INTRODUCTION:

Head injury remains the most common cause of death in children and adolescents in most developed countries, although the cause of the trauma varies by age group and by country. The etiology ranges from falls in younger children to bicycle accidents, automobile injuries, and homicide and suicide in teenagers. The category of non-accidental trauma or child abuse is addressed in Chapter 18. It is clear that with appropriate education and funding the incidence of traumatic injury to children can be limited.' In many countries the incidence of traumatic injury seems to be decreasing, for unclear reasons.2 Despite the decrease, the number of children dying in the USA per year from trauma equals or exceeds the natural death rate. Despite extensive experimental work there are no new drugs that are currently available that have been shown to have an effect in improving the outcome from head injury in children. While the mortality rate remains low, 20-40% for Glasgow Coma Score (GCS) 3-8, the degree of functional recovery is limited and most if not all patients suffer some alteration in function, neurologic, cognitive, or behavioral.

Only 7-10% of head injuries are severe as defined by a GCS of 8 or more. Only 20-30% of children with severe head injury will require a neurosurgical operation and, since the mechanism of injury and the response to trauma is very different between children and adults, the reasons for surgical intervention are different from those in adults.  

EPIDURAL HEMATOMAS

Epidural hematomas are different in children than adults with regard to the cause of the hemorrhage, the location of the hematoma, the incidence of skull fracture, and the clinical presentation. In adults the usual cause of the hemorrhage is arterial in origin; venous or bony injury are more common causes of epidural hematoma in children. Because of the difference in etiologies, which can result in a more benign clinical picture, it is becoming increasingly common to avoid surgical intervention in neurologically intact children with a small or moderately sized epidural hematoma in the frontal or parietal regions, particularly if they have an overlying skull fracture. However, risk factors for deterioration include a fracture traversing a meningeal artery, vein, or sinus, or computed tomography (CT) of the head within 6 h of injury showing an increase in the size of the lesion. All patients with acute epidural hematomas should be admitted to the intensive care unit (ICU) for observation.

Only one-third of children with an epidural hematoma will be comatose and therefore rate as having severe head injury, and in these children there may be an underlying parenchvmal injury, diffuse or focal, that is contributing to the coma. This is always the case in children who have been unconscious from the time of injury and in these two groups of children an intracranial pressure (ICP) monitor should be placed in the operating room after clot removal to allow appropriate treatment of the diffuse brain injury. In children who were awake following trauma and then deteriorated to a GCS greater than 8, the expectation is that after clot removal rapid recovery of consciousness will occur since the presumed cause of the neurologic state was mass effect and raised ICP from the hematoma. In this group an ICP monitor may not be required unless rapid recovery of consciousness does not occur following surgery.

In patients who are considered surgical candidates, appropriate preoperative planning is required despite the urgency of the situation. It is rare that any surgical intervention is required before entering the operating room since alterations in ICP can usually be controlled by mannitol and endotracheal intubation, hyperventilation, and sedation. Only in a life-threatening situation will an emergency twist drill hole in the emergency room be required.3 The danger of such a procedure in a small child is that the intracranial bleeding then begins again or increases in rate and cannot be stopped until appropriate intracranial exposure is obtained. Because of small blood volume, if the operating room is not available immediately the blood loss can lead to shock and either secondary injury or death. Thus this should never be done without adequate blood for replacement available in the emergency room. Furthermore, there is rarely a reason for blind surgery (i.e. exploratory burr-holes) unless a CT scan is unavailable within 30 mm or less and the child is deteriorating,4 or there is a fracture seen on radiography whose location is compatible with the clinical picture of epidural hematoma. In children under 8 years the location of the epidural hematoma is rarely low in the temporal fossa but is more likely to be higher and more posterior, the tear of the middle meningeal being at a suture line, often the squamosal. Thus any exploratory burr-hole surgery in the absence of a CT scan or fracture needs a different positioning of the burr holes than the classic adult location, and one hole should be over the midportion of the squamosal suture to avoid missing a hematoma that is higher than the usual temporal hole and lower than the usual parietal hole.

In patients in whom the decision is made to proceed with surgery, the prime goal, if there is a low GCS, is to relieve the ICP as quickly as possible. Despite this it is usually possible to place a cosmetically located incision with an area for craniotomy that is large enough to expose the likely site of bleeding. As with all trauma craniotomies, it is necessary to ensure that adequate venous access and blood are available, preferably cross-matched blood, before starting the procedure. A formal craniotomy is then performed, often with a single burr-hole placed over the superior or inferior margin of the hematoma. This allows immediate decompression of the hematoma and the flap is then made to include any fracture within the boundaries of the craniotomy if possible. The bone should not be rongeured away since the closure should always include reconstruction of the skull to avoid the need for later cranioplasty. Even open fractures are repaired this way. If preoperative CT reveals only the epidural hematoma, opening the dura should be avoided to minimize the risk of injury to the underlying brain. Many authors advocate 'tacking' the dura to the bony opening to minimize the risk of reaccumulation of the blood but this has never been shown to be effective or necessary and the decision is usually based on either the surgeon's usual routine or the degree of laxity of the dura after hematoma removal. If the dura does not expand then tagging sutures may be helpful. The important part of the operation is to be sure that all epidural bleeding has stopped before closure. A drain cannot be relied upon to remove continued hemorrhage adequately. Fixation of the skull may be done using wire or slowly absorbing suture. Plating systems are rarely required and add significant expense to the procedure with no defined value.

SUBDURAL HEMATOMAS

Subdural hematomas are often associated with non-accidental trauma in children. In children acute subdural hematomas of sufficient size to require surgery are rare. Smear subdural hematomas in association with diffuse brain injury or brain swelling are rarely operated on since they play little or no role in the etiology of the coma. In addition, after decompression of the hematoma, severe brain swelling often occurs, as well as additional brain edema and at times direct injury to the cortex as the surgeon tries to close the dura. The indications for subdural hematoma removal are that the midline shift is predominantly accounted for by the subdural hematoma, not associated brain swelling, or that, despite medical measures, the ICP remains high. In children the subdural hematoma can be in the inter hemispheric fissure or over the tentorium, and surgery has no role in this setting. In children with an open fontanelle the interhemispheric subdural hematoma can be drained by tapping the fontanelle.

Surgical management of acute, traumatic, subdural hematomas consists of evacuation and hemostasis. Care must be taken in opening the dura as too large a dural opening may allow edematous underlying brain to fungate out of the opening, exacerbating the injury to the brain in the area. On the other hand if the dural opening is too small, the area of hemorrhage may not be able to be seen. Hemostasis often involves finding a cortical or bridging vein that has been torn b the trauma, and coagulating it or packing with a hemostatic agent. Once hemostasis has been achieved, the dura should be closed carefully and the bone flap replaced. If brain edema is severe, such that the dura cannot be closed, then a pericranial patch, a temporalis flap, or a dural substitute should be used to achieve closure and prevent cerebral herniation through the defect. Once the dura is closed the bone flap can simply be laid on top of the dura without sutures or placed under another area of the scalp or in the abdominal wall. If the patient survives, cranioplasty will be required in the future and if the original bone is available this facilitates the process. If the patient had a GCS of 8 or less, a motor score of 5 or less or, in patients too young to be able to follow commands, an impaired level of consciousness before operation, an ICP monitor is indicated before leaving the operating room since there is almost always associated cerebral injury with accompanying raised ICP, which may require intensive medical therapy.  

FOCAL CONTUSION

Focal contusions are not often seen on head CT in children with a severe head injury, although follow-up studies of survivors show such lesions in 50% of children.5 It is rare that a contusion is considered a surgical lesion in children as the tissue can often recover and its excision does, of course, make this impossible. Most contusions contribute to intracranial hypertension and are treated by medical means. The treatment of raised ICP with multiple contusions is also by medical management. Children with a GCS lower than 8 with multiple contusions on the CT scan are at high risk for raised ICP and secondary deterioration, and despite a good GCS should be observed in the ICU.

Areas that may be safely debrided without clear neurologic sequelae follow guidelines for craniotomies in general, specifically the frontal and anterior temporal lobes with less risk in the nondominant hemisphere. Contused brain usually aggravates the injury because of local mass effect, and clinical improvement will often be greatest if local pressure can be controlled without surgery since brain that is left in place can recover whereas resected brain obviously cannot.

In the rare case where surgery is done for resection of a contusion, a generouse trauma craniotomy may be required. The clotting status of the patient should also be assessed before operation as many head-injured patients become coagulopathic. The dural opening should begin over the area of the brain that is least vital and be carried to the most vital area to he exposed. Ideally the dura is left closed over eloquent brain to prevent injury from herniation and local vascular compression. Care must be taken with brain retraction in the severely head-injured patient as retraction may exacerbate edema. A safe, direct route should be taken to the contusion and the resection done by suction with ongoing hemostasis with the bipolat. In the soft swollen brain, venous bleeding within the cerebral tissue may be difficult to control but complete hemostasis must be the goal. Lining the bed with a thin layer of a hemostatic agent such as Surgi-cel may aid in stopping oozing from the injured brain. Concerns in closure are the same as those with subdural hematomas.  

DEPRESSED SKULL FRACTURES

Open depressed fractures require surgery despite a poor neurologic state. If the dural laceration is not repaired, the occurrence of raised ICP will lead to progressive cerebral hernjation and venous infarction of cerebral tissue. The risk of intracranial infection is increased the longer the fracture is left open, and this is another reason for closure. Fractures of the skull base that produce cerebrospinal fluid (CSF) leakage from the cribiform plate or the ear rarely require surgery and almost never in the face of severe diffuse brain injury. Because CSF leakage occurs early, the ICP may remain low for the first few' days until the leakage stops and then it will rise. In this setting the ICP monitor should not be removed until 48 h after the CSF leak has stopped.

Severe basal frontal fractures of the type that occur with slow crushing injuries of the head6 can be very difficult to handle. In this setting there is the risk of injury to the anterior cerebral arteries with risk of secondary hemorrhage plus traumatic encephalocele of the floor of the frontal fossa. If the traumatic encephalocele is not repaired then secondary damage can occur to the basal frontal lobes and possibly the hypothalamus as a result of anterior and inferior herniation. On the other hand repair of these lesions in the face of acutely raised ICP is difficult and runs the risk of brain injury secondary to surgerv and retraction injury. The decision on the correct timing for surgery is difficult. If there is frank cerebral herniation through the frontal skull base on the initial CT or magnetic resonance imaging (MRI) scan, if MR angiographv suggests injury to the anterior cerebral artery, or if a large frontal hematoma is present, the best time to operate is as early as possible before severe brain swelling occurs. Whether other facial fractures should be fixed at this time will depend on the exposure, the clinical state of the child, and the experience of the operating team with this type of injury. If there are other surface compound fractures that require surgerv then all the lesions should be taken care of at the same time, if possible, with pericranial grafts to the frontal skull base and bony reconstitution of the base to prevent renewed herniation.

Other penetrating wounds should in general be debrided but this will depend on the clinical state of the child, the clotting studies, and the expectation of survival. In the urban population over the age of 10 years, gunshot wounds are now the most common cause of penetrating head injurv in children. Gunshot wounds to the brain are operated on only if there is a chance of survival and recovery.7 Most of the injury associated with gunshot wounds is caused by dispersion of kinetic energy along the brain. The speed and resulting heat of the bullet usually sterilizes the bullet but bone fragments remain a source of possible infection. With a GCS of 3, the mortality rate approaches 100% and an argument can be made that no surgery will alter the outcome. A higher GCS is an indication for debridement and closure of the dura. Skin should likewise be debrided and closed. Again, preoperative CT is valuable in assessing the course of the bullet and associated hematomas. All patients who have sustained a gunshot wound of the brain and are considered to be viable require an ICP monitor.

Dog bites and other penetrating injuries are usually associated with a preserved level of consciousness and in these cases debridement is usually performed early. For patients with penetrating head injuries caused by a sharp instrument in whom the GCS is low and surgery necessary, care must be taken not to disturb the instrument until proximal and distal control had been obtained in the operating room. If possible, preoperative CT should be done to localize the trajectory and depth of the injury. The site of injury is often around the face, specifically through the eye. In those cases, intradural and extradural exploration will maximize the ease of removal of the instrument. Should bleeding be encountered, one must be ready for rapid exploration. Every effort should be made to remove the entire foreign body and close the dura. The bone should then be reconstructed. Intravenous antibiotics are indicated for a minimum of 3 days, more based on extent of contamination.

Raised ICP without focal mass is a multifactorial pathophysiology which includes, brain swelling, diffuse axonal injury, ischemia, contusion, or edema. Appropriate treatment of this entity is non-surgical (see below). In rare cases where the raised ICP can be controlled, but only transiently, by altering medical management, decompressive craniectomy has been proposed. This is rarely necessary in children and is appropriate in only a small number of cases. Care should be taken to preserve the bone, usually in a bone freezer, for replacement after resolution of the acute traumatic episode.  

Different pathology at different times

The result of traumatic brain injury is not a single pathologic event but a series of pathophysiologic changes that vary both in severity and over time such that no two head injuries are quite the same. The surgical mass lesions usually occur early after trauma and in general the sooner they are removed the better; certainly, the better the GCS at the time of removal the greater possibility of a good outcome. The other pathologic processes that are grossly apparent on scan and at autopsy are DAT, contusions, areas of focal ischemia, diffuse brain swelling, brain edema, increased CSF, and hydrocephalus. Vascular factors are also important; although the true incidence of vascular spasm is unclear, this is certainly a cause of secondary deterioration that has not received adequate attention. Even DAT is and the very early axonal lesions may repair themselves and are then followed by secondary axonal damage due to progressive molecular events occurring within the axonal cytoplasm. Many other molecular alterations have been described and the expression of these agents varies with time after the injury. 18-25 Two of the more important molecular mechanisms, at least in animal models, have been free radical production and the release of neurotoxic neurotransmitters. Free iron and excess oxygen are triggers for free radical formation. The presence of tissue acidosis aggravates the tissue increases in the neurotoxic transmitters by further inhibiting uptake. Nitric oxide may also interfere with tissue reuptake of glutamate and aspartate. Here again, a cascade of events occurs over time. These areas of molecular change are currently the focus for pharmacologic efforts to modify the injury. As will be reiterated several times, is becoming clear that the pathology associated with death of the child may be different from that associated with survival, especially good quality of survival, and thus the benefits of chemical modulation may be seen in the quality of survivors' lives rather than in a decrease in mortality rate.

Brain swelling and brain edema may have different time courses after injury, and some of this may be dependent on the occurrence of early ischemia and hvpoxia. Brain swelling and loss of CSF spaces remains a common finding on the CT scan after head injury. Initial studies suggested this was due to vascular engorgement and also that increased cerebral blood flow (CBF) could be contributory. Experimental studies do show increased blood volume as a major component of the early swelling and clinical findings also support this thesis. Indeed, careful review of the CT scan suggests at least two patterns and maybe three, depending on the classification of the scan. In patients who present with a low GCS, 3 or 4, the brain density on the CT scan may be low, suggesting ongoing ischemia. In these children the ICP is usually over 30 torr despite resuscitation, the CBF is low, and morbidity is high. There is another group of children with normal or hyperdense brains in whom diffuse swelling is also present but in whom ICP is normal or easily controlled, and this group overlaps with those with increased CBF. Recent studies of CBF in adults and children show that CBF is usually at its lowest level in the first few hours despite adequate resuscitation and normalization of systemic parameters. Over the next several days CBF increases in those who survive and death is most likely in those with the lowest early CBF coupled with minimal increase in the following days. While CBF is increasing CMRo₂, is usually decreasing, resulting in a further increase of jugular bulb partial pressure of oxygen (PO₂). In most patients, even those with low CBF, the AVDo₂ is low and cerebral metabolism does not appear to be in excess of cerebral flow. Those with the lowest metabolism are most likely to die, yet if recovery occurs there is no apparent correlation with early metabolism and degree of recovery. Increased metabolism can occur and there are probably regional variations but ongoing evidence for ischemia is poorly substantiated. In children with increased CBF a late occurrence of further excessive CBF with increased ICP may be a sign of loss of auto regulation and is also associated with high morbidity. While the early swelling is most likely due to increased blood volume not necessarily accompanied by increased blood flow, a recent experimental study has suggested that the early swelling may be due to increased cerebral water, although the exact location of the water has not been clarified.48 In conclusion, the finding of diffuse brain swelling is more common in children than in adults. The exact cause is not proven but it is likely that there are several physiologic events that produce a similar pattern: hyperemia, which may be rarer than first suggested; vascular congestion yet without hyperemia; and increased water content, brain edema within cells. The clinical course and outcome will depend on the cause of the swelling which, in turn, is modified by the immediate posinjurv occurrence of ischemia and hypotension.

Correlation of physiologic findings with recovery becomes important since current studies suggest that early low CBF carries a very poor prognosis, and that diffuse brain swelling when resulting from ischemia, portends a poor outcome whereas when not accompanied by low CBF is associated with good recovery. Is the outcome already programmed in the first few minutes after injury or can it be modified by therapy and, if so, which pathologies can be modified by current therapy?

Cerebral contusions are rarely seen acutely yet follow-up studies of severely head-injured children show that approximately 50% will have evidence of frontal basal and anterior temporal cortical injury. There is a high incidence of ischemic injury in children who die after head trauma yet focal ischemic lesions are rare in follow-up MRI studies. DAT is a common autopsy finding in children who die after head trauma yet in follow-up MM studies is seen in only a small proportion, 15%. These findings may suggest that children who survive and recover from the injury may not simply have incurred less extensive injury than those who die but also may have different pathology.

Even in the acute ICU care phase of the head-injured child, there is not a single pathology to be considered but a continuing variation of potential pathophysiologic changes, some of which can be identified by clinical monitoring and are amenable to manipulation now, and some that we cannot identify or treat at present. An awareness of the changing pathology is the underpinning of current therapy, and modern monitoring techniques (e.g. ICP, AJD0₂, CBF, transcranial Doppler ultrasonographv and evoked responses) are designed to help identify these changes.  

Non-surgical management

Much of the pathophysiology of head trauma is non-surgical and even in children with a GCS of less than 8 who require surgery there is almost always underlying brain injury that is not treated by the operative procedure. It is rare that traumatic unconsciousness in children is caused or perpetuated by increased ICP, although raised ICP will occur in 75% of severely head-injured children. The few situations in which a direct relationship can be established between local or global pressure changes and altered state of consciousness are related to minor or moderate trauma with the delayed development of epidural or subdural hematomas or acute brain swelling. In both the former cases, if the clot is removed before the onset of coma, rapid and almost complete recovery is to be expected because there is minimal underlying primary brain injury. If clot removal is delayed then secondary brain injurv can occur related to cerebral herniation and increased ICP with decreased CBF and resultant ischemia. The current management of head trauma is based on the concept of primary and secondary injury. The primary injury is that which occurs in the few' milliseconds in which the trauma occurs and remains untreatable by any direct means. Secondary injury includes all the events that may be triggered by the primary injury but progress to cause additional damage to the brain.

The primary injury consists of damage to all layers of the cranium: scalp, skull, and intracranial contents. Intracranial contents that are injured include the neurons, the cell body, dendrites and axons, glial cells, myelin sheaths, and blood vessels. Recent studies have identified a cell structure connecting the cell membrane to the nucleus, and traction on this skeleton can affect gene expression. This mechanism may account for some of the molecular changes that occur within the cells following trauma.

Secondary injury is composed of the pathologic processes that begin at the time of the immediate injury and are progressive. They include molecular events (e.g. neurotoxic injury, lipid peroxidation, and diffuse axonal injury) as well as new pathophysiologic injuries that occur as a result of systemic hypoxia, hypercarbia, hypotension, and intracranial hypertension. In addition, secondary injury can be precipitated by vascular spasm, coagulopathy, electrolyte disturbances, and seizures. Some injuries result in a clear combination of primary and secondary pathology (e.g. epidural hematoma). The tear of the middle meningeal artery is primary but, as bleeding continues or restarts, the local pressure in the brain and the ICP rise to produce secondary damage from cerebral herniation of the temporal lobe with local brainstem ischemia and/or by diffusely raised ICP affecting cerebral perfusion and resulting in general cerebral ischemia from low CBF.

There is ongoing debate as to how often there is continuing cerebral ischemia after adequate resuscitation since the presence or absence of such pathology should have a major influence on what the most appropriate early therapy should be. While early oligemia occurs, as measured by CBF studies, there is no evidence suggesting that this represents ongoing ischemia since the AJD07 is usually low and at least global CBF is adequate for metabolism. There is no evidence that this low flow can be increased to raise metabolism, and the oligemia seems to be a reflection of severe injury with the prime change being decreased metabolism. The therapy of head injury is still focused primarily on the prevention of secondary injuries, beginning at the injury site and continuing through transport, the emergency room, radiology, and the ICU. Despite much experimental evidence, no reliable drugs are yet available to interfere with the molecular, secondary injury. To date neurotoxic blockers, calcium channel blockers, and free radical scavengers have not been demonstrated to have any benefit.

The systemic pathologic processes that are modified by therapy are hypoxemia, hypercarbia, and hvpotension. The intracranial pathology consists of focal mass lesions, usually hematomas; brain swelling; brain edema; brain contusion; diffuse axonal injury; and raised ICP either locally producing focal brain herniation or generally producing a lowering of CBF and generalized brain ischemia. Brain edema can occur in the cells themselves, neurons and glial cells (intracellular or cytotoxic), usually as a result of inadequate substrate, most commonly oxygen, or it can occur in the extracellular space, usually of the white matter, as a result of damage to the blood-brain barrier in the capillary vessels (vasogenic). Vasogenic edema is uncommon in the first 24-48 h after trauma except surrounding an intracerebral hematoma. The early low-density changes seen in the brain soon after trauma (focal low density on CT scan) are probably the result of ischemia and hypoxia that occurred at the accident scene and represent cvtotoxic edema, although exactly which cell population is primarily involved is not clear. This 'low-density' brain can occur very early after the injury and seems to occur especially in the infants who suffer nonaccidental injury. In older children it is more common to see the changes of loss of gray matter to white matter differentiation occurring 3-5 days after injury at a time when the ICP is a problem to control. Once again, the question is whether this is the result of an ischemic or hypoxic injury at the time of the trauma or progressive ischemia despite normal perfusion pressure for several days. The former seems more likely since there are no studies of CBF showing ongoing ischemia in the presence of normal ICP.

CBF studies in children have shown variable results, although most have measured some degree of hyperemia after 24 h.  Study of CBF early, 12 h following trauma, suggest two populations of injured children: one in which the early CBF is low (<40% of normal) and one in which the CBF is near normal (<75% of normal). These two populations both show an increase in CBF at 48 h but in the former group the flow is greater than 40 ml per 100 g per mm and in the latter less than 75 ml per 100 g per mm. Mortality and serious morbidity rates are much higher in the former group.6° Pressure auto regulation, chemical auto regulation, and carbon dioxide responses of the cerebral vessels seem to be intact except in the most severely damaged brains.

Cerebral metabolism after severe head injury may be decreased or increased and the changes may be regional, similar to those shown in animal models. Measurement of regional cerebral metabolism cannot be easily obtained in the acute stages of head injuries in children. Since the cerebral metabolism rate (CMR) can be calculated as the AV difference x CBF/100, the jugular venous oxygen content or saturation can give some idea of the balance between flow and metabolism, and may therefore be helpful in the selection of therapy to lower ICP or increase systemic arterial pressure (SAP).

Normal intracranial homeostasis permits reciprocal changes in volume to occur between the various intracranial components - blood volume, CSF, and brain - such that an increase in the volume of one component can be offset by a decrease in the volume of the others, and thus the ICP kept relatively constant. Also, the length of time for which the ICP can be raised is limited by this same homeostatic mechanism. Unfortunately, after severe trauma all the components of the intracranial space may be increased simultaneously or sequentially; thus the high incidence of intracranial hypertension is not surprising. The blood volume can be increased by:

1. severe hypoxia Pa0₂> 50 mmHg;

2. hypercarbia;

3. cerebral hyperemia;

4. raised ICP and vascular congestion;

5. patient position and venous distension;

6. systemic hypotension or decreased cerebral perfusion pressure (CPP);

7. defective autoregulation and decreased vascular tone.

The CSF volume can be increased by:

1. hyperemia and increased CSF production;

2. ventricular obstruction by swelling or mass effect;

3. increased outflow resistance due to

1. brain swelling and compressed arachnoid pathways

2. subarachnoid hemorrhage

3. cerebral herniation.

Brain volume can increase as a result of:

1. intraparenchymal hematoma;

2. cerebral contusion;

3. brain edema (cytotoxic, vasogenic).

Finally, there can be the addition of new volume to the intracranial space as a result of subdural or epidural hemorrhage. As a result, the normal homeostatic mechanisms are blocked and therefore minor changes in intracranial volume can result in large increases in ICP and it must be assumed that these children are functioning on the steep portion of the volume-pressure curve. In addition, because of the small intracranial volume in the child, the actual volume required to increase the ICP is small: a pressure-volume index (PVT) of 5 ml in children aged under 1 year. This accounts for the high frequency of raised ICP in children after severe head injury.

Direct vascular effects, predominantly vasospasm, are probably common in severely injured children since subarachnoid hemorrhage on the first CT scan has been reported in up to 75% of such children63 and adult studies have shown a strong correlation between subarachnoid hemorrhage and symptomatic vasospasm after trauma.'3 Care of the head-injured child and selection of monitoring and therapy are based on an understanding of the ongoing pathophysiology.  

EMERGENCY RESUCCITATION

The primary approach, which begins at the accident scene, is based on the standard ABCs of resuscitation.

Establish an open airway by cleaning the mouth, being careful not to induce gagging or vomiting. Position the patient flat to maximize perfusion of the brain in case systemic hypotension occurs or is present. Keep the head in the midline to avoid jugular compression and prevent further spinal injury in the event of existing spinal instability.

Ensure that adequate ventilation is occurring and if not instigate artificial ventilation by bag and mask or by endotracheal intubation. In children without facial injuries, bag and mask ventilation can be accomplished without extension of the neck, 'the sniffing position'. Pressure on the cricoid cartilage will avoid distending the stomach with air. The open airway and adequate ventilation will correct hypercarbia and the addition of supplemental oxygen will avoid hypoxia. Children who arrive at hospital hypoxic have a worse outcome than those who are normoxic or hyperoxic

Finally, paving attention to circulation to reverse systemic hypotension, if present, is the third of the ABCs. Insertion of an intravenous or intraosseous line may be required at the site. Isotonic or hypertonic fluid should alwavs be used for resuscitation, and never hvpotonic fluid such as dextrose and water or one-quarter strength saline. In children, glucose-containing solutions are not necessary since the sympathetic response to the injury usually induces hyperglycemia. In infants, glucose free solutions are used initially and blood glucose levels are checked as soon as possible. If the level is low, glucose is given.

Associated spinal injury is rare in childhood 1.5 per 100000 compared with 350 per 100000 for head injuries) and, while careful movement of the patient as a log is important, spine boards are not made for small children and infants and are better avoided in these under 2 years of age. At this age the large calvarium is forced into flexion by the spine board and this can compromise both the airway and the spine.

Following resuscitation, the child should be transferred to a center capable of the care of children. 

EMERGENCY ROOM CARE

A complete physical examination of the unclothed child is carried out. For young children appropriate overhead warming lights should be available to prevent a major drop in temperature, which can occur very easily in the child because of the relatively large surface area to weight ratio. Any tendency to lose heat will be aggravated by muscle paralysis and sedation, which prevent shivering and thus increase the rate of onset of hypothermia, and are present in the intubated child. Evidence of multiple trauma to chest, abdomen, or skeleton is documented. An immediate set of vital signs to document pulse rate and blood pressure is recorded. A large venous access line, Foley catheter if there is no evidence of perineal injury, and an arterial line are placed and blood is sent for stat studies of CBC, type and cross-match, Prothrombin time (PT) and partial thromboplastin time (PTT), electrolytes, and amylase levels. If an endotracheal tube is not in place, it is now placed in children with a GCS of less than 8 or those in shock. This is best done with bag and mask hyperventilation, nondepolarizing muscle paralysis, and, unless severe hypotension is present, pentothal or a similar drug to prevent the increase in ICP that occurs with intubation and to avoid vocal cord spasm. Once the endotracheal tube is in place, the aim is for a Fao₂ of over 100 mmHg and a Paco₂ of around 30 mmHg. An immediate chest radiograph is taken to document adequate position of the endotracheal tube and to evaluate any intrathoracic pathology.

Once the child is stable appropriate radiologic studies are obtained. This will involve noncontrast CT of the head and usually upper spine to C3, if a spiral scanner is available. Plain cervical spine radiographs are usually obtained, although it must be remembered that up to 60% of spine injuries in children may be SCIWORA (spinal cord injury without bony abnormalities). Usually CT of the abdomen is done if there is any suspicion of abdominal trauma. Other radiographs will depend on the history and the examination.

The period where the child leaves the emergency room to go to the radiology department is often one of least monitoring and is a time of high risk for secondary injury. Thus the use of portable monitoring that records (pulse oximetry, blood pressure, pulse rate, and ICP) is valuable. In children with a GCS of 8 or less and/or with hypotension it is easy to insert an ICP monitor in the emergency room at the time of A-line and venous line insertion. This gives an accurate measure of the effects of fluid resuscitation, position, and transport on the ICP and allows for logical therapy to maintain it below 15 mmHg. Most of the early mortality from head injury is the result of intracranial hypertension and it is easier to prevent the onset of intracranial hypertension than to try to treat it once it has occurred. Even in the event of severe intracranial hypertension, if the ICP can be controlled, recovery can be good.

INTENSIVE CARE UNIT

Correction of any residual systemic abnormality is the first step: acidosis, hypotension, hypoxia, and hypercarbia. Abnormal clotting studies are common in children and should be corrected as soon as possible.67 Hvponatremia, possibly related to overproduction of atrial naturetic factor or salt wasting, is common in severely injured children, although it may be delayed for 24 h or more, and fluid balance and electrolytes must be followed carefully to identify early changes in serum sodium levels that can then be corrected, thus avoiding severe hyponatremia and brain swelling. The drop in serum sodium concentration can be dramatic and abrupt, and may precipitate severely raised ICP

Additional monitoring will often include the insertion of an internal jugular bulb catheter. This can usually be performed quite satisfactorily with the child flat in bed and avoiding the head-lowered position, which can increase ICP. Without a CBF measurement, the AJDo2 does not give a quantitative measure of metabolism but does provide a relative measure of the match between flow and metabolism. This baseline value can then be used as an indicator of relative change in this match, and therefore aid in the selection of appropriate therapy for management of raised ICP. This makes the assumption that global CMRo2 is not changing; this may or may not be true. Repeated CBF measurements are difficult and thus there is increasing use of Doppler ultrasonography to determine flow velocity.

The means available for attempting to modify the secondary injuries have changed little over the past 15 years. Various therapies have come in and out of fashion and the cascade in which the therapies are applied has changed. The major therapies are:

1. hyperventilation;

2. patient position;

3. CSF drainage;

4. osmotic diuretics;

5. metabolic suppressants;

6. blood pressure increase;

7. hypothermia.

The new head injury guidelines for adults have generated certain controversies, particularly as to how they should or can be applied to children, and these will be discussed.  

MECHANICAL VENTILATION

Most patients with a GCS of 8 or more will receive mechanical ventilation to enable control of oxygenation and arterial carbon dioxide. The use of muscle paralysis varies from unit to unit but is a necessary adjunct to controlled respiration, in the authors' opinion, to prevent episodes of raised ICP that can result from the patient fighting the ventilator, coughing with or without suctioning; straining when episodes of decorticate or decerebrate posturing occur or with movement of the patient. The major reason given for avoiding muscle paralysis is that this results in loss of the clinical examination. If the clinical examination is considered to be important, short-acting agents can be used and reversed as often as the observer wishes. However, in the comatose child it is difficult to see what information is obtained from the clinical examination that cannot be obtained in some other way. If the ICP is not monitored then the clinical examination may be the only way to identify a change in the patient's neurological status, although such changes may be quite delayed dependent on the status of pressure autoregulation and the patency of the subarachnoid spaces. Very high ICP may be tolerated before a sudden drop in CBF occurs or hemiation, and thus the clinical examination is not a very sensitive measure of the events occurring in the intracranial space once coma is present. The events that can be treated can all be identified early by adequate monitoring of ICP, oxygen saturation, end tidal carbon dioxide, blood pressure, SAP, and heart rate, and additional information can be gained from the use of jugular bulb oxygen monitoring, CBF studies, or transcranial Doppler ultrasonography. The treatable events are accumulation of an intracranial hematoma, increased ICP not due to a surgical mass lesion, vasospasm, hyperemia, seizures, and altered systemic parameters. Currently, deterioration of function due to molecular events cannot be treated or indeed identified. Vasospasm can produce increased ICP, decreased ICP, decreased CBF, increased velocity on Doppler ultrasonography, or decreased jugular bulb oxygen. If evoked potential monitoring is done, this may also show a change.

Clinically, once the patient is comatose the most likely identifiable change is a deterioration in the motor findings, such as decorticate or decerebrate posturing, change in pupil function (this would not be masked by muscle paralysis), or changes in ventilation. Such changes become harder to identify the lower on the GCS the patient is. Episodes of increased ICP are not necessarily associated with alterations in the clinical findings or systemic parameters, and if the ICP is not monitored and only the clinical examination is used ICP pressure waves can be overlooked. It has been shown in animal models that pressure autoregulation can be modified or abolished by simply raising and lowering the ICP. Episodic increases in ICP can contribute to herniation. Thus, in our opinion, the avoidance of multiple episodes of intracranial hypertension is an important part of ICU care and this end is best attained by using muscle paralysis in parallel with controlled ventilation. Obviously all ventilators used must have automatic alarms to prevent inadvertent and unidentified disconnection.

The ventilator settings must be appropriate to maximize venous return time and avoid excess intrathoracic pressure or overdistension of the lungs that could result in pneumothorax. In patients with pulmonary contusion or adult respiratory distress syndrome (ARDS), where high levels of positive and expiratory pressure (PEEP) may be necessary, muscle paralysis has to be used.  

HYPERVENTILATION

Other than raising the head of the bed, there is no faster way to lower the ICP than by hyperventilation. The mechanism of action is presumed to be arteriolar constriction, decreased CBF associated decreased cerebral blood volume (CBV), and thus reduction in ICP by affecting the vascularcomponent. Despite a decreased CBF, the brain can still extract adequate oxygen and glucose; only when the CBF falls below 25-30% of normal is there any evidence of substrate delivery compromise, and this is in the face of normal metabolism. When the metabolic rate is lowered, lower CBF can be tolerated. Hyperventilation can produce systemic alkalosis that results in a shift of the oxyhemoglobin dissociation curve to the left, making release of oxygen at the tissue interface less efficient. Therefore, an adequate hematocrit between 35 and 40 is advisable to ensure adequate oxygen delivery to the neurons. A normal carbon dioxide response is 3.5% alteration in CBF per mmHg change in Pa co₂.,. A recent study suggests that this response is preserved, though slightly blunted, or increased after head injury in children, except in the most severe injuries where death is impending. If, indeed, damaged tissue responds less to changes in carbon dioxide than normal tissue, blood can in fact be shunted from normal brain to the abnormal brain, but this has not been reported in head injury.

The theoretic risks or adverse effects of hyperventilation are (1) too great a decrease in CBF resulting in ischemia and (2) the fact that the decreases in CBF are transient, with the effect lasting only a short time. While the first concern is a possibility, the most proximal control of CBF to the tissues is local chemical autoregulation, the ability of brain tissue to regulate CBF dependent on local tissue needs. The chemicals involved have not been clearly defined but potential candidates are potassium, calcium, adenosine, and nitric oxide. This control appears to be supreme since local changes in CBF occur all the time in response to changes in metabolic activity without affecting global CBF. Thus, if CBF is being lowered excessively by hyperventilation, local tissue factors will be released and produce local vasodilatation and avoid ischemia. Hypercarbia has been shown to result in an increase in CMRo₂. The reverse, a decrease in CMRo₂, may occur with hyperventilation, permitting an even greater decrease in CBF before ischemia.

While it has been shown in normal volunteers that continued hyperventilation results in a gradual return of CBF to normal levels over 4-6 h, the evidence is that after head injury the effects of hyperventilation are prolonged and the CBF actually becomes lower with time over the first few days.76'77 This is probably because CMR is decreased and therefore the flow required to match metabolism is decreased such that local chemical autoregulation has no stimulus to release vasodilating molecules. If there is an increased local requirement for blood flow as a result of increased local metabolism, then that vascular bed should be able to respond to the local tissue changes and increase CBF to supply the demand in that area despite systemic hvpocarbia.

It is not clear in any individual child whether cerebral metabolism is increased, decreased, or unchanged after severe head injury, and indeed the response may vary with the type and severity of the injury.  The reports that are available show varied responses. If metabolism were raised and CBF decreased, it is possible that hyperventilation could precipitate brain ischemia if local chemical control was disturbed. In reports of oligemia after hyperventilation, the ICP has been high, the CBF low, and the initial CBF very low, but with associated low CMR0, and generally a low AJD02. Children who develop very low flows appear to have devastating injuries. It is not at all clear that hyperventilation can produce ischemia under any circumstance.  It seems likely that there are different responses in different areas of the brain, as has been shown in animal models with both hypermetabolism and hypometabolism.

The potential benefits of hyperventilation are several. The decreased CBF will decrease CBV and therefore ICP. If there is hyperemia, the increased cerebrovascular resistance will decrease end capillary pressure and vasogenic edema formation. In addition, if there is hyperemia, the decreased flow may avoid tissue hyperoxia and limit free radical formation. The alkalosis may have a beneficial effect on tissue acidosis and reuptake of cytotoxic neurotransmitters. Hyperventilation may help to re-establish pressure autoregulation.  The exact match between CBF and CMR in children with head trauma is not well established but in most studies the AVDo2 is low or normal, suggesting adequate or excess global flow for the amount of global metabolism. From all of this lack of information, it is hard to be dogmatic. There is no good evidence that moderate or even severe hyperventilation is detrimental, and indeed most recent reports of outcome after head injuries in children have been from units in which some degree of hyperventilation, often to levels of 20 mmHg, are used when no other way to control ICP is working. In follow-up studies of MM after severe head injury, there is no evidence of focal ischemic lesions watershed infarcts) or global ischemia, and in these studies all the children from one institution were hvperventilated as the initial therapy, and the degree of hyperventilation was increased if ICP was hard to control. What is established is that if children develop uncontrolled intracranial hypertension they will die, and therefore it seems inappropriate to withhold a known valuable therapy because of unsubstantiated potential side-effects. Hyperventilation to around 30 mmHg is still an appropriate initial response in the comatose head-injured child with increasing degrees of hyperventilation if the ICP is high. In cases where increased hyperventilation is required, the child will be receiving metabolic depressants as well, usually barbiturates or sedatives such as Versed. One of the values of inserting the ICP monitor in the emergency room is that the presence of intracranial hypertension can be diagnosed immediately. If the ICP is 15 mmHg or less, a Paco2 between 30 and 35 mmHg is adequate. If the ICP is above 15 mmHg, the hyperventilation can be set to achieve the highest Paco2 that will lower ICP to 15-20 mmHg.

In adults the CBF appears to plateau at a Paco2 of 20 mmHg,  with further reduction in Paco2, having no effect on the CBF, presumably because of local metabolic autoregulation. In children the plateau level of Paco7 has not been established and we have seen decreases in CBF as Paco2 is lowered from 18 to 15 mmHg.78 As the ventilatory rate and volume are increased, adequate time for central venous return is required to avoid venous congestion from raised intrathoracic pressure and systemic hypotension as a result of decreased venous return. The lowest level of Paco2 that is effective has not been defined, but if the ICP remains above 30 mmHg after metabolic suppression and hypothermia the Paco2 can be lowered as much as necessary if this is the only way to control the ICP; we have used a Paco2 of 15 mmHg for 24 h or longer, with survival and recovery. This is effective because of the reduced CMR requiring a much reduced CBF for substrate delivery.  

PATIENT POSITION

Concern has been raised about the value of the head-up position in patients with increased ICP after head injury. The main theoretic concern has been that the arterial pressure generated at the intracranial arterioles may be reduced by a significant amount if the head is elevated and therefore the perfusion pressure may be lowered, with a potential decrease in CBF. The arterial pressure is usually zeroed at the level of the heart, while the ICP is zeroed at the level of the external auditory meatus. If the patient is lying in the supine position, flat in bed, then the CPP (mean arterial pressure (MAP) - ICP) has the zero point at the same level for each variable. If the head is elevated, then the calculated CPP can be inaccurate because of the discrepancy in the position of the zero reference. With the head elevated, the distance above the heart of the external meatus will vary depending on the size of the individual and the degree of head elevation. In a 170-cm individual in the full upright position, the distance from the heart to the external meatus is 30 cm; thus the measured pressure in the head would be different by 22 mmHg if the zero point were taken at the heart rather than at the meatus. In a 4-year-old child this difference would be greater than 15 mmHg, assuming a height of 100 cm. In the head-up 300 position, the difference might be 5 mmHg. Thus the calculated CPP could be off by that same amount, 5 mmHg. In a larger child or adult that value could reach 6 or 7 mmHg, one-third of the distance from the fully erect (90°) position. Raising the head will also tend artificially to increase the arterial measurement by a small amount. While of theoretic concern, papers actually measuring CBF in patients before and after head elevation show that the decrease in ICP produced by this maneuver offsets any decrease in SAP, resulting in a stable or increased CBF, and that CBF was not affected.81'82 Thus there is no reason not to use the head-up position as long as the patient has a :mal or above normal SAP for age. This problem of transducer zero can be corrected by zeroing both transducers to the same level, either the heart or the external meatus.  

CEREBRAL PERFUSION PRESSURE

In adult head-injured patients there is evidence from transcranial Doppler studies that changes in velocity of flow measured by Doppler ultrasonography can occur at a CPP below 70 mmHg, and therefore current recommendations are that CPP be maintained above 70 mmHg on the assumption that these changes in velocity are indicative of decreasing CBF. In children, the average CPP varies with age both because the ICP and the arterial pressure change with age. The ICP is probably less than 10 mmHg in a child with an open fontanelle, and rises to around 15 mmHg as the fontanelle closes. Likewise, normal MAP in a newborn is around 50 mmHg and rises to around 90 mmHg in the mid-teens. Thus the normal CPP in a 6-month-old may be 40-50 mmHg, and in a 14-year-old 70 mmHg. There is no ideal CPP that is supported by experimental data in children and therefore therapy is aimed at trying to maintain an ICP within 5-10 mmHg of normal, 15-25 mmHg and a MAP that is equal to or higher than the normal mean for age. If there is hypotension, immediate correction with blood, isotonic or hypertonic saline, and or vasopressors is required. If the ICP cannot be lowered and the CPP is dropping, then vasopressors may be carefully added to increase the MAP. This should be done with an ICP monitor in place to be sure that the rise in MAP is indeed accompanied by a rise in CPP. If the brain is tight or pressure autoregulation defective, the ICP may rise with the MAP such that there is no net gain and the rise in ICP may precipitate brain shift and herniation or result in a global decrease in CBF. The proponents of CPP as the prime parameter have failed to demonstrate that a CPP of 80 mmHg is always associated with the same CBF. Older studies suggested that this was not so and that CBF was better preserved when the CPP was maintained by a low ICP than by a high SAP.55~71'85 The use of jugular bulb monitoring of oxygen saturation and AJD02 can be helpful in trying to estimate the match between CBF and CMRo2 before therapy and for comparing the effects after therapy. If, after raising MAP, there is a decrease in AJD02, this is supportive evidence that an increase in cerebral perfusion has occurred, but assumes that no change in metabolism has taken place. There is no sound foundation for selecting a particular CPP in children and it is still advisable to treat ICP and MAP as separate variables that can be manipulated by different mechanisms.  

CONTROL OF RAISED ICP

There is still some disagreement as to whether there is clinical value to the use of ICP monitoring in children following head injury. No studies have positively shown that the use of this monitoring makes a difference to outcome. To change outcome is not the proximal reason for using ICP monitoring, just as measuring blood pressure is not believed to improve the outcome from shock. In both these clinical settings, measurement of the actual pressure, ICP or SAP, is necessary to define adequately the severity of the problem. The purpose of any monitoring technique is to supply information about a system that is not easily or sensitively gained in any other way and in which system there is reason to believe that normal physiologic homeostasis is disturbed such that the normal homeostatic mechanisms are dysfunctional. The information from the monitor is then used to modify therapy and, in essence, to take over the homeostatic role until recovery occurs. There is no purely clinical examination-based way to measure the ICP. The presence of papilledema infers that intermittent increases in ICP are occurring or have occurred, but says nothing about the actual ICP at the time of the clinical examination and the absence of papilledema does not correlate with the absence of raised ICP. Changes in clinical signs may be caused by raised ICP (e.g. pupillary dilatation) but may equally be the result of other alterations in function such as a seizure.

The ICP will be raised above normal in 75% of children with severe head injury at some time during their course in the ICU. The problem is to identify, in each patient, at which point the ICP is raised and therefore when therapy is needed. Additional information that can be obtained only through monitoring is whether the therapy given was effective in lowering the ICP by an appropriate amount. This allows a logical decision to be made as to whether further therapy is needed at that time and to identify when the ICP rises again to levels requiring further therapy. While increased ICP is not the only problem associated with severe head injury, it is associated with the majority of deaths and is a parameter that can be treated. Therefore, ICP monitoring is a recommended part of the current care of severely head-injured children. Any child who is receiving therapy to affect the ICP can be more accurately and appropriately treated when the ICP is being measured, and this is recommended in children with a GCS of 8 or less. The current most frequently utilized monitoring techniques are intraventricular via a ventriculostomv catheter or intraparenchymal using either fiberoptic or solid-state technology. The former has the advantage of allowing re-zeroing to be done, thus minimizing any artifact from inherent drift within the measurement system and is the 'gold standard' for measurement. The use of a ventricular catheter also allows withdrawal of CSF as a therapeutic modality. The use of subdural or epidural monitoring is sometimes done following surgery but the values are less certain and the pitfalls greater. The reason for monitoring the ICP is to identify increases that occur and to treat them, thus returning the ICP a more acceptable or normal level. The ideal ICP is not known but if 15 mmHg is taken as normal, the plan is to keep the ICP below 20 mmHg if possible. There are a number of therapeutic options for lowering raised ICP.  

CSF DRAINAGE

In approximately 50% of children with severe head injury the initial CT scan shows evidence of brain swelling with small ventricles and subarachnoid spaces. In these children the value of inserting a ventricular catheter to use both as an ICP monitor and as a therapeutic modality to drain CSF has to be weighed against the possible complications. These include inability to cannulate the ventricle, the production of edema or hemorrhage as a result of multiple passes of the catheter, and ventriculitis. The latter can be minimized by tunneling the catheter subcutaneously as far as possible away from the drill-hole into the cranium. In addition, if the ventricle is small there may be a dampened pressure trace that may give unreliable values; if CSF is drained, this may dampen or abolish the ICP pressure recording. Finally in this setting there is often little or no CSF to drain, making this a not very helpful therapy because of collapse of the ventricle around the catheter.

In children with normal or larger ventricles, or in whom one ventricle is enlarged because of focal mass, then CSF drainage may be the only additional therapy needed to control the ICP. In many children the ICP may be easily controlled the first few days, then start to rise and be difficult to control at a time when the child's neurologic status may be starting to improve. Repeat CT at this time frequently shows resolution of the swelling with normal or distended CSF spaces. With this CT scan the inferred pathology is increased outflow resistance secondary to subarachnoid hemorrhage. At this stage after the injury, ventricular or lumbar CSF drainage can produce rapid resolution of the problem and is always worth trying rather than more intensive therapy.  

OSMOTIC DIURETICS

The deliver of osmotic diuretics to the brain can be an effective way to lower the ICP. It has been assumed that the mode of action is withdrawal of extracellular water from the normal brain, resulting in a decreased brain volume and lowered ICP. The exact action is not clear and may even be withdrawal of CSF from the intracranial space. Rosner and Coley have suggested that the effects are bound to the presence of intact pressure autoregulation. Early studies showed effects from mannitol in the absence of autoregulanon and therefore this mechanism may or may not be functional. The most frequently used agent is mannitol given in doses of 0.25-1 g per kg intravenously. Mannitol is excreted by the kidneys and is not metabolized in the body to any significant degree, and thus tends to remove free water from the body in the urine. Thus dehydration can easily occur, resulting in alteration in viscosity of the blood and hypotension if adequate fluid replacement is not maintained. The mannitol must be delivered over 10-15 mm to prevent alteration of SAP. In children who are sedated and paralyzed, the rapid infusion of mannitol will raise the blood pressure and can result in an early increase in ICP before the ICP is reduced. This is probably due to an increase in intravascular volume related to the too-rapid infusion. The MAP rises, which can increase ICP by increasing CBF as a result of increased CP?, increasing brain turgor as a result of the increased MAP, and producing direct cerebral vasodilatation as a result of the osmotic load. Mannitol can also produce a decrease in MAP by having a direct vasodilating effect on the arterioles as a result of the increased osmotic load. This shrinks the vascular pacemaker cells and can result in fewer muscle cell contacts, with resultant vasodilatation. In general the use of mannitol should be discontinued if the serum osmolality exceeds 330 mosmol 1~ because of the risk of deposition of mannitol crystal in the renal tubules and resultant tubular necrosis. The use of mannitol is often reserved for the later, post 48 h, treatment of head trauma since there is little evidence of brain edema in the early stages and edema is the pathology for which mannitol seems most indicated. Mannitol is usually given by repeated bolus injection to boost the serum osmolality intermittently. However, it can be given as a bolus followed by a continuous infusion to try to maintain a specific serum osmolality. There is little rationale for a fixed-bolus schedule for mannitol, since the effect of a given dose may vary markedly both in the ICP decrease achieved and in the length of time over which the pressure stays down. The frequency of administration and the dosage need to be individualized for each child.

Urea is another osmotic diuretic that is still used occasionally. The final agent is glycerol. This can be given via a nasogastric tube or specially prepared for intravenous use with a special filter. Glycerol is metabolized by the liver and probably the brain, and therefore has a less dehydrating effect as its renal excretion is low. It does not produce tubular necrosis and before the serum osmolality can be increased to higher levels. The biggest complication with glycerol is that, as a result of liver metabolism, the serum free fatty acid levels can get verv high and produce pulmonary capillary obstruction. Thus serum free fatty acid levels need to be monitored carefully. The usual dosage schedule is 0.5 g kg-' as a bolus and then 0.5 g kg-' as an infusion over 30-60 mm. Like mannitol, it is repeated as needed, dependent on the ICP.  

HYPOTHERMIA

Children have a higher relative surface area to weight ratio than adults and, when paralyzed and sedated for controlled ventilation, tend to lose heat easily such that the core temperature in the ICU is often below normal. It is not necessary to reverse that as there may be value in the lowered core temperature. For each 1°C change in temperature there is an approximately 10% change in cerebral metabolism. Thus it is desirable to prevent hyperthermia since this will result in increased substrate demand in the brain, requiring increased CBF. The ability of the CBF to increase may be dampened or prevented by increased ICP, decreased CPP, and maximum existing vasodilatation such that a further increase in CBF is not possible and cerebral ischemia results in further injury. Alternatively, if vasodilatation does occur as a result of the increased temperature, this will increase the cerebral blood volume and may result in a hard-to-control increase in ICP.

Deliberate efforts to lower the core temperature to 32-33°C can be helpful in controlling otherwise impossible-to-control ICP. The decreased temperature lowers metabolism, lowers CBF and cerebral blood volume, and therefore ICP. There also may be some protective effect when waves of increased ICP occur since the lowered metabolic rate will permit the brain to tolerate greater and longer episodes of lowered CBF. If moderate hypothermia is to be used, the patient needs to be hydrated adequately, and it is better to use this modality early in the course of therapy if it is clear that high and difficult-to-control ICP is a problem. Hypothermia may reduce the requirements for other therapies.

The potential danger of this level of hypothermia is interference with phagocyte activity and pulmonary ciliarv function, with resultant pneumonia. This has been reported in Reye's syndrome but has not been a problem in the treatment of children with head injurv. However, in children with severe pulmonary contusion it may be safer to avoid hypothermia of this degree as they have an increased risk of pneumonia.  

METABOLIC DEPRESSANTS

Most children who are being treated for severe head injury by controlled ventilation are on some medication to minimize pain, usually fentanvl, and often a soporific such as Versed to allay any anxiety since we have no way to measure this in the comatose child. The use of muscle relaxants varies dependent on the treating physician. There is no question that, despite sedation, the children can fight the ventilator, resulting in increases in ICP, and muscle relaxants are generally used in addition to other medications. Neither fentanyl nor Versed has been shown to have any ICP-lowering effect and, if a sedative is required, it may be more appropriate to use a barbiturate or other soporific with known ability to lower raised ICP. In children in whom the ICP is high and difficult to maintain below 20 mmHg despite multiple medications, the use of pentobarbital at an early stage as the metabolic suppressant, anti-anxiety drug plus anti-seizure drug seems to make more sense than to use a multiplicity of drugs. The side-effects of increased vascular resistance and hypertension usually occur when such drugs are used after mannitol and in a dehydrated child and therefore, when the concerns for prolonged raised ICP are present or when early therapy such as position and hyperventilation is ineffective in lowering the ICP, barbiturates have a role. They have been shown to be capable of lowering the ICP in this setting.

ANTIEPILEPTICS 

Seizures have been reported in 30% of severely head-injured children. Many of these are single seizures and often occur around the time of the impact. Seizures occurring in the first hour are often isolated events and do not require medication. Therapy with antiepileptic drugs, usually phenytoin or now phosphenvtoin, is necessary only for repeated or prolonged seizures. The beneficial value of prophylactic antiepileptic medication has not been demonstrated in pediatric head injury. There are many pediatric head injury units that do use prophylactic anticonvulsants, and at the present time this remains the choice of the treating physicians.  

OUTCOME

The factors that are useful in predicting final outcome after severe head injury have been enumerated in a number of studies. Very young children seem to do worse than toddlers and older children, and it is likely that this is related more to the cause and therefore the pathophysiologv of the injury than to the age of the child. Some 10-30% of infant trauma under 2 years of age, at least in the USA, may be due to non-accidental or inflicted trauma. Suffice it to say that the frequency of delayed medical attention, hypoxia, and ischemia are very high and seem to be the major reason for the poor outcome.

Predictors of poor outcome for accidental injury include multiple trauma and early hypoxia or shock. Other factors include the GCS, especially the motor score which is the easiest part of the GCS to apply to children of all ages, pupillary function, and high Injury Severity Score, CT pattern, and ICP. Mortality rates in children remain high: 40-60% for children with a GCS score, of 3 or 4. The mortality rate for children with GCS of 3 remains around 70%, whereas for a GCS of 4 the rate drops to 10-20%, and that in children with a GCS of 5 or above is less than 10%, and zero in some reports.

Morbidity following pediatric trauma varies depending on the measures used to evaluate recovery: the more sensitive the measure the greater percentage of children with an identified deficit. If the Glasgow outcome scale is used to evaluate survivors, the numbers look very good with only a small percentage of children in the vegetative state, approximately 25% severely disabled, and 75% making a good recovery or being moderately disabled. Such a limited measure does not reflect the true situation. Following a severe injury, even if no head injury occurred, 50% of children will exhibit functional limitations at 6 months. In addition there are emotional and social effects on the family as a whole, with up to 40% of families experiencing a change in the family unit within the first year and almost 50% of siblings who were not injured demonstrating social, emotional, or educational difficulties. Thus, even in the absence of specific brain injury, the adjustment of the injured child and the family is prolonged and difficult. Counseling and support are required and should be an automatic part of any trauma system, otherwise the enormous expenditure of money and effort to treat the trauma in the acute stages is undermined and the goal of returning the child to a normal integrated social and educational system is not met.

Specific problems related to cerebral injury also occur but seem to be predominant in children with severe trauma. Recent findings of frequent damage to the frontal and temporal lobes may account for the neurocognitive and some of the behavioral sequelae. Following severe traumatic brain injury, frontal lobe tasks such as the Tower of London test are affected, even in children with normal MRI scans. These children do have physiologic problems such as decreased CBF in the frontal lobes. Volumetric studies have shown that the tissue loss appears to be predominantly gray matter of the frontal lobe. Thus the common complaint from the family that the child is not the same as before is probably true. With increased information about frontal circuits that may play a role in stress, depression, and possibly other higher functions, subtle changes in performance and behavior may well have a neurochemicoanatomic basis.

The effects of head injury on psychiatric disturbance has been another area of question. An increased occurrence of new psychiatric problems occurs even after moderate or minor injuries, with more reliable increases after severe injury. Whether these too will turn out to be the result of interference in the frontal-thalamic-brainstem circuits remains to be seen. Similarly, problems with memory, concentration, behavior, and impulse control all occur. The problem of neurotoxin transmitter release and other biochemical alterations may be the pathology that underlies these subtle problems, which are not easily correlated with MM abnormalities. If this is the case then the role of chemical mediators may not be measurable by looking at mortality data but only by measures of cognitive recovery.

It is clear that severe head injury almost always results in some measurable alteration in cerebral function: neurologic, cognitive, psychiatric, or social. Whether the pathology that is responsible for severe disability, the vegetative state, and death is the same as that responsible for the more subtle changes is not clear, but current evidence suggests that there are differences. The major two are the lack of evidence of secondary ischemic injury and the relatively low incidence of diffuse axonal injury that is reported in follow-up MM scans. The pathophysiology of death from head injury may be different from that associated with recovery, not because there is simply less intensity of injury but because different pathologic, processes are occurring. If this is true, it suggests that best therapy should not be based on information obtained from the study of mortality but from studies of recovery.

CONCLUSION

Severe accidental head injury in children carries a low mortality rate of approximately 20%. The outcome is dependent on:

1. severity of injury as measured by the GCS;

2. associated shock;

3. hypoxia in the emergency room;

4. initial CT scan;

5. pupil reflexes.

These factors, in combination, have the best predictive value for mortality. Recovery, if it occurs, can be equally good with any GCS. The pathology is variable over the time course of the injury and includes primary and secondary factors that can influence the outcome. Early pathology is hypoxia, ischemia, shock, and intracranial operative mass lesions. Later, brain swelling and raised ICP occur. This is then followed by brain edema, secondary ischemic swelling, increased outflow resistance to CSF, traumatic aneurysms and, later, hydrocephalus and epilepsy. The best therapy then changes over time and a clear understanding of the changing pathophysiology is important. The cause of death appears to be severe brain injury with raised ICP and usually occurs in the first 48 h. The pathology in children who die or are in a vegetative state is DAI, brain swelling, and secondary ischemic injury. In children who survive it seems that the pathology is less commonly DAI, there is little evidence of ischemia, and focal contusions of the frontal and temporal are the most frequent findings. In addition, loss of cortical gray matter volume in the frontal lobes is common.

In children who do not die, a high incidence of subtle or blatant changes in cognition, social function, and behavior is seen, and long-term support is required to maximize the recovery process. Efforts to prevent injury and support systems to maximize social recovery are essential components of any trauma system.