Biomechanical Factors

Clearly, the physics of TBI in a child is different than in an adult. The relatively large head size, reduced muscular strength in the neck, and increased flexibility of the neck may promote a greater range of biomechanical force imparted to the head. On the other hand, there is less CSF space around the brain, and the prominent bony ridges of the anterior and middle cranial fossae are less developed. These factors may contribute to a lower occurrence of focal lesions in pediatric TBI [Berney et al., 1994; Bruce et al., 1979; Levi et al., 1991; Luerssen et al., 1988].

The physical properties of the developing skull also confer advantages and disadvantages. It is thinner, and thus more prone to diffuse deformation [Margulies and Thibault, 2000]. Some authors have reported a higher risk of skull fracture in children [Berney et al., 1994], although this finding is not universal [Levin et al., 1992]. However, the open fontanels and flexible sutures may serve to dampen traumatic forces, as well as to accommodate a greater degree of intracranial swelling. The brain itself, particularly in infants and young children, has a higher water content and is incompletely myelinated [Holland et al., 1986; Paus et al., 2001]. The higher water content tends to make the brain less compressible and less compliant than in the adult. Thus, the physical properties of the immature brain may provide both unique benefits and potential vulnerabilities to TBI, a theme that will be repeated when considering other developmental distinctions.

Changes in Cerebral Metabolism

It is well known that cerebral metabolism changes during maturation. The primary source of energy for the brain changes from lactate in the perinatal period, to ketone bodies while nursing, to glucose after weaning, and on through to adulthood [Nehlig, 2004; Vannucci and Vannucci, 2000]. The ability to use alternative fuels may have important consequences for acute energy metabolism in the injured brain at different ages [Prins et al., 2004].

Even after the brain has switched to predominantly glucose metabolism, the levels of this metabolic activity continue to change throughout the developing years. Basal glucose metabolism rates peak at around 6 years of age, correlating with the time of maximal synaptogenesis. As synaptic pruning and dendritic rearrangement occur throughout childhood and adolescence, cerebral glucose metabolism declines, although it still remains elevated compared to the adult brain [Chugani and Phelps, 1986]. Also, the brain does not mature uniformly, and cerebral metabolic rates for glucose and other substances have different time courses regionally.

Distinct Neurovascular Regulation

Neurovascular control may also differ in the young brain. Early studies suggested a propensity for diffuse cerebral swelling in children following closed-head injury, occurring 2–5 times more often than in adults [Aldrich et al., 1992; Lang et al., 1994]. Recent studies report that normal children have higher baseline cerebral blood flow (CBF) than adults, with significant variation across narrow age groups [Suzuki, 1990; Zwienenberg and Muizelaar, 1999]. At birth, CBF is lower than in adults, but increases rapidly to peak by age 5 years. CBF then appears to decline through adolescence to adult levels [Suzuki, 1990]. The effect of these developmental differences on injury-induced changes in CBF and the mechanism(s) of these differences are not yet well understood. Certainly, molecular mediators of neurovascular tone, such as nitric oxide synthase, have differential expression patterns in the immature brain [Keilhoff et al., 1996; Ohyu and Takashima, 1998], and these may play a role in age-dependent changes in blood flow and response to injury.

Increased Excitatory Neurotransmission

As mentioned earlier, the number of synapses in the human brain appears to peak in early childhood, and is associated with a peak in cerebral glucose metabolism. Levels of excitatory transmitter receptors are higher in the immature brain [Fosse et al., 1989; Insel et al., 1990; Miller et al., 1990a]. The occurrence of early post-traumatic seizures is also noticeably higher in children, and particularly in infants and younger children [Annegers et al., 1980; Berney et al., 1994].

Because excessive release of excitatory neurotransmitters can result in neuronal injury and death (excitotoxicity), the fact that the developing brain is more active might suggest that blocking excitatory transmission would be neuroprotective. However, early in development, neuronal activation also plays a critical role in both survival of immature neurons and in proper wiring of cerebral circuitry. In fact, in animal models, use of glutamate antagonists early in postnatal development (while protective against some degrees of excitotoxic insult) resulted in a large degree of programmed cell death (apoptosis) [Bittigau et al., 1999; Ikonomidou et al., 1999; Pohl et al., 1999]. Such evidence suggests that optimal recovery from brain injury may require a proper balance of excitation and inhibition, and may also account for the relative ineffectiveness of glutamate antagonists following human brain injury [Biegon et al., 2004; Hardingham et al., 2002; Ikonomidou and Turski, 2002].

On-Going Cerebral Maturation

In the past, the developing brain was generally felt to recover better after many types of brain injury, including stroke, TBI, and surgical resection. Developmental plasticity does confer some advantage when the brain must recover from a focal lesion [Kennard, 1942; Kolb et al., 2000; Villablanca and Hovda, 1999], if the lesion occurs at a specific age (see chapter 13). However, more recent work clearly indicates that the very young brain may be particularly vulnerable to diffuse injuries like TBI [Anderson et al., 2005; Babikian and Asarnow, 2009]. One mechanism for this vulnerability can be explained by post-traumatic impairment in experience-dependent neuroplasticity, which is the complex task of responding to environmental stimuli and rearranging neuronal network that results in enhanced function. In a mature brain, the consequences of such impairment might be overcome with time; in the young brain, it is likely that reduced responsiveness at a critical window of development will result in long-term dysfunction [Giza et al., 2009]. As a result, studies looking at post-TBI brain development, both in experimental animals and in children, must necessarily take into account the normal trajectory of brain maturation.

Glutamate Release and Ionic Flux

Following a traumatic brain injury, there is an immediate and indiscriminate release of the excitatory neurotransmitter, glutamate [Bullock et al., 1998; Katayama et al., 1990]. This can occur due to widespread triggering of action potentials, synaptic neurotransmitter release, and membrane disruption. This flood of glutamate results in a massive efflux of potassium into the extracellular space [Katayama et al., 1990; Nilsson et al., 1993], and an influx of sodium and calcium [Nilsson et al., 1996; Osteen et al., 2001]. While hyperacute measures of extracellular glutamate are obviously unobtainable in human patients, microdialysis or CSF samples from severely injured patients (including children) in the days after injury have demonstrated glutamate elevations and have been associated with secondary injury and poorer outcomes [Bullock et al., 1998; Ruppel et al., 2001; Vespa et al., 1998].

Dynamic Changes in Cerebral Metabolism

As the injured cells attempt to restore ionic equilibrium post-trauma, membrane pumps such as the Na⁺-K⁺ ATPase are activated. An increase in cerebral glucose uptake, termed hyperglycolysis, is seen early after experimental TBI and has been postulated as a mechanism of providing additional substrate to generate the energy necessary to drive these membrane ionic pumps [Kawamata et al., 1992; Yoshino et al., 1991]. Acute studies of cerebral glucose metabolism using positron emission tomography (PET) in adult patients have also shown evidence for increased cerebral glucose uptake [Bergsneider et al., 1997]. Interestingly, there is increasing laboratory evidence that injured neurons are capable of utilizing alternative fuels to glucose [Magistretti et al., 1999; Pellerin and Magistretti, 2004; Prins et al., 2004], and it is well known that the capacity to use alternative fuels in uninjured neurons is age-specific [Vannucci and Vannucci, 2000].

After the acute period of hyperglycolysis, there is a prolonged period of diminished glucose uptake by the brain. This hypometabolism is seen in both adult and immature rats, although it lasts longer in adults (7–10 days) than in rat pups (3 days) [Yoshino et al., 1991; Thomas et al., 2000]. In traumatically injured adult patients, the period of reduced glucose metabolism has been shown to last weeks or months [Bergsneider et al., 2001].

Cerebral Blood Flow: Hyperemia? Hypoperfusion?

Blood flow also undergoes dynamic changes after injury and the pattern these changes take may depend on the type of injury and its severity. A pattern of diffuse cerebral swelling has been reported more often in children than adults [Aldrich et al., 1992; Lang et al., 1994]. This phenomenon was originally attributed to hyperemia [Bruce et al., 1981], but subsequent studies suggest that the incidence of post-traumatic hyperemia has been overestimated [Muizelaar et al., 1989]. In the earlier studies of post-TBI CBF in children, flow values were compared with normal values from a young adult control group. When CBF has been measured in normal children, however, it appears that blood flow values change dramatically across development, and are generally much higher in young children than in adults [Chiron et al., 1992; Suzuki, 1990; Zwienenberg and Muizelaar, 1999]. Subsequent studies of CBF in children suffering TBI, when compared to age-appropriate control values, have failed to show clear hyperemia [Adelson et al., 1997a; Sharples et al., 1995b]. In an experimental model of TBI, there is evidence for dramatically increased cerebral blood flow adjacent to a traumatic contusion, and this focal hyperemia was most pronounced in immature animals [Biagas et al., 1996]. While clinically it is important to recognize that diffuse cerebral swelling is more common in children, the mechanism of this swelling does not appear to be solely attributable to hyperemia.

The relation between CBF and outcome appears to depend upon the state of cerebral autoregulation. In the presence of intact autoregulation, increased CBF has been associated with better perfusion and improved outcome in children and young adults [Kelly et al., 1996; Vavilala et al., 2004]. However, after very severe injuries, in which autoregulation is dysfunctional and cerebral vasculature is pressure-passive, increased CBF can result in intractable elevations of ICP and worse outcome.

Beyond the acute phase of head injury, CBF generally declines. Post-TBI CBF can decline below normal, and it is important to monitor for ischemia, as this is associated with worse outcome [Bouma et al., 1991; Vespa et al., 2003]. Investigations are still under way to determine whether neurovascular coupling is intact during this period of reduced flow.

Altered Neurotransmission

Reductions in excitatory neurotransmitter systems have been reported after TBI in experimental animals, including impairments in glutamatergic [Giza et al., 2006; Miller et al., 1990b; Osteen et al., 2004; Sihver et al., 2001], noradrenergic [Fujinaka et al., 2003; Krobert et al., 1994], and cholinergic transmission [Gorman et al., 1996]. These changes may correlate with the injury-induced reduction in cerebral metabolism described above. Furthermore, these changes may be an underlying mechanism for post-TBI deficits in neuronal activation [Dietrich et al., 1994; Ip et al., 2003; Sanders et al., 2000], diminished activity-dependent molecular responsiveness [Giza et al., 2002; Griesbach et al., 2004], and impaired experience-dependent plasticity in developing animals [Fineman et al., 2000; Giza et al., 2005; Ip et al., 2002]. Activation studies using functional magnetic resonance imaging (fMRI), even after relatively mild TBI, have shown reduced or aberrant activation patterns in traumatically injured teenagers and young adults [Jantzen et al., 2004; McAllister et al., 1999].

Axonal Disconnection

Mechanical stretch can disrupt axonal membranes, resulting in calcium influx and microtubule and neurofilament disruption. Damage to these important cytoskeletal components impairs normal axonal transport, endangering distal axonal segments and synapses. Over time, disrupted axonal transport leads to an accumulation of transported proteins and organelles at the injury site, causing axonal blebs and, eventually, disconnection [Pettus and Povlishock, 1996; Povlishock and Christman, 1995; Povlishock and Pettus, 1996]. This type of damage is seen following experimental TBI in both immature [Raghupathi and Margulies, 2002] and adult animals [Povlishock and Pettus, 1996]. In postmortem human specimens, evidence of axonal blebbing and damage has been reported long after injury [Maxwell et al., 1997], and MRI can clearly show abnormalities of diffuse axonal injury, in pediatric and adult TBI patients [Tong et al., 2004]. Newer imaging modalities, such as diffusion tensor imaging (DTI) and MR spectroscopy, have shown white-matter abnormalities that correlate with functional impairments after pediatric TBI, even in brain regions without overt structural lesions [Ashwal et al., 2006; Babikian et al., 2009a; Ewing-Cobbs et al., 2008; Wilde et al., 2008].

Cell Death: Necrosis and Apoptosis

Mechanisms of post-TBI cell death occur across a spectrum from necrosis to apoptosis. Necrosis represents acute energy failure, inability to maintain ionic homeostasis, cell swelling, and eventual disruption of the plasma membrane. This type of cell death occurs when energy is depleted and does not require protein synthesis. Necrosis is seen as a direct result of traumatic injury, generally occurs acutely, and evolves in regions experiencing profound energy failure. When cells rupture, there is release of intracellular contents and a significant destructive inflammatory response.

Apoptosis is characterized by nuclear condensation, DNA fragmentation, and preservation of the cell membrane. Apoptosis requires energy and protein synthesis, and triggers an acute inflammatory response that may be distinct from that seen with necrosis (see chapter 14). Usually, apoptotic cell death takes longer to evolve after injury, and studies have demonstrated that apoptosis can continue to occur long after experimental trauma [Conti et al., 1998; Wilson et al., 2004]. While apoptosis is seen after TBI, it is also part of the brain’s normal maturational program. Thus apoptotic pathways are more active in the immature brain, and apoptotic cell death appears more prominently after experimental injury to the developing brain [Bittigau et al., 1999; Pohl et al., 1999].

Impaired Plasticity

An important aspect of injury in the young brain is the effect of this injury on normal neuronal responsiveness and on developmental plasticity (see chapter 13). As mentioned earlier, increased plasticity in the immature brain results in significantly better recovery than in the adult following a focal lesion, and this has been well described in rats [Kolb and Tomie, 1988], cats [Burgess and Villablanca, 1986; Villablanca and Hovda, 1999], primates [Kennard, 1938], and humans [Hogan et al., 2000; Trauner et al., 1993]. However, the benefits of youth are less apparent when the injury is more diffuse and/or occurs at a critical window of brain development. Environment enrichment (EE) is an experimental model of enhancing brain development, and animals reared in EE grow up to have larger brains [Bennett et al., 1964; Diamond et al., 1964; Rosenzweig and Bennett, 1996], increased dendritic arborization [Faherty et al., 2003; Greenough et al., 1973; Juraska, 1984; Volkmar and Greenough, 1972], and superior performance on neurobehavioral tasks [Tees et al., 1990; Venable et al., 1988; Williams et al., 2001]. While injured rat pups show less overt cell death and less behavioral impairment than adults [Gurkoff et al., 2006; Prins et al., 1996; Prins and Hovda, 2003], they do lose the ability to benefit from EE rearing. Specifically, they show a loss of EE-induced experience-dependent plasticity; cortical thickening is blocked [Fineman et al., 2000], expansion of dendritic arbors is inhibited [Ip et al., 2002], and EE-induced cognitive enhancements are absent [Giza et al., 2005]. These studies of altered developmental plasticity are also supported by the findings of Prins and colleagues using a different model [Prins et al., 2003]. Following a lesion, regrowth of entorhinal cortical axons in a nontraumatically injured brain occurs in a well-characterized fashion. In juvenile rats subjected to experimental TBI, this normal pattern of axonal regrowth was markedly disturbed, with evidence of disruption down to the synaptic level.

Studies of developmental plasticity in traumatically injured children are difficult, as the effects of both age at injury and age at assessment must be considered in the control groups. However, there is growing evidence that pediatric TBI results in severity-dependent altered trajectories of brain development [Babikian and Asarnow, 2009; Catroppa et al., 2008]. First, although children generally have better outcome from TBI than adults, those injured at the earliest ages (i.e., in infancy) actually have worse outcomes [Anderson et al., 2005; Levin et al., 1992; Luerssen et al., 1988]. Some of this may be due to different mechanisms of injury, including nonaccidental trauma, which has a very poor developmental prognosis and is predominantly an etiology seen in infancy. It is also possible that the very young brain is decidedly more vulnerable to injury, as suggested by experimental studies using hypoxia-ischemia [Ikonomidou et al., 1989] and TBI [Bittigau et al., 1999; Pohl et al., 1999]. Second, there are rare anecdotal reports of neonates with head trauma that appear to have resulted in abnormal cortical development that was subsequently confirmed pathologically after epilepsy surgery for intractable seizures [Lombroso, 2000; Marin-Padilla et al., 2002]. Third, studies of severe pediatric TBI show more persistent cognitive deficits when the injury was diffuse rather than focal [Levin et al., 2000]. Lastly, in studies of repeated, mild sports-related TBI sustained during adolescence or early adulthood, subtle but significant cognitive impairments are detected [Collins et al., 1999; Matser et al., 1999].

Epidemiology

It is estimated that over 300,000 sports-related head injuries occur annually in the United States, the majority of them mild [Thurman et al., 1998]. In high school athletes, 5–6 percent of all injuries were head injuries, the majority occurring during football play, but with almost every sport encountering some number of concussions [Guskiewicz et al., 2000; Powell and Barber-Foss, 1999]. Interestingly, after sustaining a concussion, there was a three-fold risk of a subsequent concussion during the same sports season [Guskiewicz et al., 2000]. This may be due to a concussion-induced physiological vulnerability, a manifestation of a particularly risky style of play in certain individuals, or perhaps an indication of subtle coordination and reaction time impairments in concussed individuals, setting them up for repeated injuries when they return to practice or play prior to full recovery. The vast majority of sports concussions (91 percent) do not involve loss of consciousness, and headache is the most common complaint (86 percent). Perhaps because of the mild nature of these injuries, almost one-third of these players returned to play the same day [Guskiewicz et al., 2000]. There is a common mentality among both players and coaches that one can “shake it off” and “take one for the team,” which sets the stage for recurrent injuries.

Pathophysiology

Physiological studies of sports concussion have demonstrated abnormalities in brain metabolism and activation in the absence of anatomical lesions [Bergsneider et al., 2000; McAllister et al., 1999; Vagnozzi et al., 2008] (Figure 74-11). PET scanning of mildly injured young adults has demonstrated dramatic reductions in cerebral glucose metabolism in walking, talking players that resemble abnormalities seen in comatose patients [Bergsneider et al., 2000]. Experimental studies of diffuse concussive injury in immature rats show periods of reduced glucose metabolism lasting for 2–3 days [Thomas et al., 2000].

Fig. 74-11 Reduced cerebral metabolism of glucose (CMRglc) is seen after mild or severe diffuse traumatic brain injury in human patients.

This suggests a commonality in the underlying pathophysiology of different severities of traumatic injury. GCS, Glasgow Coma Scale.

(From Bergsneider M, Hovda DA, Lee SM, et al. Dissociation of cerebral glucose metabolism and level of consciousness during the period of metabolic depression following human traumatic brain injury. J Neurotrauma 2000;17:389.)

More recently, fMRI has been used to study brain activation in mildly injured patients. Even while overall performance on particular cognitive tasks was acceptable, the task-associated activation patterns seen in these patients were decidedly abnormal [McAllister et al., 1999; McAllister et al., 2001]. A general theme of fMRI testing in concussed patients is that task-specific activation patterns become less focused and more diffuse in the injured brain [Jantzen et al., 2004]. This has been interpreted as a compensatory strategy to bypass dysfunctional circuits by utilizing less efficient alternative routes.

Evidence of postconcussive axonal dysfunction also has been reported using DTI, a newer MRI technique that quantifies the direction of diffusion along white-matter tracts. Abnormalities in DTI signal that correlate with clinical symptoms have been reported following even mild TBI in children and adolescents [Wilde et al., 2008; Wozniak et al., 2007].

Symptomatology

Symptomatology in sports concussions is similar to that described above regarding concussions. One consideration with subjective reporting of symptoms in a sports setting is that the player, being motivated to return to play, may attempt to minimize symptoms. Many athletic teams and physicians are utilizing a concussion checklist to semiquantitatively assess symptoms [McCrea et al., 2003]. This can be administered pre-injury and repeated at intervals during the recovery period. There is also increasing use of computerized neuropsychological tests that can be administered by an athletic trainer using a laptop, shortly after injury and again repeatedly during recovery. Interestingly, there is some evidence that recovery from sports concussion is prolonged in high-school students compared to college athletes [Field et al., 2003]. This raises legitimate concerns about blanket return-to-play guidelines and suggests that age-specific guidelines may be more appropriate [McCrory et al., 2009].

Sequelae

While the long-term significance of a single mild pediatric TBI is uncertain, several large studies have demonstrated multiple neurocognitive deficits in athletes who experienced an accumulation of lifetime concussions. Collins and colleagues reported impairments in executive functions and information-processing speed in college football players who experienced two or more lifetime concussions [Collins et al., 1999]. Other studies similarly showed neuropsychological deficits in amateur soccer players when compared with control athletes participating in noncontact sports [Matser et al., 1999]. Studies in adult rats have modeled this effect, demonstrating no significant deficit after a single mild injury, but definite impairment in spatial learning tasks after multiple repeated mild injuries [DeFord et al., 2002]. Individuals with previous concussions have shown a significantly longer duration of symptoms and length of recovery [Bruce and Echemendia, 2004; Guskiewicz et al., 2003]. More recent studies of retired professional football players suggest an increased risk of earlier-onset memory problems and even dementia. There also are anecdotal case reports of substantial tau protein deposition in the brains of ex-athletes who died of unrelated causes [Guskiewicz et al., 2005; McKee et al., 2009; Omalu et al., 2006]. Clearly, further work is needed to delineate the risk of chronic cognitive impairment and its relation to timing, number, and age at injury.

Skull X-Rays

Skull x-rays are still the most sensitive method to detect fractures. The presence of a skull fracture increases the likelihood of an underlying intracranial abnormality 21- to 80-fold [Quayle et al., 1997; Teasdale et al., 1990]. However, the absence of a fracture on skull x-ray does not preclude intracranial pathology. The frequency of intracranial lesions discovered on CT in the absence of skull fracture on x-ray ranges from 16 to 39 percent [Levi et al., 1991; Lloyd et al., 1997], but these results are skewed toward more severe injuries, as CT scans were obtained only in patients with “clinical indications” or who were hospitalized. Of all pediatric TBI patients with CT scans positive for an intracranial lesion, 35–50 percent occurred in the absence of a skull fracture [Lloyd et al., 1997; Quayle et al., 1997; Teasdale et al., 1990]. In patients where underlying brain injury is already suspected, the value of screening skull radiography has been questioned [Bell and Loop, 1971; Lloyd et al., 1997].

Skull fracture does occur much more frequently in infants less than 1–2 years old and is associated with intracranial hematomas, although less so than in older children [Lloyd et al., 1997; Quayle et al., 1997; Schutzman et al., 2001]. None the less, identification of isolated skull fractures in infants is still important, with the implications of diagnosing nonaccidental trauma. Also, significant scalp injuries or hematomas may overlie a skull fracture. Given this information, skull x-rays may be warranted in evaluation of TBI under the following circumstances: in settings where there is a suspicion of nonaccidental trauma, in head-injured infants below 1–2 years of age, particularly if there is scalp swelling or hematoma, or as a screening tool in older children only in areas where CT scanning is not readily available.

Computed Tomography

Noncontrast CT scanning is the neurodiagnostic test of choice in the evaluation of acute head trauma. Many studies have been performed to determine clinical indications for CT following pediatric TBI [Dunning et al., 2004; Halley et al., 2004; Homer and Kleinman, 1999; Kuppermann et al., 2009; Maguire et al., 2009; Murgio et al., 2001; Quayle et al., 1997]. The proper use of CT scanning depends upon the severity of the injury sustained and the age of the child. The rationale behind use of CT is to identify intracranial complications, particularly those that would warrant neurosurgical or medical intervention to achieve optimal outcome.

In a large, prospective series (n = 653) of children with TBI of all severities admitted to a neurosurgical department, 34.6 percent demonstrated significant intracranial abnormalities [Levi et al., 1991]. Fractures were excluded, but nonsurgical pathology, such as DAI and contusions, was included, along with acute hematomas. If only acute hematomas were tabulated, 23.7 percent of severely injured (GCS 3–8) children had a lesion, as compared with 12.4 percent of moderate (GCS 9–12) and 5 percent of mild (GCS 13–15) patients. A consecutive, prospective study of children (n = 322) seen in an urban emergency room for “nontrivial” head injury reported an 8.3 percent rate of intracranial pathology. In this case, “nontrivial” was defined as head injury with symptoms or physical findings, except for superficial scalp injuries, which were only deemed “nontrivial” in infants less than 12 months of age [Quayle et al., 1997]. Differences in these studies may be attributed to geographic area, inclusion criteria (neurosurgical admissions versus emergency department visits), distribution of injury severity, and age range. In ten studies published since 1990, each with n >100 patients, the frequency of significant intracranial pathology in all pediatric TBI ranged from 1.3 percent [Chan et al., 1990; Teasdale et al., 1990] to 34.6 percent [Levi et al., 1991]. A recent meta-analysis of 16 published papers examining variables predictive of intracranial hemorrhage in pediatric TBI reported that skull fracture was associated with a relative risk (RR) of 6.1 (that is, the presence of a skull fracture carried a 6.1-fold increased risk of intracranial bleed), focal neurological deficits (RR, 9.4), GCS score less than 15 (RR, 5.5), and loss of consciousness (RR, 2.2) [Dunning et al., 2004]. Seizure activity had a relative risk of 2.8; however, due to variability of this parameter between studies, it did not reach statistical significance. Headache and vomiting were not predictive. Given the high frequency of intracranial lesions in children with moderate or severe TBI, cranial CT is warranted in patients with a GCS <13. While the presence of the risk factors listed above should further raise suspicion of an intracranial process, the absence of a factor does not necessarily mean the risk of a lesion is reduced. Imaging recommendations are summarized in Box 74-2 and Box 74-3.

Box 74-2 Pediatric Traumatic Brain Injury >2 Years Old: Indications for CT Scanning

Definite Indications

Glasgow Coma score <15 in emergency department

Altered mental status

Skull fracture (on x-ray or clinically)

Focal neurological signs

Deteriorating neurological condition

Suspected nonaccidental trauma

Relative Indications

Loss of consciousness or post-traumatic amnesia

Seizure

Known coagulopathy

Need for anesthesia or prolonged sedation

Persistent or worsening headache, nausea, or vomiting

Box 74-3 Pediatric Traumatic Brain Injury <2 Years Old: Indications for CT Scanning

Definite Indications (High-Risk Patient)

Altered mental status, irritability

Focal neurological signs

Skull fracture (on x-ray or clinically)

Seizure

Bulging fontanel

Vomiting ≥5 times or >6 hours

Loss of consciousness ≥1 minute

Deteriorating neurological condition

Suspected nonaccidental trauma

Relative Indications (Intermediate-Risk Patient)

Vomiting 3–4 times

Loss of consciousness <1 minute

Previous irritability or lethargy, now resolved

Caregivers concerned about patient’s current behavior

High-energy mechanism of injury

Scalp hematoma, especially if large or nonfrontal

Known coagulopathy

Need for anesthesia or prolonged sedation

The indication for CT is more controversial when only mild TBI pediatric patients are considered. In this population, the true prevalence of intracranial injury is difficult to determine. Review of 12 studies published since 1986, each with n >100 patients, shows that from 0 percent [Immordino, 1986] to 19.1 percent [Wang et al., 2000] of mildly injured children will have a significant CT abnormality. Davis and colleagues reported intracranial bleeding in 7 percent of pediatric patients (n = 168) who suffered a loss of consciousness after mild TBI; however, in a subset of 49 patients with normal neurological examination, 0 percent were found to have CT abnormalities [Davis et al., 1994]. Other studies have shown lower utility of clinical variables and normal neurological examination in predicting the absence of intracranial problems in mildly injured children [Halley et al., 2004; Quayle et al., 1997]. A formal review of this question [Homer and Kleinman, 1999] concluded that it is clear that only a very small number of mildly injured children will have significant intracranial damage, but that clinical history and examination are relatively insensitive predictors of those who will. Intracranial lesions requiring surgical intervention do occur, albeit very rarely, even when the neurological examination is normal. Therefore, the American Academy of Pediatrics consensus guideline for mild head injury states that CT imaging (in conjunction with observation – see management, below) is a reasonable management option, but is not required in children older than 2 years of age who experience brief (<1 minute) loss of consciousness and who have no abnormal physical or neurological findings at the time of medical evaluation [American Academy of Pediatrics, 1999]. While there are currently no definitive guidelines for children with mildly abnormal GCS (13–14) or normal GCS (15) with loss of consciousness lasting over 1 minute, careful observation in all and liberal use of CT is recommended (see Box 74-2) [Dietrich et al., 1993; Halley et al., 2004; Savitsky and Votey, 2000]. A recently developed and validated clinical prediction rule shows promise in identifying the subset of children with mild TBI who do not require CT scanning [Kuppermann et al., 2009]; the specifics of this rule are discussed later (and see Figure 74-16 below).

Fig. 74-16 Clinical algorithm for CT scanning in children with minor head injury (GCS 14-15).

A. Algorithm for children less than 2 years old. B. Algorithm for children over 2 years old. *Includes motor vehicle crash with ejection, death of another passenger, or rollover; pedestrian or bicyclist without helmet struck by motor vehicle; fall >0.9 m (3 ft) for A, >1.5 m (15 ft) for B; head struck by high-impact object. Clinically important TBI (ciTBI) includes death, neurosurgery, intubation for more than 24 hours for TBI, hospitalization for more than 48 hours for TBI. ED, emergency department.

(From Kuppermann N et al., Identification of children at very low risk of clinically-important brain injuries after head trauma: a prospective cohort study. Lancet 2009; 374: 1160–1170.)

Recommendations for CT scanning are even broader when children under the age of 2 years are concerned. In this age group, skull fractures are more common, and abnormalities on neurological examination may not be as easily detected as in older children [Berney et al., 1994; Dietrich et al., 1993]. An expert review of the literature has resulted in a proposed guideline for stratification of risk in head-injured infants younger than 2 years of age [Schutzman et al., 2001]. Indicators of high and intermediate risk stratification are presented in Box 74-3. In general, this algorithm recognizes that low-risk patients (low energy mechanism of injury, no signs or symptoms, more than 2 hours since injury, and age over 12 months) may be discharged from the emergency department without CT scanning, but with continued observation by reliable caretakers. Kupperman and colleagues have also devised and validated a CT rule for children less than 2 years of age that shows very good negative predictive value for clinically significant TBI [Kuppermann et al., 2009] (see Management of mild TBI and Figure 74-16, below).

Magnetic Resonance Imaging

MRI has limited usefulness in the acute management of pediatric TBI, primarily because of the longer time needed for scanning. MRI is also less able to visualize bony structures and fractures, and confers no advantage over CT in diagnosing large (surgical) hematomas. MRI is better able to delineate structures in the posterior fossa, as well as subacute hematomas that may appear isodense on CT. The ability of MRI to distinguish the approximate age of an intracranial hemorrhage makes it useful in cases of nonaccidental trauma (Table 74-2; see Subacute and chronic SDH). DWI has been used to demonstrate areas of restricted diffusion related to post-traumatic secondary hypoxic-ischemic injury [Suh et al., 2001]. Particular MRI sequences (T2* GRE, and more recently, SWI) are especially good at detecting microhemorrhages that indicate foci of DAI [Tong et al., 2004] and might not be seen on conventional CT. MR spectroscopy performed subacutely (within the first week) in patients with persistent coma has been highly accurate at predicting outcome [Ashwal et al., 2000; Holshouser et al., 2000], and is discussed later in this chapter (see Figure 74-17 below), under the section on prognosis.

Fig. 74-17 Examples of children after traumatic brain injury with good or bad outcomes.

Anatomic MRI images: A, Patient with good outcome. B, Patient with poor outcome. Magnetic resonance spectroscopy: C, Patient with good outcome. D, Patient with poor outcome. Note the relative reduction of the N-acetylaspartate (NAA) peak and the presence of a lactate peak in the patient with the poor outcome. Cho, choline; Cre, creatine; Lac, lactate.

(From Ashwal S et al. Predictive value of proton magnetic resonance spectroscopy in pediatric closed head injury. Pediatr Neurol 2000;23:119–120.)

Advanced MRI techniques have shown promise in characterizing pediatric TBI and recovery from injury, but are currently used primarily for investigational purposes [Ashwal et al., 2006; Suskauer and Huisman, 2009; Difiori and Giza, 2010]. fMRI (altered patterns of network activation [Lovell et al., 2007; McAllister et al., 1999]), MR spectroscopy (abnormal cerebral metabolites [Babikian et al., 2006; Babikian et al., 2009a]), DWI (restricted diffusion [Babikian et al., 2009b]), and DTI (perturbations in diffusion along white-matter tracts [Ewing-Cobbs et al., 2008; Wozniak et al., 2007]) have all been reported to show changes after TBI, including pediatric TBI.

Angiography

Angiography is useful for diagnosing traumatic vascular lesions, particularly dissections of the great cervical arteries, traumatic intracranial aneurysms, and cavernous carotid fistulas. In some cases, MR angiography is being recommended as a screening test for suspected dissection in children, in preference to conventional angiography [Chamoun et al., 2008].

Ultrasound

An open fontanel in younger infants provides a window for ultrasound examination. While able to visualize more centrally located intracranial lesions, ultrasound is less sensitive at detecting epidural or subdural hematomas, and is not generally utilized in an acute trauma setting. Use of ultrasound is reviewed in more detail in Chapter 11.

Sedation and Neuromuscular Blockade

Sedatives and analgesics may be useful to reduce struggling and exacerbation of elevated ICP, promote optimal ventilation, alleviate pain, and allow for adequate nursing care (suctioning, catheter placement, etc.). A theoretical benefit of sedatives is to minimize increases in cerebral metabolic activity provoked by pain or stress. In addition, some sedatives possess anticonvulsant or antiemetic properties. A potential adverse effect of sedation is hypotension, which can exacerbate poor cerebral perfusion if not addressed promptly. A recent expert literature review found insufficient evidence to recommend a specific choice of sedative medications in the setting of pediatric TBI [Adelson et al., 2003d]. Sedatives with published reports of use in pediatric TBI include fentanyl, morphine, sufentanil, remifentanil, diazepam, ketamine, propofol, etomidate, and barbiturates. However, the majority of these reports were small series, often with mixed adult and adolescent patients. There is anecdotal evidence of ICP reduction with remifentanil [Tipps et al., 2000], ketamine [Albanese et al., 1997], propofol [Farling et al., 1989; Spitzfaden et al., 1999], and barbiturates. However, the use of a continuous propofol infusion in children has been associated with a lethal syndrome of metabolic acidosis [Bray, 1998; Canivet et al., 1994; Cray et al., 1998; Parke et al., 1992], and has not been approved for pediatric sedation by the FDA.

Neuromuscular blockade may aid in the management of TBI patients by promoting ease of ventilation, reducing ICP, and stopping metabolic demands from skeletal muscle. Complications of neuromuscular blockade include an increased risk of pneumonia, cardiovascular depression, immobilization stress (if used with inadequate sedation), unnoticed seizure activity, and prolonged paralysis/myopathy. Myopathy appears to be associated with the combination of nondepolarizing agents and corticosteroids [Martin et al., 2001; Rudis et al., 1996]. When using iatrogenic paralysis, it is important to monitor the depth of blockade periodically using a “train of four” stimulation. Vecuronium, pancuronium, and doxacurium have all been used for pediatric neuromuscular blockade in the setting of TBI.

Hyperventilation

Hyperventilation reduces blood and tissue PCO₂ levels, resulting in vasoconstriction, decreased cerebral blood volume, and reduced ICP. These effects can be seen within minutes, making hyperventilation an important tool in managing acute neurological deterioration or impending herniation.

However, excessive hyperventilation has been shown, in both adults [Kiening et al., 1997; von Helden et al., 1993] and children [Skippen et al., 1997], to result in reductions in cerebral blood flow that may reach ischemic levels. The routine use of prophylactic hyperventilation (PCO₂ <35 mmHg) in severe pediatric TBI is no longer recommended.

Controlled hyperventilation (PCO₂ 30–35 mmHg) may be useful to control elevated ICP that is refractory to sedation, neuromuscular blockade, CSF drainage, and hyperosmolar therapy. More aggressive hyperventilation (PCO₂ <30 mmHg) should be considered a second-tier therapy for intractably increased ICP. The risk of potential ischemic complications must be considered, and the patient monitored appropriately for these problems. Jugular venous oxygen saturation measurements via a jugular bulb catheter should be utilized. This methodology provides a more global measure of cerebral tissue perfusion and can be monitored frequently as the patient’s condition and treatment change. Alternative methods of measuring for ischemia include CBF studies and brain-tissue oxygen monitoring. CBF studies have the advantage of providing some regional data, but are usually limited to the actual time of testing. Brain-tissue oxygen levels can be monitored continuously over time but has the disadvantage of being limited to measurements in the vicinity of the probe [Figaji et al., 2009].

Despite concerns regarding the potential for hyperventilation-induced ischemia, hyperventilation remains a potent modulator of ICP. In the setting of acute deterioration or imminent herniation, aggressive hyperventilation may still be employed for short periods of time. With the advantage of fast onset of action, hyperventilation may provide a window during which diagnostic testing may be completed and additional first-tier ICP measures may be optimized.

Hyperosmolar Therapy

Two agents have been recommended for hyperosmolar therapy in pediatric TBI: 3% saline and mannitol. The theory behind hyperosmolar therapy is to raise the solute concentration of the serum, thus creating an osmotic gradient whereby free water is drawn out from the brain’s extracellular space. One drawback of the evidence-based medicine approach is made apparent when reviewing this therapy. Which agent is preferable, one with a long history of effectiveness but little in terms of blinded, placebo-controlled clinical trials, or a newer intervention with demonstrated efficacy in small but appropriately designed clinical trials? Mannitol, an agent whose clinical use for ICP management has become almost ubiquitous, appears clearly effective in many older adult studies [James, 1980b; Marshall et al., 1978; McGraw et al., 1978; Mendelow et al., 1985; Muizelaar et al., 1984; Schwartz et al., 1984], as well as those with mixed populations of children and adults [Nara et al., 1998; Rosner et al., 1995; Smith et al., 1986]. Unfortunately, the pediatric component of the mixed studies was not separately defined. Hypertonic saline has been more recently investigated [Peterson et al., 2000], and several studies used prospective data collection specifically in pediatric patients to document the effectiveness of this agent in controlling refractory ICP [Simma et al., 1998], yet this therapy lacks widespread clinical implementation.

Mannitol exerts its beneficial effects through several mechanisms. First, it reduces blood viscosity, improving blood flow and allowing for a reduction in cerebral blood volume [Levin et al., 1979; Muizelaar et al., 1983; Muizelaar et al., 1986]. Second is the osmotic effect described above, drawing free water out of the brain (across an intact blood–brain barrier) back into the bloodstream [James, 1980a]. A less well understood effect of mannitol is its ability to act as an antioxidant [Kontos and Hess, 1983]. Mannitol is administered with a bolus infusion, at 0.25–1.5 g/kg up to every 4 hours. Overall, mannitol begins to exert its ICP-reducing effects within 10–20 minutes, and these effects can last for several hours.

A caveat to remember when using mannitol is that, in the presence of a leaky or damaged blood–brain barrier, mannitol can potentially enter the injured area and cause a rebound osmotic effect [Kaieda et al., 1989; McManus and Soriano, 1998]. Partially for this reason, bolus rather than continuous administration of mannitol is recommended. Hypertonic saline exerts its effect through several mechanisms, including reducing blood viscosity, osmotically drawing free water out of brain parenchyma, restoring cellular membrane potential and volume [McManus and Soriano, 1998; Nakayama et al., 1985], reducing inflammation [Qureshi and Suarez, 2000], and improving cardiovascular function. The effectiveness of hypertonic saline in pediatric patients has been demonstrated in several recent studies [Khanna et al., 2000; Peterson et al., 2000; Simma et al., 1998]. Despite reducing ICP, none of these studies investigated whether the therapy resulted in better outcomes. Dosing of hypertonic (3%) saline was on a sliding scale titrated to ICP and serum sodium, but the infusion rates ranged from 0.1 to 1.0 mL/kg/hr.

Potential adverse effects of 3% saline include osmotic rebound, central pontine myelinolysis, and subarachnoid hemorrhage. For the four published pediatric 3% saline studies described, higher serum osmolalities were well tolerated (mean 365 mOsm/L, highest 431 mOsm/L) and none of the potential complications occurred.

With either of the hyperosmolar therapies described above, euvolemia is the goal, although serum osmolality is generally kept below 320 mOsm when using mannitol. Other osmotic agents used to treat brain edema have been less thoroughly evaluated, including urea and glycerol. While older management schemes attempted to reduce brain edema by systemic dehydration using diuretics, this strategy is no longer supported. In conclusion, hyperosmolar therapy is effective in controlling ICP elevations. The particular agent used (mannitol versus hypertonic saline) is a decision left to the treating physician.

ICP Monitoring – Indications and Treatment Threshold

Outcome from pediatric and adult TBI depends on avoiding secondary insults, such as ischemia and hypoperfusion [Chesnut et al., 1993a; Narayan et al., 1981]. Elevated ICP can lead to global hypoperfusion, as well as precipitating herniation, which can cause tissue ischemia via compression and, in some situations, by impinging upon cerebral arteries. Many of the specific therapies for severe TBI are directed at preventing excessive ICP; thus, it logically follows that monitoring ICP is an important component of TBI management [Becker et al., 1977; Marshall et al., 1979; Narayan et al., 1982]. Although there has been no randomized clinical trial specifically validating the benefits of ICP monitoring in children, many studies have demonstrated benefits of aggressive management of elevated ICP in head-injured children [Bruce et al., 1981; Cho et al., 1995; Peterson et al., 2000; Taylor et al., 2001].

There are three commonly used types of ICP monitoring methods. First are monitors that measure ICP via an external pressure transducer and include ventricular catheters, as well as subdural and subarachnoid catheters or bolts. These measure ICP by maintaining a continuous fluid connection with the intracranial space. They are accurate and can be recalibrated, but erroneous readings can result from obstruction of the connecting tubing or from inconsistent placement of the external transducer in relation to the patient’s head. The second type of monitor is a pressure transducer that is located at the tip of an inserted catheter. These monitors are not dependent upon maintaining a long, continuous fluid connection between an external gauge and the intracranial space; however, once placed, they cannot be readily recalibrated. The third type of monitor is a fiberoptic sensor placed at the catheter tip and this carries similar caveats to use of catheter-tip mounted transducers.

A pediatric study comparing Camino catheters (a tip-mounted fiberoptic sensor) with ventricular catheters found that they similarly measured ICP, although there were slight differences between the measured ICP from the two devices, dependent upon the patient’s GCS score [Gambardella et al., 1993]. Several adult studies addressing these issues have been published and summarized in published adult TBI management guidelines [Brain Trauma Foundation et al., 2000]. A ventricular catheter with an external sensor is considered the simplest and most reliable method to measure ICP and has the added benefit of allowing therapeutic removal of CSF. Catheter-tip mounted devices (pressure transducers or fiberoptic) are more expensive, but similarly accurate when placed within the ventricular catheter. Catheter-tip devices used for intraparenchymal placement may experience measurement drift, and external transducers coupled to subarachnoid, epidural, or subdural devices were felt to be less reliable; these modalities may be useful when ventricular access is problematic. ICP monitoring is an invasive procedure that requires a burr hole at a minimum, although suture lines may be used in infants/neonates. It is thus important to have standard indications for use of monitors in those children who are at greatest risk of having elevated ICP following TBI. In one series, 19 of 22 (86 percent) children with a GCS score ≤8 (i.e., severely injured), who received ICP monitoring, had ICP values >20 mmHg [Shapiro and Marmarou, 1982]. In a large series of severely injured patients (n = 654) from the National Traumatic Coma Data Bank, 72 percent had ICP >20 mmHg during the course of their monitoring [Marmarou et al., 1991].

These studies suggest that clinical determination of severe injury (GCS ≤8) delineates a group of patients at high risk of developing elevated ICP. Examination of less severely injured patients can provide greater sensitivity in determining clinical deterioration. However, particularly in young children and infants, clinical examination and GCS are less reliable than in older children or adults [Humphreys et al., 1990; Shapiro and Marmarou, 1982]. Using CT findings to predict elevated ICP can also be problematic in children, who are less likely to present with mass lesions than adults. Despite the presence of open fontanels and flexible sutures, infants can still develop elevated ICP [Cho et al., 1995]. Given this information, ICP monitoring is recommended in children with a GCS ≤8. In addition, ICP monitoring may be reasonable in selected patients with moderate TBI, particularly those whose clinical examination may be clouded by concomitant sedation, paralysis, or anesthesia. ICP monitoring can also still be beneficial in infants, particularly those with severe injuries due to abusive head trauma.

Normative values for ICP are generally felt to be lower in children, but the consequences of exceeding particular ICP levels have only been incompletely examined. Several retrospective studies in children have shown worse outcome in children with ICP elevations above either 20 mmHg [Esparza et al., 1985; Pfenninger et al., 1983] or 30 mmHg [Cho et al., 1995]. Other studies have shown cerebral compliance and CBF to be inversely proportional to ICP [Shapiro and Marmarou, 1982; Sharples et al., 1995a]. In general, it is recommended to institute ICP-lowering therapy in pediatric patients whose ICP exceeds 20 mmHg [Adelson et al., 2003c].

Complication rates for fiberoptic monitoring in children (n = 98) have been reported, with 7 percent of catheters becoming colonized (all with Staphylococcus aureus; average duration of catheter placement >7 days) and 13 percent having mechanical failure (on average by 9.5 days) [Jensen et al., 1997]. The complication rate was significantly different for catheters placed in the emergency department or in the intensive care unit. No pediatric reports of significant brain injury, hemorrhage, or seizure due to an ICP monitor placement or monitoring were found, suggesting a very low complication rate.

Recently, some debate over the utility of ICP monitoring following TBI has arisen after publication of two retrospective observational studies [Cremer et al., 2005; Salim et al., 2008]. In the pediatric study, results were compared between two groups (ICP monitor versus no ICP monitor) managed in one center between 1996 and 2006 [Salim et al., 2008]. The ICP-monitored group had significantly more severe overall injuries (Injury Severity Score [ISS] 25 versus 18, p = 0.001) and more severe head injuries (head Abbreviated Injury Score [head AIS] >3, 81 percent versus 55 percent, p = 0.01), as well as more frequent use of central venous catheter placement and use of mechanical ventilation. This group also had longer intensive care unit and hospital stays, and more systemic medical complications. After statistically correcting for the difference in TBI severity, there was no difference in survival between the two groups. No subgroup analysis was done for different types of TBI or age at injury. The adult study was performed at two centers with different TBI management protocols, including ICP monitoring [Cremer et al., 2005]. This study also found no survival benefit but did not perform subgroup analysis related to the type of TBI. These studies have suggested that a controlled clinical trial of ICP monitoring be considered. However, due to the limitations mentioned above, as well as the substantial existing physiological data suggesting the importance of ICP monitoring, there has been no consensus to limit the use of ICP monitoring in adult or pediatric TBI.

Cerebral Perfusion Pressure

Cerebral perfusion pressure (CPP) is defined as the difference between the mean arterial pressure (MAP) and ICP. Reduced CPP can occur during hypotension, intracranial hypertension, or both. Retrospective pediatric studies have shown a strong correlation between CPP <40 mmHg and death [Downard et al., 2000; Elias-Jones et al., 1992]. In one study of 118 children with ICP monitors placed within 24 hours of TBI, outcome was compared to the mean CPP over the first 48 hours of monitoring. None of the 22 children with a mean CPP <40 mmHg survived [Downard et al., 2000]. When outcome was assessed in comparison with CPP decile (i.e., 40–50 mmHg, 50–60 mmHg, etc.), no further relationship was found. Mean CPP <40 mmHg was thus interpreted as a limit of survivability in severe pediatric TBI [Downard et al., 2000]. More recently, using receiver operating curve (ROC) analysis and a novel pressure–time index (PTI), critical CPP thresholds of 48, 54, and 58 mmHg have been proposed. These values demarcate absolute CPP levels for the 2–6-year, 7–10-year, and 11–16-year age groups respectively, below which secondary brain injury contributes significantly to worsened global outcome [Chambers et al., 2006].

While low CPP is associated with poor outcome in children and adults, the benefits of CPP-directed therapy are uncertain. In adults, CPP of >80 mmHg has been associated with better outcome in two large, retrospective studies [Changaris et al., 1987; McGraw, 1989]. In one study, improved neurological outcome at 6 months was observed in patients managed for CPP and cerebral oxygen extraction (using jugular venous oxygen saturation monitoring) [Cruz, 1998]. However, a large randomized adult trial (n = 189) directly comparing CPP-directed therapy with ICP-directed therapy failed to show an outcome benefit of the former, and also reported an increased risk of acute respiratory distress syndrome in the CPP group [Robertson et al., 1999].

Since systemic blood pressure varies by age, it is reasonable to assume that optimal CPP will be age-dependent. Based upon the available evidence, maintaining CPP above 40 mmHg at all times in pediatric TBI patients is recommended, and the range of CPP from 40 to 65 mmHg most likely represents a spectrum of optimal CPP across the pediatric age range. Use of vasopressor agents may be associated with an increased risk of pulmonary complications. Further studies will be necessary to better delineate optimal age-specific CPP ranges.

Cerebrospinal Fluid Drainage

Physical drainage of CSF is the most direct means of reducing ICP, and has been done via intraventricular or lumbar catheters. In most studies, it is difficult to separate CSF drainage from other first- and second-tier means of ICP reduction. However, in one prospective adult TBI study, which compared patients with medical management only versus those with medical management with ventriculostomy (after initial removal of surgical lesions), it was found that the group without ICP monitoring had a four-fold higher mortality rate (53 percent versus 12 percent), and a one-third lower rate of independent living (20 percent versus 59 percent) [Ghajar et al., 1993]. In a pediatric study, risk of sudden increases in ICP was assessed using the pressure–volume index (PVI) technique [Shapiro and Marmarou, 1982]. This method relies upon the observation that brain compliance does not change linearly as fluid volume is added to the intracranial space. As the threshold for brain compliance is neared, addition of even small volumes of fluid results in dramatic increases in ICP. Reduced PVI (less than 80 percent of predicted normal) was shown to predict intracranial hypertension accurately. All patients were managed with CSF drainage, which was found to reduce ICP and increase PVI. Based on such findings, CSF drainage can be considered a reasonable treatment option to reduce elevated ICP in children.

Barbiturates

Barbiturate coma has benefit in cases of intractable ICP elevation by several potential mechanisms. Barbiturates are well known to suppress cerebral metabolism [Piatt, and Schiff, 1984]. They can reduce cerebral blood flow and volume by coupling with reduced cerebral metabolism [Kassell et al., 1980]. There is also some evidence that barbiturates decrease markers of free radical-induced damage in neonates with perinatal asphyxia [Singh et al., 2004].

Efficacy for barbiturates in refractory intracranial hypertension comes primarily from a study that included patients from 15 to 50 years of age [Eisenberg et al., 1988]. In this study, ICP control was achieved in 32 percent of the barbiturate group, which was about twice the response rate in the conventional treatment group. The 1-month survival rate was 92 percent among barbiturate responders, compared to only 17 percent of nonresponders.

The use of barbiturates in children with TBI has been described in several case series. One study reported cardiovascular complications and a high rate of requiring pressor support (91 percent), but did not mention whether there was any effect on cerebral hemodynamics or ICP [Kasoff et al., 1988]. In a different series of children with refractory ICP, addition of pentobarbital therapy achieved ICP control (<20 mmHg) in 52 percent [Pittman et al., 1989]. Dosing of pentobarbital is best titrated to burst-suppression on the EEG rather than serum levels, so continuous EEG monitoring should be used. Intravenous loading doses of pentobarbital range from 5 to 15 mg/kg, followed by either slow continuous infusion 1 mg/kg/hr, or 5 mg/kg intermittent doses every 4–6 hours.

The major complications of barbiturate therapy are myocardial depression and hypotension. Infants are perhaps particularly prone to these adverse effects. There is experimental evidence in neonatal rodents that even brief blockade of synaptic activity can result in a significant increase in neuronal apoptosis [Bittigau et al., 2002; Ikonomidou et al., 1999; Olney et al., 2002; Pohl et al., 1999].

The use of barbiturates is an option in patients with intractable elevation of ICP. Proper EEG and cardiovascular monitoring should be used at all times. The use of barbiturate coma in traumatically injured neonates and infants should be approached with caution.

Temperature Control and Hypothermia

Temperature control can be used either to avoid hyperthermia or to institute hypothermia. There is substantial experimental and adult clinical literature that suggests hyperthermia can exacerbate brain injury by increasing metabolism, triggering seizures, elevating levels of excitotoxins, promoting inflammation, and so on [Busto et al., 1989; Dietrich et al., 1996; Ginsberg and Busto, 1998; Wass et al., 1995]. It is recommended that fever/hyperthermia be avoided in brain-injured children.

Since the original report of hypothermia to treat TBI in children [Hendrick, 1959], interest in reducing temperature therapeutically has periodically been considered. Hypothermia can substantially reduce cerebral activity, and has been anecdotally associated with neurological preservation after prolonged drowning [Young et al., 1980]. In adults, several single-center studies showed either physiological benefits (improved CPP, reduced ICP) or trends toward better neurological outcomes [Clifton et al., 1993; Marion et al., 1993; Marion et al., 1997; Shiozaki et al., 1993], as did a recent meta-analysis [Polderman, 2008]. However, a large multicenter hypothermia trial that excluded children under the age of 16 years failed to show efficacy in improving outcome [Clifton et al., 2001]. There was a trend toward better outcomes in patients who were hypothermic at presentation and in younger patients, but there was no overall benefit.

More recently, a prospective randomized multicenter trial of therapeutic hypothermia for severe pediatric TBI was completed [Hutchison et al., 2008]. This study found a trend toward higher mortality (21 percent versus 12 percent, p = 0.06) and morbidity (unfavorable outcome, 31 percent versus 22 percent, p = 0.14) in the hypothermia group compared to the normothermia group. The hypothermia group also showed more systemic hypotension and use of vasopressors during the rewarming period than the normothermic patients. While this study utilized a 24-hour period of hypothermia with rapid rewarming, other investigators have reported that the short-duration/rapid-rewarming hypothermia protocol is associated with rebound increases in ICP [Adelson et al., 2005]. Unfortunately, a second multi-center hypothermia trial for pediatric TBI (‘cool kids’ trial) that uses a longer period of hypothermia with a more gradual rewarming period [Adelson, 2009] was halted in 2011.

Surgical Management of ICP

Decompressive craniectomy has been used to alleviate intractable intracranial hypertension and malignant cerebral edema. This procedure has been reported to lower ICP on average by 9–47 mmHg in children and adolescents [Cho et al., 1995; Polin et al., 1997; Ruf et al., 2003; Taylor et al., 2001].

One study divided 23 pediatric victims of nonaccidental trauma into a surgical group (with ICP >30 mmHg) and two medical groups (ICP >30 mmHg and <30 mmHg). Children in the surgical group and the medical group with less severe ICP elevation had significantly better outcome than the medically managed patients with ICP >30 mmHg [Cho et al., 1995].

Taylor et al. prospectively studied 27 children and randomly assigned them to best medical management with ventricular drainage (n = 14) or bilateral decompressive craniectomy with best medical management (n = 13) [Taylor et al., 2001]. Mean ICP was reduced by 9 mmHg postoperatively. Over a similar 48-hour period, the decompressed group had fewer episodes of ICP >30 mmHg (9 versus 59), fewer episodes of ICP >20 mmHg (107 versus 223), and a narrower range of ICP (11–25 mmHg versus 11–44 mmHg) when compared to the medical-only group. The surgically managed patients also tended towards a better Glasgow Outcome Scale score, although follow-up times were not standardized and mild pre-existing cognitive impairments were identified in two of the medically managed patients. Other studies reporting outcome after decompressive craniectomy were case series without a control/medical-only group.

The most commonly used surgical approach for decompressive craniectomy is a combination of bilateral craniectomy and expansion duraplasty [Cho et al., 1995; Dam et al., 1996; Guerra et al., 1999; Polin et al., 1997]. The studies reported above had another feature in common: surgery was undertaken less than 48 hours post-injury. Expert review of the literature suggested that severely injured children with intractably elevated ICP who are more likely to benefit from decompressive craniectomy could be selected on the basis of meeting at least some of the criteria listed in Box 74-4. In clinical practice, unilateral craniectomy has been performed ipsilateral to the hemisphere with greater edema or a surgical lesion, although there are no definitive published studies comparing bilateral with unilateral decompression in pediatric TBI.

Acute Anticonvulsant Therapy

Early after TBI, an acute seizure may act as a secondary injury, increasing cerebral metabolic demand, depleting neuronal energy reserves, causing release of excitotoxins, inducing hypoxia by impairing respiration or in association with aspiration, or, particularly in generalized tonic-clonic seizures, increasing body temperature and raising ICP. The incidence of early post-traumatic seizures (PTS) is higher in children than in adults, and, following severe pediatric TBI, early seizure rates of 20–39 percent have been reported [Annegers et al., 1980; Lewis et al., 1993; Ratan et al., 1999]. This risk is increased with lower GCS and younger age [Hahn et al., 1988a; Hahn et al., 1988b; Hendrick and Harris, 1968; Lewis et al., 1993; Ratan et al., 1999]. Most early PTS in children occur within the first 24 hours.

In adults, prophylactic phenytoin and carbamazepine reduce the risk of early PTS [Glotzner et al., 1983; Temkin et al., 1990; Chang and Lowenstein, 2003]. In children, one study determined early seizure prophylaxis based on caregiver preference [Lewis et al., 1993]. Fifty-three percent of untreated children and only 15 percent of the treated children experienced early seizures. A larger pediatric study retrospectively described 128 severely injured children (out of a total of 477 of all severities) and compared treatment differences between centers. The use of prophylactic anticonvulsants for early seizures ranged from 10 to 35 percent, and the overall risk of an early PTS in the entire cohort was 9 percent. However, the risk of early seizure by treatment group or severity group was not reported. It was noted that the group treated for early seizures had an improved survival rate [Tilford et al., 2001].

More recently, levetiracetam, a newer antiepileptic medication, has been evaluated as a possible alternative to phenytoin. The availability of an intravenous form, good side-effect profile, and lack of hepatic metabolism all make it an attractive medication. Two small head-to-head trials have shown no difference in the incidence of seizures between levetiracetam and phenytoin. One of these studies showed an increase in interictal EEG abnormalities with levetiracetam [Jones et al., 2008; Szaflarski et al., 2009].

Given the apparent effectiveness of prophylaxis against early PTS in adults, the potential risk of complications from early seizures, the increased risk of early seizures in children, and the frequent difficulty in identifying clinical seizure activity in infants and very young children, early prophylaxis for post-traumatic seizure activity in children should be considered for use in the first week after injury. Fosphenytoin can be administered intravenously more rapidly and has fewer infusion-related side effects than phenytoin. Loading doses of either drug range from 10 to 20 mg/kg, and maintenance doses are typically 5–7 mg/kg/day, divided at least into two doses per day. (In the case of fosphenytoin, 1 mg = 1 PE, or phenytoin equivalent.) For acute seizures occurring before early prophylaxis can be initiated or breaking through prophylaxis, intravenous bolus doses of lorazepam (0.05–1.0 mg/kg/dose) may be administered until fosphenytoin levels become therapeutic. Levetiracetam may also be considered, although a pediatric-specific trial remains to be carried out.

Prevention of Post-Traumatic Epilepsy

Despite the effectiveness of early prevention, several prospective, randomized adult studies have revealed the inability of anticonvulsant prophylaxis to prevent subsequent development of late post-traumatic epilepsy [Temkin et al., 1990; Young et al., 1983]. Given the potentially different mechanisms underlying the genesis of early and late PTS, this finding is not surprising.

In a trial of anticonvulsant prophylaxis for late PTS, 41 children were randomized in a double-blind, placebo-controlled study [Young et al., 1983]. Phenytoin or placebo was administered; those with phenytoin sensitivity were treated with phenobarbital. Patients were then followed for 18 months, and 12 percent of the treated group had a late PTS, as opposed to 6 percent of the untreated group (not significant). Importantly, compliance with treatment was poor. Therapeutic levels were present in only 50 percent of treated patients at 3 months’ follow-up, and by 1 year only 10 percent were therapeutic.

In addition to the above clinical data showing no efficacy for late PTS/pulmonary thromboembolism prophylaxis, there are also substantial animal data demonstrating that the immature brain is particularly sensitive to anti-excitatory pharmacological agents like anticonvulsants and anesthetics. Excitatory activity is necessary for survival of developing neurons and for on-going experience-dependent plasticity. Immature rodents show major increases in neuronal apoptosis and development of learning deficits when treated with many common anesthetics and anticonvulsants [Bittigau et al., 2002; Jevtovic-Todorovic et al., 2003; Olney et al., 2004]. (Topiramate and levetiracetam were notable exceptions [Kaindl et al., 2006].) These animal data are supported by human studies showing reductions in IQ for children treated with phenobarbital for febrile seizures [Farwell et al., 1990], and more recently showing an apparent dose-dependent increased risk for learning disabilities in children treated with anesthetics [Wilder et al., 2009].

Given the available data in adult and pediatric patients with late PTS and the potential concerns around use of agents that block neurotransmission in the developing brain, anticonvulsants cannot be recommended routinely for long-term prophylaxis after pediatric TBI. After the acute period, anticonvulsants can be gradually weaned. If a late PTS occurs, it should be evaluated and treated as a new-onset seizure [Adelson et al., 2003f].