients whose ICP could not be controlled, 92% died, and only 3% had a satisfactory outcome.32 In another series, the number of patients who died or remained institutionalized more than doubled when ICP was higher than 15 mm Hg (57% versus 23%) than when it remained below that level.42 In addition to its effects on survival, intracranial hypertension affects brain function in survivors. Neuropsychologic assessment done 1 year after severe head injury revealed that hyperemia during the acute phase correlated better with poor intellectual and memory function than did low-flow states, and that patients who had experienced intracranial hypertension had more memory deficits than those whose ICP had remained normal.43

A variety of monitoring techniques have been used over the past 30 years. Lundberg first used a ventriculostomy coupled to a fluid-filled manometer.18 This system was cumbersome and carried a considerable risk of causing infection. As electronic transducers became available, closed monitoring systems came into use. Although a few implantable transducers have been used experimentally, they are not in regular clinical use today. The two most common ICP monitoring systems used now are fluid-filled catheters and rigid bolts44 attached to an electromechanical transducer or a fiberoptic system that detects deflection of a sensing mirror to measure ICP.45 The former devices are generally less expensive, but are prone to blockage and may require more experience on the part of the clinician to keep them functioning optimally. The latter may provide more accurate data for longer periods without the need for intervention by doctors or nurses. The fluid-filled systems may be more appropriate in hospitals where ICP monitoring is common and is applied in several patients simultaneously; in such settings, medical personnel become familiar with the system and can recognize and correct problems as they arise. The fiberoptic systems may be more appropriate for facilities where fewer patients require ICP monitoring.

We currently use either subdural or intraventricular catheters for monitoring ICP, which we place either in the operating room at the time of surgery or at the bedside in the ICU (Fig. 93-11). We prefer an intraventricular position, so that CSF can be drained when necessary to lower ICP. An intraventricular catheter can be placed through a craniotomy opening in the operating room or by twist-drill ventriculostomy in the operating room or ICU when the ventricles seem to be large enough to permit entry. When the ventricles are compressed by brain swelling, we prefer to use a subdural catheter, which can only be placed in the operating room. (Because a twist-drill hole is perpendicular to the skull and brain, it precludes catheter insertion parallel to the inner table of the skull over the brain surface.) Placement of an ICP device requires meticulous sterile technique; if any part of the system is contaminated during insertion or attachment, it must be discarded and replaced. Intracranial infection is a devastating complication when added to the brain dysfunction for which monitoring is required. ICP monitors should be placed by neurosurgeons, since they are most familiar with manipulating brain tissue and are least likely to induce brain hemorrhage, a risk that is always present and that can be fatal. Epidural catheters must be placed through a burr hole in the operating room and are used less frequently than subdural or intraventricular monitors.

We attach the ICP catheter to pressure transducers connected to monitors, which provide continuous digital and waveform displays of ICP as well as of arterial, central venous, or pulmonary vascular pressures as needed in either the operating room or the ICU. The data can be sampled by a central ICU computer so that these and other parameters can be analyzed over a few hours or days. Usually a good pressure tracing is easily recognized, having a regular pulse pressure of 3 to 7 mm Hg (the pulse pressure can be greater at elevated ICP). The ICP baseline gently drifts up and down with respiratory or ventilatory excursions as intrathoracic pressure changes cause slight alterations in the intracranial venous pressure and ICP. A flat ICP tracing cannot be interpreted; a grave error in ICP monitoring is accepting a meaningless digital readout number as the actual ICP when the tracing clearly does not reflect a valid waveform. A flat tracing usually is due to blockage by blood or brain occluding the catheter tip or to a kink somewhere along the catheter. After the extracranial portion of the monitor is inspected for proper placement, it may be necessary to inject a small amount (0.25 mL initially) of an irrigating solution to clear the catheter tip; a larger amount, even only 1 mL, can markedly increase ICP when the brain is tight. It is helpful to turn the system stopcock from the injection position back to the monitoring position while the irrigant is being injected. If the stopcock is turned after the injection is completed, brain under pressure may immediately occlude the catheter tip when the injection pressure is stopped. Even if a catheter reoccludes and produces a flat waveform, a valid estimate of the ICP usually can be made. After a small amount of irrigant is injected, the ICP will be briefly elevated and have large pulse pressures, but will drift asymptotically toward the true ICP before becoming a flat line again.

The ICP catheter and transducer are filled with an antibiotic solution containing gentamicin, 1 mg/mL, which is stable in solution for 4 to 6 days and has wide antibacterial activity. It is important to maintain a closed system as much as possible; this requires tight connections throughout the monitoring system. It is imperative that no CSF leak around the catheter; any such leakage greatly increases the risk of infection. If leakage is noted, a new purse-string suture must be placed around the catheter's exit from the scalp. If the monitoring device is sufficiently flexible, it should be brought out through a stab wound a short distance from the incision made to insert it to further reduce infection risk. Our intracranial infection rate for treatment of a variety of open or penetrating brain injuries is <2%, so the infection rate attributable to ICP monitoring is extremely low. Infection rates vary from 0.7% to 6.3% in other series using different devices.46,47 In 378 patients, several different fluid-filled systems for monitoring ICP were compared regarding their tendency to become blocked or infected. The rigid Richmond bolt42 became occluded 16% of the time; subdural or intraventricular catheters became blocked only 3% of the time, but the intraventricular catheters were associated with complications of infection or intracerebral hemorrhage, and led the authors to recommend the subdural catheter as the best of the methods.48 However, intraventricular catheters allow withdrawal of CSF, which can be a critical method of ICP control. Some prefer to measure the change in ICP produced by withdrawal of a measured volume (ΔV) of CSF to determine the intracranial compliance (ΔV/ΔICP); we have not found this additional measurement to provide a sufficient increment in useful data to warrant the time and risk involved.

Blood Flow and Metabolic Rate

Cerebral blood flow and cerebral metabolism have been measured in a number of units engaged in clinical research. The common techniques rely on measurement of washout curves from the brain of nitrous oxide or radioactive xenon. The equipment and techniques for conducting CBF measurements are complicated and expensive, and this fact has discouraged widespread adoption of these methods. Furthermore, in the case of xenon CBF measurement, the number of studies that can be carried out in a given patient is limited by radiation safety concerns. The incidence of ischemic levels of blood flow is highest within 12 hours of severe injury, and blood flow usually increases to normal levels by 24 hours.49 Knowledge of CBF may be useful in identifying patients in whom hyperemia is contributing substantially to increased ICP, and thus to determine if treatment with hyperventilation may be used safely.CBF, the cerebral metabolic rate for oxygen (CMRO2), and the arterial-jugular oxygen content difference (AVDO2) are related by the following equation:

CMRO2 = (AVDO2 × 3 CBF)/100

When CMRO2 is relatively constant, AVDO2 varies inversely with CBF; therefore, it has been suggested that AVDO2 could be used as an index of the adequacy of cerebral blood flow.50 The jugular oxygen content may be measured by advancing a central venous catheter rostrally in the internal jugular vein upward to the base of the skull. The normal AVDO2 is 4 to 9 vol%. Higher values indicate increased oxygen extraction and potential ischemia. More recently, fiberoptic catheters have been developed that will give a continuous readout of jugular venous oxygen saturation. Episodes of desaturation have been correlated with a poor outcome.51,52

Hemodynamics

Hemodynamic monitoring in the acute phase of head injury management should include arterial pressure monitoring and detailed measurement of fluid input and output. When repeated large doses of mannitol are administered, invasive hemodynamic monitoring may be necessary to maintain an accurate assessment of intravascular volume status in the face of massive diuresis and consequent fluid replacement (see below).

Medical Management

Manipulating Intracranial Pressure

The medical management of acute head injury focuses primarily on the detection and treatment of raised intracranial pressure. Although the ICP is less than 10 mm Hg in normal individuals, 20 to 25 mm Hg is generally accepted as the point at which ICP requires treatment in head injury patients. Transient elevations above this level are common when patients are restless or coughing; therefore, persistence of the ICP elevation for 5 to 10 minutes in a nonagitated patient should be observed before treatment is initiated. In all cases, a cause for the elevation should be sought. There may be a problem with ventilation or obstruction of jugular venous return. Unless contraindicated by other injuries, patients should be positioned head-up 30° with the neck neutral between flexion and extension. If the ICP elevation persists after adequacy of ventilation and unimpeded jugular venous return have been assured, it is incumbent on the physician to rule out a surgically treatable cause of the ICP elevation with a CT scan. In cases of unoperated cerebral contusion, an increase in ICP may signify the need for surgical evacuation of the contused brain.53

Once it has been determined that the ICP increase has no surgically treatable cause, management depends on pharmacologic reduction of ICP and manipulation of the patient's PCO2. If the patient is not ventilated, intubation and mechanical ventilation should be instituted with moderate hyperventilation (i.e., PCO2 30 to 34 mm Hg). Hyperventilation to levels of 25 mm Hg or below has been shown to have an adverse effect on outcome.54 If the patient is already ventilated, sedation with morphine or an analog should be given. The anesthetic agent propofol has come into increasing use for this indication. A recent randomized clinical trial has shown that it is safe and efficacious for sedation of intubated and ventilated head-injured patients.55 If this measure is insufficient to reduce the ICP, then pharmacologic paralysis should be induced. In patients who are febrile, reduction of the core temperature to 37°C or lower by means of a cooling blanket frequently results in reduction of raised ICP. Often these measures alone are sufficient.

If the ICP elevation persists, and there is a ventricular drain in place, intermittent drainage of CSF may be instituted. It may be necessary to repeat drainage several times an hour. Continuous drainage should not be used, because drainage will often cease as the ventricles undergo progressive compression, and during ventricular drainage the ICP is not being measured. One is then in the unfortunate situation of not reducing the ICP and not being aware of the true ICP. If ventricular drainage is not feasible or effective, then mannitol must be used.

Mannitol, an osmotic diuretic, is the single most useful pharmacologic agent in the management of raised ICP following head injury. The dosage is 0.5 to 1.0 g/kg of a 20% solution infused as a bolus. Typically, the effects of mannitol on ICP will last from 2 to 6 hours, depending on the severity of the underlying pathologic process. Mannitol may be administered repeatedly as long as the serum osmolarity does not exceed 320 mOsm/L. At osmolarities greater than this, the effectiveness of the drug is reduced, and there is an increasing risk of renal toxicity.56 With frequent repeated doses of mannitol and the consequent diuresis, the patient may become systemically dehydrated, resulting in hypotension and increased serum osmolarity. Systemic dehydration is not necessary to achieve reduction in ICP. Therefore, the excess urine loss caused by mannitol should be replaced. The amount and type of replacement fluid should be calibrated to maintain high normal intravascular volume, a serum osmolarity of <320 mOsm/L, and normal serum electrolytes. With this regimen, it may be possible to give mannitol at intervals of 2 hours for several days at a time using aggregate volumes of 2 to 3 L of mannitol per day.

Furosemide can be used to reduce intracranial hypertension and probably works by inducing diuresis, increasing intravascular oncotic pressure, removing brain water, and possibly reducing CSF production.57 The combination of mannitol and furosemide may be more effective in lowering ICP than either drug alone. This synergistic effect is not from an altered renal excretion of mannitol; furosemide may help to sustain the elevated serum osmolarity induced by mannitol, or may sustain the osmotic gradient across the blood-brain barrier induced by mannitol.57 However, some caution is advised, since the combined use of mannitol and furosemide can cause rapid dehydration, hypotension, and a reduced CPP with the risk of brain ischemia.

In situations in which the combination of measures listed above is insufficient to control ICP, or where mannitol can no longer be used because of extant or incipient renal failure, pentobarbital may be a useful adjunct in the management of ICP.58 It is not a first-choice agent, since it is clearly less effective than mannitol in lowering raised ICP,59 may cause harmful cardiovascular side effects, and has no demonstrated protective effects on the brain when administered prophylactically to head-injured patients with or without raised ICP.60 The dose regimen varies in the published literature, and pentobarbital may be given by bolus infusion or by drip. A typical approach is to administer a bolus of 10 mg/kg over 30 minutes, followed by a continuous infusion at 1 to 3 mg/kg per hour. The end point of administration is reduction of ICP or an unacceptable decrease in arterial blood pressure. Adequate hemodynamic monitoring (i.e., an arterial line and central venous catheter) must be in place before pentobarbital is given, and adequate intravascular volume should be assured prior to administration. Infusion of norepinephrine or other vasoactive agents is often necessary to maintain an adequate arterial pressure. Administration of high doses of pentobarbital will result in pupillary dilation and unreactivity, so that no clinical neurologic examination is possible. Brain stem auditory evoked responses (BAEPs) are preserved, and their recording may be a useful adjunct to the management of patients in barbiturate coma.61

The medical management of intracranial hypertension after head injury is generally successful in the vast majority of patients with severe head injury.58,62,63 In one series, when standard medical management was unsuccessful, pentobarbital coma was induced, but it only controlled intracranial hypertension in about 30% of these patients. A subtemporal decompression was then performed, resulting in a significant further reduction in ICP and a 60% survival, compared with <20% survival when subtemporal decompression was not done.62 Other surgical options available for the treatment of refractory ICP include subfrontal decompressive craniectomy, removal of an existing bone flap, or internal decompression by removal of a swollen or damaged frontal or temporal lobe. The controversy surrounding decompressive craniectomy has been alluded to above. These surgical adjuncts perhaps deserve consideration for a very small number of head-injured patients who do not respond to other avenues of ICP control. We have used surgical management in young patients with refractory ICP who have a reasonable potential for neurologic recovery (i.e., GCS ≥7).

Management of Seizures

Seizures may aggravate an existing brain injury; therefore, their prompt treatment is essential. Factors that increase the risk of late epilepsy include severe injuries, intracranial hematomas, and the presence of seizures early after injury. Ordinarily, patients in these risk categories are given prophylactic phenytoin at the time of admission, and the medication is continued for 6 to 12 months, depending on the injury type and on whether or not any seizures have occurred. Prophylactic anticonvulsants have only been shown to be effective in the early posttraumatic period.64 A discussion of acute seizure management is provided elsewhere in this text (see Chap. 64).

Nutrition

The caloric requirements imposed by a head injury are comparable to those of a burn covering 20% to 40% of the body surface area. The requirements are increased by motor posturing and reduced by barbiturate coma or muscle relaxants. Enteral feeding via nasogastric tube is usually possible and is certainly desirable early after injury, unless there has been major abdominal trauma, in which case parenteral alimentation may be used. Although the nitrogen loss resulting from a severe head injury may be mitigated by early feeding, it may not be possible to reverse it consistently.65

Fluid and Electrolytes

SIADH is usually successfully managed by restricting free water intake to 1 liter or less per 24 hours. Demeclocycline may be a useful adjunct when the syndrome persists beyond a few days. Hypertonic saline (3%) is occasionally necessary to correct profound hyponatremia in the presence of serious neurologic symptoms (e.g., seizures). Diabetes insipidus may be managed with desmopressin acetate (DDAVP). The dosage is 1 to 2 μg (0.25 to 0.5 mL) intravenously two to four times daily as necessary to control urine output. Serum and urine osmolarity and electrolyte measurements are necessary to distinguish true diabetes insipidus from excessive diuresis caused by mobilization of fluids used during resuscitation or as a result of the use of mannitol for control of ICP.

Rehabilitation

Attention to the rehabilitation needs of the patient should begin on or shortly after admission to the critical care unit. In the early days after admission, this consists of proper positioning, regular turning, skin care, and movement of patients' limbs through their full range to minimize or prevent late joint contractures and skin decubiti, either of which may significantly retard recovery; more active rehabilitation begins once consciousness has been regained. Sensory stimulation programs to facilitate the return of normal consciousness are attractive in theory, but they have not been tested by randomized clinical trials. As the patient's level of consciousness improves, the goals of rehabilitation therapy change from maintenance of normal limb posture and movement to the retraining of simple and then progressively more complex physical and mental activities. Although at this time the patient will normally be out of the ICU, it is important that these measures be initiated in the ICU.

The mortality from severe head injury varies from 30% to just over 50% in most published series.3,6,14,32,33,40 Factors associated with a decreased likelihood of survival are low GCS, advanced age, the presence of significant intracranial hematomas, and the presence of significant systemic injuries in addition to the head injury. Approximately 50% of patients dying from head injury do so from uncontrollable ICP, and this occurs early in the patient's course. In 30% to 60%, there is a good outcome or moderate disability. In contrast, the number of patients with a good outcome or moderate disability following hypoxic-ischemic coma was 13% in one large series.66 In addition to those patients who die, small numbers will be left in a persistent vegetative state or alert but totally dependent. The numbers of such patients have not been increased in those series of patients in which mortality was reduced by early aggressive management of head injury.6 Recovery may continue for up to 18 to 24 months after head injury, although the most significant gains are made during the first 6 months.3