Diffuse Axonal Injury

Histopathologic findings of DAI have been observed in 25% of patients who die of TBI.¹⁷ When more sensitive indicators of axonal injury are used, such as β-amyloid immunohistochemistry, as many as 90% of patients who die of TBI have DAI changes.¹⁸ Current thought about the development of DAI is that axotomy may not be complete immediately after the trauma. Secondary or delayed axotomy may predominate in many circumstances.

DAI is primarily a pathologic diagnosis, with microscopic damage scattered throughout the brain, but clinical correlations have been described. Traumatic loss of consciousness for less than 6 hours is considered a concussion and is usually associated with amnesia for the events related to the injury. Traumatic coma for longer than 6 hours is generally attributed to DAI. Three levels of DAI severity based on clinical criteria are recognized: (1) mild DAI—coma of 6 to 24 hours’ duration; (2) moderate DAI—coma lasting longer than 24 hours without decerebrate posturing; and (3) severe DAI—coma lasting longer than 24 hours with decerebrate posturing or flaccidity. Severe DAI has a 50% mortality.¹⁹

Hematomas/Contusions

Subdural hematoma is the most common focal intracranial lesion and occurred in 24% of patients in the TCDB with severe closed head injury.⁶ The hematoma is located between the dura and brain, usually results from a torn bridging vein between the cortex and draining sinuses, and is typically situated in the frontoparietal region. An acute subdural hematoma, identified within 72 hours after trauma, generally appears on CT as a high-density, homogeneous crescent-shaped mass paralleling the calvaria (Fig. 334-2). However, up to 10% of acute subdural hematomas may be isodense with brain because of the low hemoglobin content.²⁰

FIGURE 334-2 Categorization of primary brain injury (Traumatic Coma Data Bank categories). CT, computed tomography; GOS, Glasgow Outcome Scale; OR, odds ratio.

(From Marshall LF, Marshall SB, Klauber MR, et al. A new classification of head injury based on computerized tomography. J Neurosurg. 1991;75:S14-S20.)

The mortality rate in patients with subdural hematomas is high.²¹ The following CT scan findings are particularly predictive of outcome: hematoma thickness, midline shift, the presence of underlying brain swelling or contusions,^22,23 obliteration of the basal cisterns, and the presence of traumatic SAH.²¹

Subdural hematomas cause brain damage by increasing intracranial pressure (ICP) and by shifting brain structures. Reductions in cerebral blood flow (CBF) below ischemic thresholds²⁴ and marked reductions in cerebral oxygenation^25,26 have been observed preoperatively and are rapidly reversed by surgical evacuation of the hematoma. The presence of blood in the subdural space may also have direct toxic effects, but experimental studies have not consistently demonstrated this mechanism.²⁷

The primary treatment of subdural hematoma is prompt surgical evacuation, and the longer the delay between injury and surgery, the more severe the ischemic damage that can occur. Surgical indications for acute subdural hematoma include (1) any subdural hematoma greater than 10 mm in thickness or with a midline shift of greater than 5 mm and (2) a subdural hematoma less than 10 mm in thickness and with a midline shift of less than 5 mm if the GCS score is less than 9 and decreases 2 or more points between the time of injury and hospital admission, asymmetric or fixed and dilated pupils, or ICP exceeding 20 mm Hg.²⁸ Mortality in patients arriving at the hospital in coma with subsequent surgical evacuation is between 57% and 68%.²⁸

Epidural hematomas, or collections of blood between the skull and the dura, are less common than subdural hematomas. Epidural hematomas occurred in 6% of patients with severe closed head injuries in the TCDB series.⁶ Although patients with subdural hematomas are generally comatose immediately, only a third of patients with epidural hematomas are unconscious from the time of injury, a third have a lucid interval, and a third are never unconscious.²⁹ An epidural hematoma is almost always associated with a skull fracture (91% in adults and 75% in children).³⁰ The blood comes from torn dural vessels, usually arterial, from the fractured skull bone or occasionally from torn venous sinuses. On CT (see Fig. 334-2), an epidural hematoma is characterized by a biconvex, uniformly hyperdense lesion. Associated brain lesions are less common than with subdural hematomas. Epidural hematomas may also develop in delayed fashion³¹ or on the contralateral side after evacuation of an initial epidural hematoma.³²

The outcome of patients with an epidural hematoma depends on their neurological status at the time of surgery. The mortality rate varies from 0% for patients who are not in coma, to 9% for obtunded patients, to 20% for patients in deep coma. The following parameters were found to be significantly related to outcome in patients with epidural hematoma: age, time from injury to treatment, immediate coma or lucid interval, presence of pupillary abnormalities, GCS motor score on admission, CT findings (hematoma volume, degree of midline shift, presence of signs of active hematoma bleeding, associated intradural lesion), and postoperative ICP.³³

The primary treatment of an epidural hematoma is prompt surgical evacuation. An acute epidural hematoma with a volume greater than 30 cm³ should be surgically evacuated regardless of the patient’s GCS score. An epidural hematoma less than 30 cm³ in volume, less than 15 mm in thickness, and with less than a 5-mm midline shift in a patient with a GCS score higher than 8 and no focal deficit may be managed nonoperatively with serial CT scanning and close neurological observation.³⁰ Mortality in patients undergoing surgery for evacuation of an epidural hematoma is approximately 10%.³⁰

Intracerebral blood can take the form of a hematoma or a contusion. Intracerebral hematomas are more common and occurred as the primary lesion in 10% of patients with severe closed head injuries in the TCDB series.⁶ Occasionally, it may be difficult to differentiate a traumatic intracerebral hematoma from spontaneous hemorrhage; however, the presence of associated contusion, fracture, or an air-fluid level in the sinus helps in identifying the hematoma as traumatic. A zone of surrounding hypodensity denotes contusion or edema.

Most intracerebral hematomas are visualized as hyperdense mass lesions (see Fig. 334-2). They are generally located in the frontal and temporal lobes and can be detected on a CT scan immediately after the trauma. However, delayed intracerebral hematomas may also be manifested during the hospital course. A delayed hematoma is one that is seen on a repeated CT scan within 24 to 48 hours of the injury or operation but is not present on the initial CT scan. Commonly, a delayed hematoma is associated with clinical deterioration. Expansion of the intracerebral hemorrhage occurs in half the patients within the initial 24 hours, and larger hematomas exhibit the greatest expansion. Furthermore, the earlier the initial CT scan is obtained, the more likely it is that significant growth of the hematoma will be observed on subsequent scans.³⁴ Indications for surgery are (1) signs or neurological deterioration referable to the parenchymal lesion and medically refractory intracranial hypertension or signs of a mass effect on CT, (2) frontal or temporal contusions greater than 20 cm³ with a midline shift of 5 mm or cisternal compression in a patient with a GCS score of 6 to 8, and (3) any parenchymal lesion with volume greater than 50 cm³.³⁵ Factors that determine outcome in patients with intracerebral hematomas include GCS score, hematoma volume, and the occurrence of hypoxia.³⁶

Hemorrhagic contusions were present as the primary lesion in 3% of patients with severe closed head injuries in the TCDB series.⁶ Single contusions are located either below the region of the impact or opposite the region of impact. Contusions appear as heterogeneous areas of brain necrosis, hemorrhage, and infarction and represent mixed-density lesions on CT scan.³⁷ Multiple focal contusions have a “salt and pepper” appearance on CT. Ischemia may play a role in the pathogenesis of contusions, and regional CBF (rCBF) values averaging 17.5 ± 4.0 mL/100 g per minute have been measured in pericontusional tissue.³⁸

Classification of Primary Brain Injury

A number of injury severity schemes have been developed for predicting mortality after trauma and other critical illnesses. In general, these schemes have been developed in other patient populations and do not accurately predict outcome in neurologically injured patients. Rocca and colleagues studied 70 patients with TBI and found that the GCS was superior to the Acute Physiology Score (APS), the Simplified Acute Physiology Score (SAPS), and the Therapeutic Intervention Scoring System (TISS).³⁹ Alvarez and coworkers studied 401 patients with TBI and compared the predictive ability of the Acute Physiology and Chronic Health Evaluation (APACHE II), SAPS II, Mortality Probability Models (MPM II), and the GCS.⁴⁰ The MPM II system provided a good estimation of mortality. The SAPS II and APACHE II systems did not calibrate well, and the logistic regression model containing the GCS as an independent variable and developed in this group of patients was not as well calibrated as MPM II. Of all these systems, the GCS is the most commonly used neurological injury scale for adults because of its high interobserver reliability and generally good prognostic capabilities,⁴¹ but it is a poor discriminator for less severe TBI, which accounts for 80% to 90% of all cases.

Schemes that predict outcome well after TBI generally include age, severity of injury, and type of injury. For the severity-of-injury scale, the GCS, either the sum score or, for comatose patients, the motor score alone, is used. For the type of injury, a classification that separates focal and diffuse injuries is typically used. The CT classification of head injury based on information obtained from the initial CT scan was derived from the TCDB data.¹⁰ The scheme uses the status of the mesencephalic cisterns, the amount of midline shift, and the presence or absence of surgical masses. Diffuse injuries were defined in all patients with no mixed- or high-density lesions greater than 25 cm³. The category of diffuse injuries was divided into four subgroups: diffuse injury I included all head injuries with no visible pathology, diffuse injury II included all diffuse injuries with cisterns present and less than 5-mm shift, diffuse injury III included all diffuse injuries with compressed or absent cisterns but less than 5-mm shift, and diffuse injury IV included all diffuse injuries with more than 5-mm midline shift. Example of these diffuse injury categories are shown in Figure 334-2. The category of mass lesions, which included all patients with mixed- or high-density lesions greater than 25 cm³, was divided into those with the mass surgically evacuated (including operated subdural, epidural, and intracerebral hematomas) and those with nonevacuated mass lesions.

This CT classification scheme provided a better assessment of the risk for intracranial hypertension and a fatal outcome. Patients with diffuse injury I had the lowest mortality rate of 10%, whereas the diffuse injury II, III, and IV groups had mortality rates of 14%, 34%, and 56%, respectively. The mortality rate of patients with evacuated hematomas was 39%.

The IMPACT database studies have confirmed the usefulness of the TCDB classification but have also shown that individual CT characteristics such as traumatic SAH and the individual hematoma type (subdural versus epidural) have added prognostic value.²¹ Probably the most descriptive scheme at the present time is to use a combination of the TCDB classification plus these individual CT characteristics.

Traumatic Brain Swelling/Intracranial Hypertension

Brain edema and vascular engorgement have been used interchangeably to describe the brain swelling that accompanies severe TBI. Although the relative contribution of these two entities to the swelling process is controversial, there is increasing evidence that edema plays the predominant role. In experimental models of brain trauma, the water content of the brain is increased whereas cerebral blood volume is decreased, thus suggesting that edema is the major component of brain swelling after trauma.⁴² Clinical studies using magnetic resonance imaging (MRI) techniques to quantify water and blood content have confirmed these experimental findings in humans.⁴³

Brain edema can be vasogenic (secondary to opening the blood-brain barrier [BBB]) or cellular. Most experimental evidence, including recent data using diffusion-weighted imaging techniques to differentiate types of edema formation, suggest that the early, immediate increase in brain water after trauma is probably vasogenic whereas the gradual increase in brain water that occurs during the first few days after injury is cellular.^44,45

The clinical manifestation of brain swelling is intracranial hypertension, which develops in 50% of patients in coma caused by severe head injury. It is a misconception that ICP will always be low after operative evacuation of a large intracranial hematoma. Intracranial hypertension occurs in 50% to 70% of patients after evacuation of an intracranial hematoma.^46,47 This postoperative intracranial hypertension may be due to a postoperative hematoma, either at the site of the operation or at a new site, progressive swelling of a focal contusion, diffuse brain swelling, and other systemic complications. The incidence of intracranial hypertension is greater after evacuation of an intracerebral hematoma, 71%, than after evacuation of a subdural or epidural hematoma, 39%.⁴⁶ In patients with no mass lesions, elevated ICP has been observed to occur during the hospital course in 30%⁴⁶ to 80%⁴⁸ of patients.

The association between the severity of intracranial hypertension and poor outcome after severe head injury is well recognized. In one series, 77% of patients with ICP below 15 mm Hg had a favorable outcome as compared with just 43% of patients with ICP above 15 mm Hg.⁴⁹ Miller and coauthors reported that the mortality rate increased from 18% to 92% and the frequency of good outcomes decreased from 74% to 3% in patients with normal ICP versus patients who had intracranial hypertension that could not be reduced below 20 mm Hg.⁵⁰ Similarly, Saul and Ducker reported a 69% mortality rate in patients with ICP greater than 25 mm Hg as opposed to a mortality rate of 15% if ICP remained less than 25 mm Hg.⁵¹

The relationship between elevated ICP and a poor outcome is not simply a reflection of the severity of the initial neurological injury. Severe intracranial hypertension can result in secondary injury to the brain through ischemia produced by reduced cerebral perfusion pressure (CPP), and it can also distort and compress the brainstem. Although no randomized clinical trial has addressed this question, several clinical series have suggested that reduction of ICP to less than 20 mm Hg does reduce mortality after severe head injury.^46,51–53

Effects of Trauma on the Cerebral Vasculature

Many investigators have studied CBF after TBI and emphasized various patterns of CBF. A few studies have closely examined the evolution of TBI over time with serial measurements of CBF.^54,55 These serial studies in particular have helped to develop the overall picture of the postinjury evolution of CBF.

Martin and coworkers described a phasic pattern of cerebral hemodynamic changes in 125 severely head-injured patients with a GCS score lower than 8 who were studied prospectively with intravenous ¹³³Xe-CBF measurements, cerebral metabolic assessment, or transcranial Doppler (TCD) evaluations (or any combination of the three studies).⁵⁴ An early hypoperfusion phase occurred during the first 24 hours after injury and was characterized by low CBF and normal middle cerebral artery flow velocity (MCA-FV). This was followed by a hyperemic phase that occurred in about 40% of patients between postinjury days 1 and 3 when CBF was transiently increased with a rapidly rising MCA-FV but a normal hemispheric index. Later (postinjury days 4 to 15), a vasospasm phase occurred that was characterized by low-normal CBF values accompanied by high MCA-FV and an elevated hemispheric index. Hlatky and colleagues found a significant reduction in CBF during the first 12 hours after TBI; in patients with CBF lower than 18 mL/100 g per minute, intracranial hypertension played a major causative role in the reduction in CBF, and for levels of CBF between 10 and 40 mL/100 g per minute, the presence of regional hypoperfusion was a more important factor in reducing the average CBF.⁵⁶ These CBF patterns, which have been described primarily with global CBF measurements, can also occur regionally. Hypoperfusion, in particular, can occur in brain tissue surrounding a focal contusion or underlying a subdural hematoma. Schroder and associates observed rCBF values averaging 17.5 mL/100 g per minute in pericontusional brain.³⁸ McLaughin and Marion observed CBF values within contused brain and in pericontusional brain that were significantly lower than those in the rest of the brain.⁵⁷ Focal hyperemia has been observed in 38% of patients, particularly in tissue adjacent to intraparenchymal or extracerebral focal lesions.⁵⁸

Causes of Secondary Ischemic Insults

Any pathophysiologic process that impairs cerebral energy metabolism can be a secondary ischemic insult after head injury. As shown in Figure 334-1, these processes can be divided into two general categories: (1) those that decrease the cerebral delivery of energy substrates and (2) those that increase cerebral energy consumption.

Cerebral oxygen delivery is the product of CBF and arterial oxygen concentration (CaO₂). CBF is normally closely regulated by the cerebral metabolic rate and will also be affected by CPP and arterial PCO₂. However, in disease states, CBF may be uncoupled from the metabolic rate and can be elevated or reduced from normal. CaO₂ is calculated from arterial oxygen saturation and the hemoglobin concentration and is normally around 20 mL/dL (9 µmol/mL). Both hypoxia and anemia can reduce CaO₂.

Cerebral oxygen consumption or CMRO₂ has a normal value of 3.4 mg/100 g per minute (1.5 µmol/g per minute). Numerous conditions are known to alter CMRO₂. Head injury, anesthesia, and hypothermia are known to decrease CMRO₂, whereas fever and seizures increase it.^63,70–76 Recent studies using [¹⁸F]fluorodeoxyglucose-labeled positron emission tomography (FDG-PET) to study glucose metabolism after severe head injury have suggested that the cerebral metabolic rate of glucose (CMRGlc) may be elevated regionally and perhaps even globally at times when CMRO₂ is normal or reduced.⁷⁷ This “hyperglycolysis” may indicate mitochondrial damage and inability of the brain to metabolize oxygen normally. If energy metabolism is dependent on glucose metabolism, energy failure as a result of glucose depletion is another possible cause of secondary ischemic insults after severe head injury.

Techniques for Monitoring Intracranial Pressure

Complications

The two major complications of ICP monitoring are ventriculitis and hemorrhage. Infection may be confined to the skin wound, but ventriculitis occurs in 1% to 10% of patients. Factors that predispose to ventriculitis are intraventricular hemorrhage, SAH, cranial fracture with leakage of CSF, craniotomy, systemic infections, catheter manipulation, leaks, and irrigation. The duration of catheterization has been correlated with an increasing risk for CSF infections during the first 10 days of use. Although prophylactic catheter exchange remains a practice option, the available data suggest that this procedure does not reduce the risk for infection.⁸⁹ Systemic prophylactic antibiotics are ineffective in reducing the incidence of infections, but antibiotic-impregnated ventriculostomy catheters reduce the risk for CSF infection from 9.4% to 1.3%.⁹⁰ Systemic prophylactic antibiotics and routine catheter exchange are not recommended in the current TBI guidelines (see Table 334-3—level III recommendation). The best strategies for reducing the risk for ventriculitis associated with ICP monitoring are meticulous aseptic technique during catheter insertion, the use of antibiotic-impregnated catheters, and minimization of the duration of monitoring.

The second major complication of ICP monitoring is intracerebral hemorrhage. Although the risk for hemorrhage has been shown to be consistently low (1% to 2%), it is an important complication to recognize and treat.⁸³ Patients with coagulopathies are at greater risk for the development of this complication. Some evidence suggests that patients with an international normalized ratio of 1.6 or less have a very low risk for hemorrhage if a parenchymal catheter is placed.⁹¹

Normal Intracranial Pressure Values

Normally, resting ICP is less than 10 mm Hg. Sustained ICP of greater than 20 mm Hg is clearly abnormal. ICP between 20 and 40 mm Hg is considered moderate intracranial hypertension. ICP greater than 40 mm Hg represents severe, usually life-threatening intracranial hypertension. These thresholds for the severity of intracranial hypertension assume normal BP. When a temporal mass lesion is present, however, herniation can occur at ICP values of less than 20 mm Hg.⁹²

Different authors have recommended treatment of ICP when greater than 15, 20, or 25 mm Hg. No studies have clearly established one of these values as the most appropriate threshold for treatment. Marmarou and coworkers observed an increasingly worse outcome with greater durations of time that ICP was higher than 20 mm Hg.⁹³ Contant and associates observed a worse outcome with greater durations of time that ICP was higher than 25 mm Hg.⁹⁴ The current TBI guidelines (see Table 334-3) recommend treating ICP above 20 mm Hg (level II recommendation) or using clinical and CT characteristics in addition to the ICP value to determine the need for treatment (level III recommendation).¹³

Indications for Intracranial Pressure Monitoring

Monitoring of ICP can result in serious complications and is therefore indicated only in patients at significant risk for the development of intracranial hypertension. Indications from the current TBI guidelines are listed in Table 334-3.^13,95 ICP should be monitored in all salvageable patients with severe TBI, defined as a GCS score of 3 to 8 after resuscitation and abnormal findings on CT (level II recommendation). ICP monitoring is indicated in patients with severe TBI and normal CT findings if two or more of the following features are present at admission: age older than 40 years, unilateral or bilateral motor posturing, or systolic BP lower than 90 mm Hg (level III recommendation). Patients with a GCS score higher than 8 might be considered for ICP monitoring if they require treatment that would not allow serial neurological examinations, such as prolonged anesthesia for surgery to treat multiple injuries or prolonged pharmacologic paralysis for ventilatory management, or if they require a treatment that might raise ICP, such as positive end-expiratory pressure (PEEP). A severe coagulopathy is the only major contraindication to ICP monitoring.

Monitors of Cerebral Perfusion

Cerebral Blood Flow

Techniques for Monitoring Cerebral Blood Flow

Recently, newer technologies have made measurement of CBF more feasible in critically ill patients. Measurement of global CBF by the classic Kety-Schmidt technique in which nitrous oxide is used as the diffusible indicator can be performed at the bedside with a minimum of expense and equipment. Measurement of rCBF by stable xenon–CT or perfusion CT is possible and can even be performed in the ICU with a portable CT scanner (examples are shown in Fig. 334-3). However, these measurements of CBF are intermittent and require the patient to be hemodynamically stable during the time required for the measurements. Therefore, transient reductions in CBF or reductions in CBF in an acutely unstable patient are difficult to document with these technologies.

FIGURE 334-3 Examples of cerebral blood flow imaging with stable xenon–enhanced computed tomography (CT) (left) and CT perfusion (right). The top image is from a patient with a temporal contusion. The bottom image is from a patient with a hemispheric infarction after evacuation of an acute subdural hematoma.

The stable xenon–CT method relies on the radiodensity of xenon and its inertness and rapid diffusion into tissues. An initial non–contrast-enhanced scan helps in choosing the levels for CBF measurement. Baseline scans, usually at four levels, are followed by multiple scans at each of the four levels at 1-minute intervals during the inhalation of 28% to 33% xenon. The baseline pre-xenon scans are subtracted from the subsequent xenon scans to provide quantitative enhancement values in Hounsfield units. Clearance curves proportional to brain and arterial xenon concentrations from an end-tidal analyzer are converted into CT enhancement units. Blood flow is then calculated from the Kety-Schmidt equation for each CT voxel. Studies in normal volunteers have indicated that xenon at a concentration of 30% has an effect on cerebral hemodynamics and causes arterial vasodilation and enhancement of CBF.^100,101 This probably occurs as a result of an increase in metabolism and a vasodilatory effect of xenon. Some studies have observed a small increase in ICP averaging 6.9 ± 7.7 mm Hg during CBF measurements with xenon-CT,¹⁰² whereas others have not.¹⁰³

Perfusion CT imaging uses a nondiffusible indicator technique to calculate blood flow. The imaging study is performed by collecting a continuous (kinematic) sequence of scans after an intravenous bolus of an iodine contrast agent. Each image is subtracted from a baseline image to develop a time enhancement curve. Perfusion is then calculated by the slope method.

Local Cerebral Blood Flow Sensors for Brain Implantation

Two methods for continuously measuring local CBF are now commercially available, the thermal diffusion method and the laser Doppler method. Both methods are invasive and require the probe to be placed on the surface of the brain at surgery or into the brain through a bur hole. In addition, both methods measure CBF in only a small volume of brain, which may or may not be representative of the whole brain. However, the continuous nature of the measurements gives a dynamic picture of brain perfusion. Although there is extensive documentation of the reliability of these methodologies in the laboratory, experience in the ICU is currently limited.^104–107 Nonetheless, especially for patients undergoing a craniotomy, these methods may become practical for monitoring CBF postoperatively.

Thermal diffusion flowmetry (TD-CBF) uses heat transfer as a tracer for the measurement of blood flow. The sensor consists of two gold disks embedded in a 3-mm Silastic leaf or two bands on an intraparenchymal catheter. One disk is heated slightly above brain temperature (to a maximum of 44°C), whereas the other is neutral. The leaf probe is laid on the surface of the brain, and the catheter version can be implanted into the brain parenchyma. The difference in temperature between the two disks is monitored and converted to blood flow in mL/100 g per minute by the monitor and displayed digitally. Carter and Atkinson¹⁰⁸ modified a thermal sensor described by Brawley¹⁰⁹ and were able to quantitate the flow measured by the thermal sensor. They derived a mathematical formula by comparing TD-CBF with ¹³³Xe-CBF¹¹⁰ and further confirmed its reliability through comparison with hydrogen clearance.¹¹¹ In TBI patients, the CBF values obtained with the catheter probe have good correlation with those obtained with stable xenon–CT.¹¹²

Laser Doppler flowmetry (LDF) is a technique in which CBF is measured indirectly from the magnitude of the frequency shift of monochromatic light by a moving column of blood. The most accurate recordings are obtained when the probe is in direct contact with the region of interest but away from obviously visible vessels, which will bias the capillary flow. The most useful application of LDF is for continuous monitoring of CBF in the ICU. Its high temporal resolution allows the detection of transient hemodynamic events, although the information involves relative changes in local CBF.

Cerebral Blood Flow Adequacy

Measures of cerebral oxygenation, such as SjvO₂ and brain tissue PO₂ (PbtO₂), have been used in place of quantitative CBF measurements because they give an indicator of the adequacy of CBF relative to cerebral metabolic requirements. Because cerebral metabolic requirements may be reduced after TBI, normal CBF values may not apply. When CBF is low (25 to 30 mL/100 g per minute), it can be difficult to decide whether it is an appropriate response to lower cerebral metabolic requirements or whether the brain is hypoperfused. A measure of cerebral oxygenation can be helpful in making this distinction. If the brain is hypoperfused, oxygen extraction will be increased and SjvO₂ will be reduced. If CBF is appropriate for the brain’s metabolic requirement, SjvO₂ will be normal. Frequently, this information is more clinically useful than the absolute CBF values.

Summary of Monitoring Options for Secondary Cerebral Ischemia

It is clear from the description of the available devices that none are ideal in the head injury setting. No single monitor gives continuous high-resolution rCBF information. At the present time, a reasonable strategy is to use intermittent measurements of rCBF, preferably by the stable xenon–CT or perfusion CT method, to obtain information about whether regional abnormalities in flow exist. If there are no hypoperfused regions, a global monitor such as SjvO₂ should suffice. If there are significant regional abnormalities in flow, it may be better to use PbtO₂ or one of the local CBF probes as a local monitor of the hypoperfused area in addition to a global monitor such as SjvO₂.

Figure 334-4 illustrates how this strategy might be used to identify and guide the treatment of focal ischemia after a head injury. This patient was initially admitted with a subdural hematoma (Fig. 334-4, left CT scan), which was evacuated promptly on arrival at the hospital. At surgery, a PbtO₂ catheter was placed near contused brain underlying the evacuated hematoma. Initially, PbtO₂ was normal, but over postinjury hours 36 to 46, PbtO₂ gradually decreased to 5 mm Hg with no changes in any other physiologic parameters. Note that SjvO₂, which is a measure of global oxygenation, remains low-normal (≈55%) despite the very low PbtO₂ and that CPP was 65 to 70 mm Hg. Administering 100% oxygen raised PbtO₂ to 15 mm Hg (Fig. 334-4, top graph). A CT scan (Fig. 334-4, middle scan) showed increased swelling around the contusion, and stable xenon–CT showed a large area with rCBF values of less than 10 mL/100 g per minute at the contusion site. Dopamine was started and phenylephrine added later to increase blood pressure. As CPP increased with this treatment, PbtO₂ returned to normal (Fig. 334-4, bottom graph), and a repeated stable xenon–CT scan (Fig. 334-4, right scan) showed significant improvement in CBF.

FIGURE 334-4 Example of identification of focal ischemia and treatment guided by brain tissue PO₂ (PbtO₂). This patient was admitted with a subdural hematoma (left computed tomography [CT] scan), which was evacuated promptly on arrival at the hospital. A PbtO₂ catheter was placed near contused brain underlying the evacuated hematoma. Initially, PbtO₂ was normal, but over postinjury hours 36 to 46, PbtO₂ gradually decreased to 5 mm Hg. Note that jugular venous oxygen saturation (SjvO₂), which is a measure of global oxygenation, remains low-normal despite the very low PbtO₂ and that cerebral perfusion pressure (CPP) was 65 to 70 mm Hg. Administration of 100% oxygen raised PbtO₂ to 15 mm Hg (top graph). A CT scan (middle scan) showed increased swelling around the contusion, and xenon-CT showed a large area with regional cerebral blood flow (CBF) values greater than 10 mL/100 g per minute at the contusion site. Dopamine was started and phenylephrine was added later to increase blood pressure. As CPP increased with this treatment, PbtO₂ increased (bottom graph), and repeated xenon-CT (right scan) showed significant improvement in CBF.

In this case, the progressive focal ischemia in the area of the contusion could possibly have been suspected from the changing appearance of the CT scan; however, stable xenon–CT and PbtO₂ provided a convenient way to grasp the severity and extent of the ischemia, and PbtO₂ also provided a good method for monitoring the effectiveness of the treatments applied in the ICU. The challenge with this strategy for monitoring is placing the local monitor in an area of the brain that is vulnerable to the development of ischemia.

Monitoring for Cerebral Causes of Secondary Ischemic Insults

Monitoring for Systemic Causes of Secondary Insults

Minimize Venous Outflow Resistance: Head Elevation, Head Position

Elevation of the head of the bed while keeping the head in a neutral position to minimize compression of venous return from the brain has been standard neurosurgery practice for the management of ICP in the past. However, the ideal head position for a patient with head injury has been disputed in recent years. Rosner and Coley advocated keeping the patient’s head flat as part of an overall treatment program intended to maximize CPP.^172,173 Other studies have shown a reduction in ICP without a reduction in either CPP or CBF in most patients with elevation of the head to 30 degrees.^174–176 In 22 patients with severe head injury, elevation of the head to 30 degrees reduced ICP and BP without changing CPP or CBF in most patients.¹⁷⁵ Meixensberger and coworkers observed that elevation of the head to 30 degrees reduced ICP and increased CPP but did not change PbtO₂.¹⁷⁶ Therefore, unless the patient is hypovolemic and head elevation results in hypotension, the reduction in ICP afforded by 30 degrees of head elevation is probably advantageous. If head elevation is used, it is important to remember that both the ICP and BP transducers should be zeroed at the same level (i.e., at the level of the foramen of Monro).¹⁷²

In multiple trauma patients in whom abdominal injury may also be present, increased intra-abdominal pressure may also impede venous return from the brain, decrease BP, and increase ICP. Experimental and clinical studies suggest that abdominal decompression can significantly improve control of ICP when abdominal compartment syndrome is present.^177,178

Sedation/Analgesia

Sedative/analgesic drugs blunt the effect that the routine nursing care of patients has on ICP. In general, the benzodiazepines cause a coupled reduction in CMRO₂ and CBF, with no effect on ICP, whereas the narcotics have no effect on CMRO₂ or CBF but have been reported to increase ICP in some patients.^179,180 A combination of morphine and lorazepam is a frequently used regimen that is well tolerated and provides good sedation/analgesia.

In choosing a sedative agent for a head-injured patient, it is important to remember to avoid drugs with hypotensive side effects (Table 334-4).^179–190 Hypovolemia predisposes to the hypotensive side effects of sedatives and should be remedied before administering sedative agents. Propofol has the advantage of a short half-life, which allows intermittent neurological examination, but it is a potent systemic vasodilator and can cause hypotension that exceeds the reduction in ICP, so CPP can be significantly reduced.¹⁹¹ Propofol infusion syndrome can result as a consequence of the use of high doses of propofol. Some clinical features are hyperkalemia, hepatomegaly, lipemia, metabolic acidosis, myocardial failure, rhabdomyolysis, and renal failure. Extreme caution must be taken when using doses greater than 5 mg/kg per hour or when use of any dose exceeds 48 hours in critically ill adults.¹⁹²

Dexmedetomidine is an α₂-agonist that provides adequate sedation without altering the respiratory drive while facilitating frequent neurological examinations. The dosage range for a loading infusion is 0.1 µg/kg infused over a 10-minute period, followed by a 0.2- to 0.7-µg/kg/hr continuous infusion for 24 hours. It is a safe and effective sedative agent in neurosurgical patients, although a loading infusion should be avoided and higher maintenance doses may be required to ensure adequate sedation.¹⁹³

Treatment of Systemic Hypertension

Systemic hypertension associated with head injury is common and is characterized by an increase in systolic BP that is greater than the increase in diastolic pressure. It is associated with a hyperdynamic state, including tachycardia and increased cardiac output. Systemic hypertension is associated with sympathetic hyperactivity.¹⁹⁴ It is unwise to reduce systemic BP in patients with hypertension associated with untreated intracranial mass lesions because cerebral perfusion is being maintained by the higher BP. Treatment of systemic hypertension (systolic BP >160 mm Hg) during the postoperative course after a head injury is recommended by many neurosurgeons. Because autoregulation is frequently impaired after severe head injury, systemic hypertension may increase CBF and ICP¹⁹⁵ and exacerbate cerebral edema.¹⁹⁶ However, this issue is controversial, and others have emphasized the importance of maintaining CPP, even at the expense of higher ICP.^173,197

If hypertension is treated in a head-injured patient, the choice of antihypertensives is important. Frequently, systemic hypertension will resolve with sedation. If antihypertensive drugs are required, vasodilating antihypertensive drugs, including hydralazine, nicardipine, and nitroprusside, consistently increase ICP.^198,199 Sympathomimetic-blocking antihypertensive drugs, such as β-blocking agents^200,201 (propranolol, esmolol, or labetalol), or centrally acting α-receptor agonists (clonidine or α-methyldopa) are preferred because they reduce BP without affecting ICP.¹⁹⁴ Agents with a short half-life have an advantage when BP is labile. Nicardipine has been studied or recommended for the management of hypertension in many neurovascular settings, including TBI. Nicardipine meets the criteria as a short-acting continuous-infusion agent with a reliable dose-response relationship and favorable safety profile. In a controlled, double-blinded, randomized pilot clinical trial, nicardipine was shown to reverse and prevent vasospasm based on Doppler FV in patients with moderate and severe TBI.²⁰²

Airway Protection/Controlled Ventilation

TBI patients in coma often cannot protect their airway and should be intubated initially to prevent aspiration and airway obstruction.²⁰³ Respiratory dysfunction, reported by clinical observation and measured by abnormal blood gas values, is also common after head injury. Mechanical ventilation is usually needed in the acute recovery period to allow deeper levels of sedation to be used and to reduce the risk that periodic hypoventilation will exacerbate the intracranial hypertension.

In a series of 225 patients with severe head injury, hypoxia (PO₂ <60 mm Hg) was seen on admission to the hospital in 35% of the patients, and hypercapnia was present in 8%.¹⁴¹ Kohi and colleagues found that 36% of 67 comatose head-injured patients had either hypoxia (PO₂ <65 mm Hg) or severe respiratory dysfunction requiring mechanical ventilation on admission.²⁰⁴ In another study, 9% of 363 patients admitted while in coma from head injury were hypoxic (cyanotic or PO₂ <60 mm Hg) on admission.²⁰⁵ In the TCDB studies, hypoxia was present in 19% of the comatose head-injured patients at arrival in the emergency department.⁸ Pneumonia developed as a complication during hospitalization in 41% and pulmonary insufficiency occurred in 28%.¹⁴²

After experimental impact and missile head injury, an increasing duration of apnea has been observed with increasing severity of the brain injury.^206,207 Immediate deaths in these models are generally caused by apnea. There is little information on apnea after human head injury, probably because the actual impact is typically not observed. Apnea at the scene of the accident, 3 and 12 minutes in duration, was reported in 2 patients who eventually recovered well with the early institution of cardiopulmonary resuscitation and artificial ventilation.²⁰⁸ Pfenninger and associates measured arterial blood gases before resuscitation at the scene of the accident (6 to 21 minutes after injury) in 33 patients with head injury.²⁰⁹ An elevation in PCO₂ suggesting apnea or hypoventilation was strongly correlated with the severity of the injury in these patients. These studies suggest that apnea probably occurs in human head injury similar to that observed in experimental models.

North and Jennett examined breathing patterns in 227 spontaneously breathing patients with neurosurgical diagnoses, most commonly trauma.²¹⁰ Sixty percent of the patients had some type of breathing abnormality, including periodic respirations, tachypnea, and irregular breathing. All 12 patients with medullary lesions had abnormal breathing patterns, 8 of whom had irregular breathing. In contrast, periodic breathing was not correlated with any particular anatomic site of the neurological injury.

Treatment of these ventilatory abnormalities during the acute phase after injury is by controlled ventilation to prevent inducing rapid changes in PCO₂. Sedation and occasionally paralysis may be necessary to control ventilation. Hypoxia and hypercapnia can dramatically raise ICP. Patients with severe head injury can have periodic episodes of hypoventilation that precipitate episodes of intracranial hypertension. Controlled ventilation helps prevent these episodes of hypoventilation and intracranial hypertension. Early tracheostomy in trauma patients who remain ventilator dependent has been shown to reduce length of stay in the ICU.^211,212

Treatment of Fever

Fever is common during recovery from a head injury. In experimental studies, postinjury fever worsens the outcome after fluid percussion injury.¹⁵⁹ Fever is a potent cerebral vasodilator and can raise ICP. In addition, fever can raise cerebral metabolic requirements. Elevated rectal temperatures should be treated with antipyretics, cooling blankets, or both. Infectious causes of fever should be investigated with appropriate cultures and treated with antibiotics.

Prevention of Seizures

The risk for posttraumatic seizure is approximately 15% with severe head injury.²¹³ The risk for seizures is related to the severity of the injury. In a study of 4541 patients with head injury, the standardized incidence ratio for the development of seizures was 1.5 after mild injuries but with no increase over the expected number after 5 years, 2.9 after moderate injuries, and 17.0 after severe injuries. In multivariate analysis, significant risk factors for later seizures were brain contusion with subdural hematoma, skull fracture, loss of consciousness or amnesia for longer than 1 day, and age 65 years or older.²¹⁴

The use of anticonvulsants to prevent seizures is controversial. Although seizures can dramatically increase the cerebral metabolic rate, there is not a clear relationship between the occurrence of early seizures and a worse neurological outcome.²¹⁵ Young and coworkers found no difference in the incidence of seizures with prophylactically administered phenytoin.²¹⁶ Tempkin and coauthors reported the results from a double-blind study in which 404 severely head-injured patients randomly received phenytoin or placebo for 1 year.²¹⁷ Phenytoin reduced the incidence of seizures during the first week but not thereafter. Therefore, it is reasonable to give all severely head-injured patients in the emergency department phenytoin, 15 mg/kg intravenously at a rate that does not exceed 25 mg/min, followed by a daily maintenance dose adjusted to keep plasma levels in the therapeutic range. If no early seizures occur, the phenytoin is tapered and discontinued after 1 week.

Levetiracetam, which is now available in parenteral form, may be an alternative to phenytoin for seizure prophylaxis. Levetiracetam has the advantages that it does not require serum monitoring or have significant pharmacokinetic interactions. In a recently published case series of 32 patients with severe TBI, it was found that levetiracetam was as effective as phenytoin in preventing early posttraumatic seizures but was associated with an increased tendency for seizure activity on EEG analysis.²¹⁸

Prevention of Ventilator-Associated Pneumonia

TBI patients have a high incidence of pneumonia, 40% in some series, and this complication is often associated with aspiration, which may occur even before receiving medical attention. Because of the high incidence, the possibility of aspiration should be considered in every comatose TBI patient. In prehospital series, clinical assessment of the presence of aspiration by the paramedic is usually reliable, and it most commonly occurs before their arrival at the scene.²¹⁹ Therefore, rapid-sequence intubation in the field does not prevent the occurrence of aspiration, although it may reduce the incidence. Once in the emergency department, CT scanning of the thorax may be a more sensitive method than routine chest radiography for identifying aspiration. The subsequent occurrence of pneumonia does not alter mortality in TBI patients, but it does significantly extend the need for ventilation and prolong the ICU stay.²²⁰ Studies of antibiotic prophylaxis for ventilated TBI patients, either cefuroxime, 1500 mg intravenously for two doses after intubation, or ampicillin-sulbactam, 3 g every 6 hours for 3 days, have demonstrated significant reductions in the incidence of subsequent pneumonia.^221,222

Beyond antibiotic prophylaxis, the most effective way of minimizing the risk for ventilator-associated pneumonia is to avoid or limit the duration of intubation and mechanical ventilation. In many clinical conditions, noninvasive ventilation can effectively provide the ventilatory support needed and is associated with a lower risk for pneumonia. For a patient with severe TBI, this is not usually an alternative because the major indication for intubation is to protect the airway. However, some strategies to reduce the duration of ventilation and intubation should be applied even in this population. Daily interruption or lightening of sedation to evaluate readiness to wean from ventilatory support can be readily timed to occur with the daily neurological examination. Decisions about whether a patient is ready to be weaned from ventilatory support can then be based on both neurological and pulmonary status. Daily spontaneous breathing trials, which have been successful in other patient populations in shortening the duration of ventilation, have not been shown to do so in neurosurgical patients.²²³ The decision about whether patients can protect their airway is more likely to determine the timing of extubation.

Other general recommendations to minimize the risk for ventilator-associated pneumonia include the use of oral rather than nasal intubation, continuous aspiration of subglottic secretions, maintenance of endotracheal tube cuff pressure at values of at least 20 cm H₂O, remaining in a semirecumbent position (30 to 45 degrees), emptying condensate from the ventilator tubing, and not allowing condensate to contaminate the endotracheal tube.²²⁴

Prophylaxis for Thromboembolism

Venous thromboembolism is a common complication after major trauma, with an observed incidence as high as 58% if prophylaxis is not used.²²⁵ Within the general trauma population, risk factors for thromboembolism include spinal cord injury; pelvic, femoral, or tibial fracture; surgery; blood transfusion; and older age,²²⁵ as well as major head injury in some studies,.²²⁶

Both venous compression devices and low-dose heparin are effective in reducing the risk for thromboembolism and subsequent pulmonary embolism. Dennis and colleagues found a reduction in the incidence of thromboembolism from 8.98% to 2.9% with prophylactic treatment.²²⁷ Both low-dose heparin and sequential venous compression prophylaxis were equally effective in this study. Because TBI may be a relative contraindication to anticoagulation, sequential compression devices are preferred by many,²²⁸ and this is the class II recommendation in the third edition of the head injury guidelines.²²⁹ Other guidelines recommend low-molecular-weight heparin prophylaxis as soon as it is safe²³⁰; however, there is little convincing information available regarding when prophylactic heparin would be safe in a patient with traumatic intracranial hemorrhage. The finding of stable hemorrhage on CT scans separated by 12 to 24 hours has been suggested to indicate that low-molecular-weight heparin could be started safely.²³¹

The use of prophylactic vena cava filters in high-risk trauma patients also remains controversial. Evidence that vena cava filters actually prevent pulmonary emboli is lacking, and this type of prophylaxis probably has long-term consequences that are not justified by the small increase in protection provided against pulmonary embolism.²³⁰ The recent introduction of retrievable vena cava filters has made this option for prophylaxis popular in some trauma centers, although most authorities do not recommend these filters as routine practice for prophylaxis.²³⁰

Prophylaxis for Gastric Ulcers

Stress ulcers are a common complication in critically ill patients. Endoscopic evidence of mucosal damage can appear within 24 hours of a severe brain injury, and 17% of these early erosions can progress to clinically significant hemorrhage.²³² The major risk factor for the development of gastric bleeding is the severity of the brain lesion.²³² Other common risk factors in trauma patients include burns involving greater than 25% of body surface area, respiratory failure, hypotension, sepsis, jaundice, peritonitis, coagulopathy, and hepatic failure.²³³

The mechanism of stress ulceration is not completely understood, but increased gastric acidity plays a major role. Studies have suggested that intracranial injury, particularly that involving the diencephalon and brainstem, results in increased production of gastrin and gastric acid.²³⁴ In a study of high-risk postoperative neurosurgical patients, the major risk factors were a gastric pH lower than 4 and a high daily volume of gastric output.²³⁵

Prevention of stress ulceration has traditionally been directed at neutralization of gastric acid or reduction of gastric acid production with H₂ blockers. Both methods have been shown to be effective in reducing the incidence of gastric bleeding, but administration of antacids, which is most effective if titrated to stomach pH, is more time-consuming.^235,236 H₂ blockers, however, can have central nervous system effects that may be troublesome in neurosurgical patients.

One potential side effect common to both these preventive strategies is an increase in the risk for nosocomial pneumonia, probably as a result of aspiration of colonized gastric secretions. When gastric pH is lower than 4, bacterial colonization of the stomach occurs. This may result in an increase in the incidence of nosocomial pneumonia, although the data are conflicting. Sucralfate, which is a complex of sucrose, sulfates, and aluminum hydroxide, is an alternative that achieves its protective effects by mechanisms other than a decrease in gastric acidity. Sucralfate strengthens the gastric mucosa and increases mucosal blood flow, and it has been demonstrated to be at least as effective as antacids in preventing clinical gastric bleeding.²³⁶

In a retrospective cohort study conducted by Ecker and colleagues, the use of H₂ blockers in neurosurgical patients was associated with an increased incidence of thrombocytopenia.²³⁷ Therefore, the use of proton pump inhibitors or sucralfate may be a better choice for gastrointestinal prophylaxis in neurosurgical patients.

Prophylactic Antibiotics to Prevent Meningitis

In a study of posttraumatic bacterial meningitis, the incidence of meningitis was found to be 18% and 9% when associated with otorrhea and rhinorrhea, respectively, versus 0.38% in the absence of a CSF leak.²³⁸ Tenney found a higher risk for infection with rhinorrhea than with otorrhea.²³⁹ The organisms causing posttraumatic meningitis are evenly divided between gram-positive and gram-negative bacteria.²⁴⁰ The efficacy of prophylactic antibiotics in patients with traumatic CSF fistula is controversial. Antibiotics are recommended only when symptoms or signs of meningitis develop.

Nutritional Support

Patient with severe head injury are hypermetabolic and catabolic, similar to patients with multiple trauma.²⁴¹ In a large retrospective study of hospitalized TBI patients, those who were not fed within 5 and 7 days after TBI had a twofold and fourfold increased likelihood of death, respectively. The amount of nutrition in the first 5 days was related to death; every 10-kcal/kg decrease in caloric intake was associated with a 30% to 40% increase in mortality rates adjusted for injury severity.²⁴²

A number of studies have measured protein and caloric requirements in head-injured patients.²⁴³ In these studies, caloric expenditure after TBI averaged 140% of normal resting energy expenditure (REE) based on age, gender, and body size. REE can be calculated from the Harris-Benedict equation:

For a 70-kg, 25-year-old man, normal REE is 1700 kcal/day. The projected caloric expenditure is 2400 kcal/day. In patients who are paralyzed or in barbiturate coma, caloric expenditure is 100% to 120% of normal resting levels. The average nitrogen loss for a head-injured patient is 0.2 g/kg per day or 14 g/day in a 70-kg male. This is 2 to 3 times the normal fasting nitrogen loss and can result in a 10% decrease in lean body mass in 1 week.

These projected losses should be replaced, preferably with enteral feedings, beginning as soon after injury as possible, but certainly within 72 hours. The goal should be to gradually increase feedings and be at full caloric replacement by the end of the first week after injury (see Table 334-3—level II recommendation in the TBI guidelines). As the rate of enteral feedings is increased, maintenance fluids should be decreased to maintain balance between total fluid input and output.

Enteral feedings are preferable to parenteral feedings because gut integrity is better maintained and this may reduce the risk for sepsis.^244,245 If gastric emptying is impaired, feeding via jejunal tube is usually successful.^246,247 In the more chronic phase, percutaneous gastrostomy is a safe and effective method of establishing a permanent route for enteral feeding and is less costly than open gastrostomy.²⁴⁸ Even pentobarbital coma, used to lower ICP, does not completely preclude the enteral route of feeding.²⁴⁹

Management of Fluid/Electrolytes

Electrolyte abnormalities occur commonly in patients with head injury, at least once in the course of 59% of the patients in the TCDB.¹⁴² Abnormalities in sodium, both hyponatremia and hypernatremia, are among the most common. Hyperglycemia is also common and is part of the stress response to injury. Meticulous attention should be paid to fluid administration and fluid balance.

Hyponatremia Syndromes

The differential diagnosis of hyponatremia in a head trauma patient is usually reduced to two entities, the syndrome of inappropriate antidiuretic hormone secretion (SIADH) and cerebral salt wasting (Fig. 334-5). The diagnosis of SIADH requires the following: (1) hyponatremia (serum sodium <135 mEq/L), (2) hypo-osmolarity (serum osmolarity <280 mOsm/L), (3) urine osmolarity greater than serum osmolarity, and (4) inappropriately high urine sodium concentration (>40 mEq/L). Patients with SIADH are normovolemic or slightly volume expanded. In contrast, patients with salt-wasting syndrome are hypovolemic. Cerebral salt wasting also results in high urine sodium levels (>40 mEq/L). Volume status is the critical clinical distinction between the two syndromes. SIADH is caused by enhanced secretion of ADH, whereas cerebral salt wasting is due to elevated levels of circulating natriuretic factors, possibly including atrial natriuretic factor.

FIGURE 334-5 In head-injured patients, most hyponatremia can be attributed to either cerebral salt wasting or the syndrome of inappropriate antidiuretic hormone secretion (SIADH).

For SIADH, fluid restriction is the primary treatment. Limiting fluid intake to 800 to 1000 mL/day will result in a steady rise in serum sodium concentration. If fluid restriction is impractical, which is often the case in a critically ill patient, demeclocycline is an alternative. Demeclocycline is an antibiotic that inhibits the action of ADH on the collecting duct of the kidney and thereby produces a water diuresis. For patients with severe hyponatremia or significant symptoms, such as seizures or increased ICP, hypertonic (3%) saline may be administered slowly. Concomitant administration of furosemide can reduce the risk of inducing hypervolemia and also increase the free water diuresis.

For salt-wasting syndrome, replacement of intravascular volume with normal saline is the initial treatment. One continues to replace sodium deficits with normal or hypertonic saline. Sodium losses in urine may be replaced in the more chronic phase with enterally administered salt tablets.²⁵⁰

Pharmacologic Paralysis

Data from the TCDB suggested that routine paralysis of all patients with severe head injury increases the risk for pulmonary complications and prolongs the ICU stay.²⁷² However, increases in ICP as a result of agitation, posturing, or coughing should be prevented by narcotics and nondepolarizing muscle relaxants that do not alter cerebrovascular resistance (Table 334-5).^273–283 A reasonable regimen is morphine and lorazepam for analgesia/sedation and cisatracurium or vecuronium as a muscle relaxant, with the dose titrated by twitch response to stimulation.^284,285 Although neurological status cannot be closely monitored while the patient is paralyzed, the muscle relaxants can be withheld once a day, usually before morning rounds, to perform a brief neurological examination.

One complication of neuromuscular blockade that has received little attention in patients with TBI is critical illness myopathy and neuropathy. Weakness in a critically ill patient can be categorized into three general syndromes (Fig. 334-6). Critical illness myopathy was initially described in 1977,²⁸⁶ and numerous cases have been reported since that time.²⁸⁷ The syndrome seems to occur most commonly in asthmatics who require mechanical ventilation for status asthmaticus and who receive a combination of corticosteroids and neuromuscular blocking agents. β₂-Agonists, methylxanthines, and aminoglycosides may also play a role in the development of this syndrome. Prolonged neuromuscular blockade can occur with neuromuscular blocking agents, particularly in patients with kidney or liver dysfunction, and in patients who also receive aminoglycosides.²⁸⁸ Finally, an acute polyneuropathy has been observed in patients with sepsis and multiple organ syndrome.²⁸⁹ There is considerable overlap in these typical syndromes, and a combined myopathy/neuropathy may also occur. Patients with TBI in whom neuromuscular blocking agents are given to control ICP are a high-risk group for these syndromes, but the symptoms are often difficult to differentiate from those of their neurological injury.²⁹⁰ In the proper setting, however, an electromyogram and muscle and nerve biopsy can be helpful in distinguishing weakness as a result of these syndromes from the underlying brain injury.

FIGURE 334-6 Diagram of weakness syndromes in critically ill patients.

The underlying mechanisms of these syndromes are not known, but the association with neuromuscular blocking agents and critically ill patients is strong. Recommendations to minimize the risk for occurrence of these complications include limiting the use of neuromuscular blocking agents, limiting the dose of neuromuscular blocking agents by train-of-four monitoring, measuring creatine phosphokinase daily while neuromuscular blocking agents are given, and stopping administration of the neuromuscular blocking agents at least once a day to observe motor responses.²⁸⁸

Hyperventilation

The relationship of hyperventilation and global CBF has been examined in a series of 171 head-injured patients during the first 10 days after injury.²⁹¹ Of 1212 CBF measurements in these patients, 132 (11%) were less than 25 mL/100 g per minute. Of the 132 low CBF values, 71 (54%) were appropriately reduced relative to the lower CMRO₂, whereas 61 (46%) were associated with increased oxygen extraction or increased cerebral lactate production (or both), thus suggesting a relative inadequacy of perfusion. The incidence of inadequate CBF steadily increased as PCO₂ decreased and was 2%, 4%, 8%, and 23% when PaCO₂ was greater than 30, 25 to 30, 20 to 25, and less than 20 mm Hg, respectively. The incidence of inadequate CBF was twice as high during the first 24 hours than on any other day (8% versus 4%). The incidence was higher in patients with reduced CBF (11%) than in patients with normal or elevated CBF (2%).

Routine hyperventilation (to a PaCO₂ of 20 to 25 mm Hg) has been shown to have a detrimental effect on outcome in one randomized trial²⁹² and is not recommended in the current TBI guidelines (see Table 334-3—level II recommendation). The authors of this study recommended using hyperventilation only in patients with intracranial hypertension rather than as a routine in all head-injured patients. In general, hyperventilation should not be used during the first 24 hours after injury and only if PbtO₂ or SjvO₂ is monitored to make certain that ischemia is not being produced (see Table 334-3—level III recommendation).

In patients who have been chronically hyperventilated, abruptly returning the PaCO₂ to normal can result in a dramatic increase in ICP. Muizelaar and colleagues showed in an experimental study that this phenomenon occurred after 24 hours of hyperventilation and was associated with vasodilation of cerebral vessels as CSF pH equilibrated at the new lower PaCO₂ level.²⁹³ Hyperventilation should be withdrawn over a period of several days to avoid this increase in ICP.

Drainage of Cerebrospinal Fluid

Although removal of 1 mL of CSF normally does not change ICP by more than 1 to 2 mm Hg, in patients with elevated ICP, drainage of 1 to 2 mL of CSF through the ventriculostomy catheter can temporarily lower ICP. This modality can be an important adjunctive therapy for lowering ICP.²⁹⁴ However, as the brain becomes more swollen, the ventricles collapse, less CSF is available for drainage, and the effectiveness of this modality is reduced.

Osmotherapy

Before a method for measuring ICP became available, dehydration therapy was common in the management of head-injured patients. However, normovolemia does not result in an increased rate of intracranial hypertension and eliminates severe electrolyte abnormalities and renal failure.²⁹⁵ Fluid overload, however, should be avoided. Fluid restriction may be necessary in patients with hyponatremia caused by SIADH.

Barbiturate Coma

Barbiturate coma is another treatment modality that has been used to lower ICP in head-injured patients.^49,302,303 In addition to lowering ICP, extracellular concentrations of lactate and excitatory amino acids are reduced after treatment with barbiturates.³⁰⁴ Although routine use of barbiturates in unselected patients has not been consistently effective in reducing morbidity or mortality after severe head injury,^305,306 a randomized multicenter trial demonstrated that instituting barbiturate coma in patients with refractory intracranial hypertension resulted in a twofold greater chance of controlling ICP.³⁰⁷

Because of the hypotensive complications associated with barbiturates and because neurological evaluation is unavailable during treatment, barbiturate coma is generally reserved for patients with intracranial hypertension resistant to other modalities. Pentobarbital is given in both loading and maintenance doses. The loading dose is 10 mg/kg given over a 30-minute period, followed by 5 mg/kg each hour for three doses. This regimen typically provides a therapeutic level after the fourth dose. The maintenance dose is 1 to 2 mg/kg per hour adjusted so that either the serum level is in the therapeutic range of 30 to 50 µg/mL or the EEG has a burst suppression pattern. Winer and coauthors reported that plasma and CSF pentobarbital levels do not accurately reflect the physiologic effects of pentobarbital and recommended monitoring the EEG instead of pentobarbital levels.³⁰⁸ Pulmonary wedge pressure and cardiac output are monitored in all patients. Hypotension caused by pentobarbital is treated first by volume replacement and then with dopamine if necessary. Laboratory studies suggest that for the treatment of hypotension associated with barbiturate coma, volume resuscitation may be better than dopamine.³⁰⁹ In these studies, dopamine infusion increased cerebral metabolic requirements and partially offset the beneficial effects of barbiturates on metabolism.

The mechanism of ICP reduction by barbiturates is not entirely clear but is usually considered to be hemodynamic because of the immediate effect on ICP. Studies by Messeter and associates have suggested that the reduction in ICP with barbiturates is closely tied to the retention of CO₂ reactivity by the brain.^72,310 Studies by Cruz showed that outcomes are significantly worse in patients who have a reduction in SjvO₂ to less than 45% with the initiation of barbiturate coma.³¹¹ Complications occurring during treatment of intracranial hypertension with barbiturate coma have been reported to include the following: hypotension in 58%, hypokalemia in 82%, respiratory complications in 76%, infections in 55%, hepatic dysfunction in 87%, and renal dysfunction in 47% of patients.³¹²

In a study of 67 patients in barbiturate coma for refractory intracranial hypertension, the most consistent cerebral effect was a reduction in CMRO₂ by an average of 31%.³¹³ The change in CMRO₂ with the loading dose of pentobarbital was closely related to the pretreatment value (n = 67, r² = 0.65, P < .001). In the 30 patients with a good ICP response to pentobarbital infusion, pretreatment CMRO₂ and AVDO₂ were greater, and CMRO₂ and AVDO₂ decreased more with the loading dose of pentobarbital than in patients with a partial or no ICP response. Outcome was significantly better in patients with a good or partial ICP response to pentobarbital, with 21% of these patients having a good recovery or moderate disability at 3 months after injury as opposed to a persistent vegetative state or death in 100% of the nonresponders.

In summary, barbiturate coma can be a useful agent for acutely reducing ICP in selected patients. Patients with overwhelmingly severe injuries are not likely to benefit, partly because their CMRO₂ is already markedly reduced by the injury and partly because their outcome is already predetermined by the injury. Patients with systemic hypotension are not likely to have a good response because hypotension limits the amount of barbiturates that can be given.

Hypothermia

Therapeutic hypothermia has been investigated as a possible neuroprotective strategy for the prevention or reduction of brain injury as a result of several causes. Possible mechanisms include reductions in the cerebral metabolic rate, increased ICP, cerebral edema formation, frequency of epileptic discharges, and opening of the BBB.³¹⁴ Hypothermia also inhibits the inflammatory response and the release of glutamate, nitric oxide, and free radicals associated with TBI.³¹⁵ In a systematic review, McIntyre and coauthors reported an overall beneficial effect of moderate hypothermia (32°C to 33°C) in patients with severe TBI: a 19% reduction in deaths and a 22% reduction in poor neurological outcomes relative to normothermic patients.³¹⁵ The review indicated that the therapy should be continued for at least 24 hours, with temperature ranging between 32°C and 33°C and rewarming period lasting less than 24 hours. Evidence from six randomized controlled trials did not clearly demonstrate that hypothermia is associated with consistent and statistically significant reductions in mortality.^316–320 However, patients treated with hypothermia have a favorable neurological outcome. Preliminary findings suggest that hypothermia may have a better chance of reducing mortality when cooling is maintained for more than 48 hours. Based on these results, hypothermia is limited to a recommendation level III in the guidelines for the management of TBI.³²¹ There are two different methods to induce hypothermia: systemic and selective hypothermia. Most of the studies conducted have involved systemic hypothermia; the most frequent risks associated with this method include cardiovascular and pulmonary complications, infections, and increased rates of thrombocytopenia. Methods to induce selective hypothermia include surface cooling, intranasal selective hypothermia, endovascular cooling, and epidural cerebral cooling. From these choices, only surface cooling and intranasal cooling have been tested in humans.³²²

Decompressive Craniectomy

Decompressive craniectomy has been used to treat uncontrolled intracranial hypertension of various origins, including cerebral infarction, trauma, SAH, and spontaneous hemorrhage. Patient selection, the timing of surgery, the type of surgery, and the clinical and radiologic severity of the brain injury are all factors that determine the outcome of this procedure. The effectiveness of very early decompressive craniectomy was evaluated in one randomized clinical trial that included 27 children who suffered TBI. There was a reduced risk ratio for death of 0.54 (95% confidence interval [CI], 0.17 to 1.72) and a risk ratio of 0.54 for death, vegetative status, or severe disability 6 to 12 months after injury (95% CI, 0.29 to 1.07).³²³ Other available studies in adults are either case series or cohorts with historical controls. Their results suggest that decompressive craniectomy effectively reduces ICP in most (85%) patients who have intracranial hypertension refractory to conventional medical treatment.^323,324 In another case-control study, Polin and coworkers reported improved outcomes in a selected group of patients with diffuse brain swelling who underwent bilateral frontal decompressive craniectomy.³²⁵ A favorable outcome was observed in 46% of the patients who underwent surgery within 48 hours of admission. Brain oxygenation measured by tissue PO₂ and blood flow estimated by MCV-FV are also usually improved after decompressive craniectomy.^326,327 The results of a recently published case series of 33 patients with severe TBI who underwent decompressive craniectomy showed that the long-term (3 years) results justify the use of decompressive craniectomy and that good clinical results are seen in up to 40% of patients who were otherwise most likely to die,³²⁸ although further studies will be required to define the parameters of use, including the timing of surgery, the physiologic threshold, the choice of procedure, and the age limit. Randomized controlled trials in patients with TBI are ongoing (Rescue ICP and DECRAN).³²⁹

Treatment of Hypotension

Hypotension in a head-injured patient is treated in a manner similar to any other critically ill trauma patient, except that the goal must be to provide CPP that is adequate for the injured brain. Initially, intravascular volume should be assessed, and if low, volume should be given. A central venous pressure monitor or in some cases a Swan-Ganz catheter is appropriate to assess intravascular volume. Low intravascular volume should be treated by infusion of crystalloids, colloids, or blood as appropriate for the clinical situation. There is uncertainty about the best choice of fluids, and both crystalloid-based and colloid-based resuscitation strategies have been advocated. From the Saline versus Albumin Fluid Evaluation (SAFE) study, a post hoc follow-up study of patients with TBI (the SAFE-TBI study) was undertaken.³³⁵ A higher mortality rate was observed in patients with severe TBI who received 4% albumin than in those who received saline, thus suggesting that saline is preferable to albumin during the acute resuscitation of patients with severe TBI.³³⁶

Hypertonic saline has been found to allow replacement of intravascular volume with less of an increase in ICP than is the case with saline or dextran.²⁹⁹ In one multicenter study of hypertonic saline for resuscitation of trauma patients, the head-injured patients treated with hypertonic saline had significantly better survival to discharge.³³⁷ If volume replacement does not provide an adequate BP, inotropic or pressor agents, or both, should be used.

The underlying cause of the hypotension must be sought and treated. Hypotension in an adult is only occasionally caused by head injury alone, and another mechanism should be sought.³³⁸ Blood loss from another injury, an associated spinal cord injury, cardiac contusion or tamponade, and tension pneumothorax are other causes of hypotension to be considered.

Treatment of Hypoxia

Treatment of hypoxia is similar to that used for other critically ill trauma patients. Increasing FIO₂ will often improve oxygenation. When pulmonary edema is present, oxygen supplementation may not be sufficient to improve arterial PO₂. The addition of PEEP may increase functional residual capacity and improve the ventilation-perfusion mismatch. As a result, PEEP may allow the inspired oxygen concentration to be decreased and hence reduce the potential for oxygen toxicity. However, PEEP can have adverse effects on ICP and BP. PEEP can raise ICP by increasing intrathoracic pressure, central venous pressure, and cerebral venous pressure. In addition, by decreasing venous return to the heart, BP can be reduced. The combined effect on ICP and BP is a reduction in cerebral perfusion.

The severity of the effect of PEEP on ICP depends on both intracranial and pulmonary compliance. In patients with normal intracranial compliance, ICP does not change significantly because the brain is able to compensate for the increased cerebral venous blood volume caused by an increase in intrathoracic pressure. In contrast, when intracranial compliance is decreased, even small increases in cerebral venous pressure induced by PEEP may cause dangerous increases in ICP. Apuzzo and associates studied the effects of PEEP in 25 head-injured patients.³³⁹ ICP increased in 12 of the patients, all of whom had decreased intracranial compliance. CPP decreased to less than 60 mm Hg in 50% of the patients with increased ICP. Shapiro and Marshall found that 6 of 12 head-injured patients had a 10–mm Hg increase in ICP with the addition of 4 to 8 cm PEEP.³⁴⁰ In most cases, elevating the patient’s head reduces the effect of high airway pressure on ICP.³⁴¹

The extent to which any given amount of PEEP is transmitted to the venous system also depends on the severity of the pulmonary injury. When the lungs are poorly compliant, exposure to PEEP does not usually cause marked increases in venous pressure or ICP.^342,343

Removal of PEEP must also be judicious. Sudden cessation of PEEP may precipitate a sudden increase in central blood volume and a rise in arterial blood pressure and CPP, thereby resulting in an increase in ICP.^342,344

Treatment of Anemia

Anemia can easily be remedied by transfusing packed red blood cells. The optimal hematocrit for improving oxygen delivery after head injury is not entirely clear. Studies that examine the relationship between CBF and hematocrit have demonstrated that CBF and tissue oxygen delivery increase as the hematocrit and viscosity decrease until the hematocrit falls to 33%. Below a hematocrit of 33%, tissue oxygen delivery decreases.³³⁰ In normal adults, adequate oxygen-carrying capacity is usually met by a hemoglobin concentration of 7 g/dL or a hematocrit of 21% when intravascular volume is normal. However, because cerebral vascular responses are impaired after TBI, it cannot be assumed that the same hematocrit will be adequate in a head-injured patient. In a recent study of 35 patients with severe TBI and SAH who had hematocrits less than 25%, packed red blood cell transfusion increased local brain tissue oxygen saturation in 75% of the patients.³⁴⁵ Current evidence suggests a favorable outcome with a restrictive transfusion practice in the general ICU population. The optimal transfusion threshold in critically ill neurological patients is unknown. In selected patients, a hematocrit of greater than 25% to 30% may be required for maximal oxygen delivery to the brain.³⁴⁶

Treatment of Seizures

For a head-injured patient who is actively convulsing, the seizure activity should be arrested immediately. Diazepam, 5 to 10 mg intravenously, or lorazepam, 2 to 3 mg intravenously, is given initially to stop the seizure. The action of these drugs is short-lived, and the seizures may recur if other antiepileptics are not given. Intravenous phenytoin should be administered at 25 mg/min until a loading dose of 15 to 20 mg/kg has been given. Maintenance doses of phenytoin, 300 to 400 mg/day, are administered to prevent subsequent seizures.

Treatment of Cerebral Vasospasm

Symptomatic vasospasm after TBI is treated similar to vasospasm after SAH. Nimodipine may be effective in reducing the neurological consequences of vasospasm after head injury. It is not clear that the beneficial effect of nimodipine after head injury is via an effect on vasospasm. However, the only subgroup of head-injured patients who have been found to benefit are those with SAH.³⁴⁷

Hypervolemic hemodilution and induced hypertension are the primary therapies for vasospasm. The patient should undergo volume expansion. Usually, a Swan-Ganz catheter is indicated to optimize cardiac output. If the patient is still symptomatic after volume expansion, a pressor agent should be added to raise BP. The choice of pressor agents is controversial, as discussed under the topic of treatment of secondary ischemia. Information about cardiac output obtained from the Swan-Ganz catheter can be helpful in this decision. If cardiac output is normal or elevated, a pressor agent such as phenylephrine or norepinephrine is likely to be more effective in raising BP. If cardiac output is reduced, dopamine may be a better choice, but it should be remembered that dopamine has been observed to have an unpredictable effect on CBF in some studies.^332,334 In patients with vasospasm caused by a blast-type injury, aggressive open surgical and endovascular treatment may improve outcome. In a recent report of vasospasm associated with blast injury, patients received medical management and aggressive endovascular interventions (endovascular nicardipine, angioplasty, or both); although no early improvement in neurological function was noted, there was a clear improvement in blood flow velocities measured by TCD.⁶⁸