Introduction

Traumatic brain injury (TBI) is one of the leading causes of disability in the USA with an estimated 13.5 million individuals affected by varying severities of TBI [1]. TBI is also a leading cause of acquired disability and mortality in the pediatric population. In 2014, over 837,000 TBI-related emergency department visits, hospitalizations, and deaths were estimated in American children by the Centers for Disease Control and Prevention [2]. Many survivors live with significant disabilities that are associated with a major socioeconomic burden. In 2010, the economic impact of TBI in the USA was estimated to be $76.5 billion in direct and indirect costs [3]. Accurate prognostication from physiatrists throughout the continuum of care from acute hospitalization through rehabilitation is important as it enables improved patient and family counseling, prioritizes rehabilitation goals, and justifies the allocation of healthcare resources.

TBI by nature is heterogeneous with no two injuries exactly alike. The patient’s premorbid state of health, in addition to primary and secondary injury factors, affects how the patient will respond to the TBI. The defining outcome is also difficult to standardize with definitions of a “good” versus a “poor” outcome variable between studies. The Glasgow Outcome Scale (GOS) is popularly employed for its simplicity, but is limited by its broad categories, which stratifies patients recovering from TBI into five groups depending upon their ability to perform activities of daily living and the amount of supervision required (Table 1) [4]. However, it can be argued that five outcome categories are insufficient to represent the wide range of mental and physical disability that a patient can have following TBI [5]. To better represent the patient’s possible disability, the Glasgow Outcome Scale Extended (GOSE) further splits severe disability, moderate disability, and good recovery into upper and lower categories to allow for greater differentiation between levels of recovery [5, 6]. This gives greater sensitivity for detecting changes in condition over time and improves accuracy when assessing ongoing treatment or care needs (Table 2) [7]. Currently, there are no universally accepted scoring systems that reliably predict outcomes in patients following TBI.

Table 1 The Glasgow Outcome Scale
Full size table
Table 2 Glasgow Outcome Scale Extended
Full size table

The literature suggests that the majority of physicians do not feel knowledgeable to accurately assess prognosis in TBI [8]. Traditional predictors include demographic factors such as age, Glasgow Coma Scale (GCS) scores, length of coma, length of posttraumatic amnesia (PTA), and the presence of structural abnormalities on neuroimaging. Novel and emerging predictors include biomarkers and advanced magnetic resonance imaging (MRI). This article reviews the most recent work related to outcome prediction following moderate-to-severe TBI to improve physiatrists’ ability to provide valuable prognostic information to patients and families (Table 3 and Fig. 1).

Table 3 Summary of predictor variables in adults
Full size table
Fig. 1
figure 1

Predictor variables of outcome in adults with TBI

Full size image

Overview of Potential Predictor Variables

Age

In the adult population, advanced age has been associated with worse outcomes in patients with severe TBI [9,10,11,12]. While some studies have identified threshold values for poor outcomes, other studies have treated the outcome as a continuous function of age [10, 11, 13, 14]. Hukkelhoven et al. examined four prospective series including 5600 patients and showed a linear association between age and both mortality and unfavorable outcome, which is defined as severe disability or vegetative state (VS) on the GOS [15]. It has been hypothesized that the adult brain has a decreased capacity for repair as it ages due to the declining number of functioning neurons combined with a diminished cognitive reserve [16, 17]. Overall, studies have shown that in the adult population, there is a linear association between advancing age and poorer functional outcome following severe TBI [15].

In the pediatric population, there are competing conceptions regarding the vulnerability of the developing brain to injury versus a child’s enhanced capacity for neuroplasticity and recovery, which contributes to the lack of consensus on the impact of age on outcome upon TBI [18]. In general, children and adolescents have superior functional outcomes and rates of survival compared to adults sustaining severe TBI, though reports describing outcomes based on a child’s age at injury are inconsistent [19•, 20]. Several studies evaluating intellectual function in TBI suggest lower IQ scores on initial testing and reduced recovery in infants and preschool-aged children compared to older children and adolescents over a 1–2-year follow-up period [21]. Keenan et al. reported high but relatively stable rates of attention deficit hyperactivity disorder and affective symptoms in children 6–11 years of age for 2 years after TBI; however, preschoolers demonstrated increasing symptoms over time, indicating variable symptom trajectories based on age at injury [22•]. This observation may be attributable, at least in part, to the lower demands on abstract or higher-level cognitive skills placed on children at younger developmental levels. Andruszkow et al. noted that children aged 0–7 years who sustained moderate-to-severe TBI had more favorable outcomes at 10-year follow-up compared to (a) older children and (b) adults as measured by the GOS but found no differences based on age at the time of injury for other physical and psychological outcome measures [23]. This finding may be impacted by the lower sensitivity of the GOS for detecting common impacts on children with TBI in comparison to more detailed and tailored measures such as the GOSE for Children [24]. Levin et al. reported variations in word fluency recovery based on age at time of injury as well as injury pattern (focal vs diffuse TBI). Younger children (mean age 7 years) with severe diffuse TBI displayed less improvement in word fluency over time compared to older children and adolescents, but younger children with focal left frontal lesions had less pronounced word fluency impairments compared to older children with similar focal injuries [25]. Interpretation of the literature is confounded by variable distributions of age categories and timing of follow-up as well as the potential influence of abusive head trauma that may have a delayed presentation in younger children and result from repeated insults [20].

Glasgow Coma Scale Score

The GCS has been widely used as a measure to record the level of consciousness and severity of injury in patients who have sustained a TBI (Table 4). GCS is an essential aspect of the primary trauma assessment and is monitored throughout the course of the acute hospitalization. The role of GCS in prognosticating outcomes following severe TBI has been extensively studied [26, 27]. There has been variability in the degree of correlation of initial GCS score with prognosis, with some studies finding a strong correlation of initial GCS score and long-term prognosis measured via the GOS, while other sources have noted a weak and somewhat inconsistent correlation [27,28,29,30,31,32,33]. This variability in outcome may be related to the fact that GCS is affected by numerous factors unrelated to the injuries such as alcohol, illicit drug intoxication, and sedative use. These limitations are most notable in patients with severe injuries as they are intubated and sedated, making accurate assessment more challenging [34,35,36]. Additionally, the timing of GCS measurement can impact the reliability of prognostic value, with the admission GCS found to be more reliable in terms of prognostication compared to the preadmission GCS [34,35,36].

Table 4 The Glasgow Coma Scale
Full size table

In particular, the best motor response on admission is found to be the most useful GCS predictor of long-term functional outcome in patients with severe TBI [37]. Several studies have suggested that utilizing the motor component as an alternative to total GCS with similar prognostic value [38•, 39, 40]. This is most notable in those with severe injuries where verbal and eye responses are less reliable; thus, most of the predictive power of GCS is derived from the motor score [41•, 42]. While lower admission GCS score, particularly the motor response component, is associated with worse outcome, studies have shown mixed results regarding long-term outcome based on GCS alone given the wide range of confounding factors that may alter GCS scoring on admission.

IMPACT Prognostic Calculator

The IMPACT database includes patients with moderate-and-severe TBI (GCS ≤ 12) from eight randomized controlled trials and three observational studies conducted between 1984 and 1997 [43]. The intent of the IMPACT database was to produce a model that could be used to predict outcome after TBI at the point of admission to hospital. The investigators created three prognostic models: (1) the core model (age, motor score component from GCS, and pupillary reactivity), (2) the extended model (core model with addition of secondary insults [hypoxia and hypotension], CT findings, traumatic subarachnoid hemorrhage, and epidural hematoma), and (3) the lab model (extended model with additional information on hemoglobin and blood sugar). Validation of the model has shown that all IMPACT predictors have statistically significant associations with 6-month GOS in univariate and multivariable analyses. A poor outcome (GOS 1–3) occurred especially for those with GCS motor scores 1 or 2. Pupillary reactivity, hypoxia, and hypotension also had strong prognostic effects. Lower hemoglobin levels and higher glucose levels were associated with poor outcomes, but the effects were more moderate than other variables such as age [44].

Radiographic Imaging

Computed Tomography

Computed tomography (CT) is the preferred imaging modality for the acute evaluation of patients with moderate-to-severe TBI due to its availability, rapid image acquisition, and ability to detect fractures of the skull and intracranial hemorrhage that require immediate neurosurgical attention. In patients with TBI, the outcome is better with the absence of intracranial abnormalities [45]. The presence of basal cistern collapse is an indicator of increased intracranial pressure and is associated with an unfavorable outcome following severe TBI [43, 46,47,48,49,50,51,52]. The Marshall CT classification system is a prognostic system that incorporates anatomical CT scan findings [53]. The Marshall CT classification uses CT scan findings on the status of the mesencephalic cisterns, the degree of midline shift, and the presence or absence of local lesions to categorize patients into six different groups (Table 5). The presence of a midline shift of > 5 mm on initial brain CT scan and a high- or mixed-density lesion > 25 cm3 in volume have both been correlated with early death [54]. In patients with TBI, the Marshall criteria have been correlated with 6-month outcomes with all patients in the diffuse brain injury I group experiencing a good recovery (GOS 4 or 5). In the diffuse brain injury II group, older age > 40 and the presence of multiple parenchymal lesions on CT scans were significantly correlated with poor outcomes (GOS 1–3). For the diffuse brain injury III and IV groups, the only significant prognostic factor was the GCS score with lower initial GCS of 3 to 5 correlated with poor outcomes (GOS 1–3). Outcomes were unfavorable in most patients with intracerebral hematoma [55].

Table 5 Marshall CT classification system for a head injury
Full size table

Magnetic Resonance Imaging

While CT is the initial imaging technique of choice in the setting of moderate-to-severe TBI, MRI is more sensitive in detailing traumatic lesions of the brain parenchyma, especially in the posterior fossa and identifying non-hemorrhagic lesions [51, 56,57,58,59, 60•]. Several published reports have proposed that brain stem lesions are a marker of poor prognosis following TBI. If MRI is available in the first few months following TBI, bilateral brainstem lesions make a good recovery unlikely on GOS [58–59, 60•, 61,62,63]. Anterior and non-hemorrhagic lesions on MRI showed the highest positive predictive value for a good outcome on the GOS [60•].

Quantitative Diffusion-Weighted MRI

Diffusion-weighted imaging (DWI) is a specific MRI sequence, based on the changes in the diffusion of water molecules in brain tissue, which is sensitive to brain ischemia. Restricted diffusion is seen when there is cell death due to significant tissue injury and combining images obtained with different amounts of diffusion weighting producing a diffusion coefficient (ADC) map. The ADC is sensitive to acute and subacute TBI and has been shown to provide prognostic information [64•]. If MRI is completed in the initial 2 weeks following injury, a threshold ADC < 400 × 10−6 mm2/s in 0.49% of brain tissue predicts a poor outcome (discharged to a skilled nursing facility or died) versus a good outcome (discharged to inpatient rehabilitation or home) [64•]. While the initial investigation has shown some preliminary promising results, large prospective studies that track patient outcome over a longer period are necessary to better elucidate the prognostic role of quantitative diffusion-weighted MRI in moderate-to-severe TBI.

Diffusion Tensor Imaging and Magnetic Resonance Spectroscopy

Combining diffusion tensor imaging (DTI) with magnetic resonance spectroscopy (MRS) has been shown to be 86% sensitive and 97% specific for predicting unfavorable outcomes (death, persistent VS, or minimally conscious state (MCS)) 1 year after TBI [65]. DTI is a relatively new imaging technique that can be used to evaluate white matter in the brain detailing the orientation and direction of white matter fiber tracts. This method involves quantifying the orientation and directional uniformity of water diffusion in brain tissue. In white matter, the mobility of water is restricted in directions perpendicular to the axons oriented along the fiber tracts, and thus, the orientation and direction of these fibers can be traced. MRS differs from conventional MRI in that the information from protons in water is suppressed, allowing proton signals from other metabolites to be measured. The use of multimodal magnetic resonance techniques combining DTI with MRS imaging performed in the subacute period following severe TBI provides valuable information on both primary and secondary brain injuries, assessment of metabolic changes within the brain, and prognostic information on the unfavorable outcome at 1 year [65].

Coma

The duration of coma has been shown to have a near-linear relationship with duration of recovery time following TBI with diffuse axonal injury without focal cortical injury [66]. Prolonged duration of coma portends a worse outcome following severe TBI with a length of coma greater than 4 weeks making good recovery unlikely, while severe disability is improbable when it is less than 2 weeks [4, 67••].

Further evidence has demonstrated that a more favorable prognosis is observed in those who have both traumatic etiologies and diagnosis of a MCS as opposed to a VS at the time of inpatient rehabilitation admission [68,69,70]. Additionally, it has been shown that patients admitted to inpatient rehabilitation at a VS or MCS level of recovery were able to generally progress beyond the post-confusion level. This occurred even if patients had MCS of 3 months or longer. Nearly half of patients with long-term follow-up achieved recovery to safe, daytime independence at home, and 22% returned to work or school within 2 years after injury. A noteworthy proportion (17%) returned to productive pursuits at or near their previous level of functioning. Overall, there was a favorable prognosis for the continued evolution of recovery to post-confusion levels for patients with prolonged, severe disorders of consciousness [69].

Posttraumatic Amnesia

PTA is a period of time after a TBI when the brain is unable to form continuous day-to-day memories. The Galveston Orientation and Amnesia Test (GOAT) is a tool developed to evaluate cognition serially during the subacute stage of recovery from TBI [71]. This scale measures orientation to person, place, and time and memory for events preceding and following the injury. The duration of PTA is defined as the period following a coma in which the GOAT score is < 75. PTA is considered to have ended if a score ≥ 75 is achieved on two consecutive administrations [72, 73]. The orientation log (O-Log) is a brief measure of orientation as an alternative to the GOAT. An advantage of this brief measure is ease of administration, provision for cueing if the patient is not able to respond or responds inaccurately, consistent scoring across items, and exclusion of questions pertaining to personal information or hard-to-verify amnesia-related items [74]. The duration of PTA is defined as the period following a coma in which the O-Log score is < 25. PTA is considered to have ended if a score ≥ 25 is achieved on two consecutive administrations.

Classically, PTA duration has been a powerful prognostic factor and has been particularly useful for inpatient rehabilitation physicians managing patients recovering from severe TBI as many recovery thresholds are crossed during a stay at an inpatient rehabilitation facility [4, 67••]. The longer the duration of PTA, the worse the patient outcome with severe disability unlikely when PTA is less than 2 months, though good recovery is unlikely when PTA duration is greater than 3 months [67••]. More recently, upon further evaluation of the relationship between PTA duration and patient outcome, PTA duration ending within 4 weeks resulted in severe disability being unlikely at year 1, with good recovery being the most likely at year 2. Conversely, if PTA lasted beyond 8 weeks, good recovery was highly unlikely at year 1, and severe disability was about as likely if not more likely than moderate disability at year 2 [75••]. In addition to improved outcome from year 1 to year 2, there is evidence that patients with a shorter PTA duration may experience continued late improvement between year 2 and year 5 [76].

Expanding on the GOS as a measure of the level of disability and global neurological functional outcome following TBI, the GOSE is the recommended core global measurement in TBI research [7, 77,78,79,80]. The PTA duration and its effect on GOSE scores at 1, 2, 5, and 10 years post-injury have been assessed. Compared to previous threshold values outlined, patients with PTA duration less than 18 days were found to have increased GOSE scores at the 1-, 2-, and 5-year intervals compared to those with PTA greater than 19 days [81].

Further illustrating the importance of PTA duration and its impact on functional outcomes, a study followed patients at 1-, 2-, and 5-year intervals post-injury after having moderate-to-severe TBI and being discharged from a rehabilitation facility [82•]. The functional independence measure (FIM) instrument was used to evaluate outcomes, allowing for interpretation of the required hours of care and subsequently the burden of care [83, 84]. Patients were divided into either having non-extremely severe PTA (≤ 28 days) or extremely severe PTA (> 28 days). Ultimately, those with non-extremely severe PTA were noted to have higher FIM scores at all time points relative to those with extremely severe PTA [82•]. These results can further supplement the previously described threshold values discussed earlier.

PTA duration has a significant impact on other prognostic measures, including a return to work (RTW) and intelligence. Specifically, PTA duration of 12.1 days for those with moderate or severe TBI predicted early RTW (i.e., within 6 months of injury), whereas patients with PTA duration of 26.2 days had a late RTW [85•]. These findings were elaborated further with decreased probability of employment at 1-, 2-, and 5-year follow-up of patients with moderate-to-severe TBI with increased PTA duration [86].

Regarding intelligence, a large meta-analysis showed that PTA duration can help predict potential declines in intelligence. Patients with severe TBI and increased PTA duration were found to have significantly larger depressions in IQ domains, specifically full-scale IQ, performance IQ, and verbal IQ [87]. Previously, evidence has suggested that predictive value of PTA duration for intelligence is superior to that of either GCS scores or coma duration [88, 89].

The above evidence suggests the importance of using PTA duration to prognosticate patients with severe TBI with respect to multiple domains. Despite evidence demonstrating the importance of the length of coma and prognosticating TBI, PTA duration has been described as a significant unique predictor of patient FIM score at discharge, whereas duration of coma was not [90]. PTA duration remains the most robust injury severity predictor of long-term outcomes following TBI, demonstrating to predict the degree of recovery [67••, 75••], the severity of disability [75••], ability to anticipate traffic hazards [91], independent living status after 1 year [92], and RTW [85•, 86, 93].

Biomarkers

S100B

S100B is an astrocytic protein specific to the central nervous system with serum and cerebrospinal fluid (CSF) levels increasing following TBI. Some studies have questioned the validity of S100B levels as a reliable prognostication tool citing confounding extracerebral origins from long bone fractures, variable collection timing in relation to injury, and strategies employed in the literature to correlate S100B levels to a patient outcome [94,95,96,97,98,99,100,101,102]. Serum and CSF levels are elevated early across the first 6 days following severe TBI and diminish over time. It has been shown that subjects with higher CSF S100B concentration in the first week following injury had a higher acute mortality and worse GOS and Disability Rating Scale (DRS) scores at 6 months post-injury. Higher mean and peak serum S100B levels were predictors of acute mortality following severe TBI [103].

Glial Fibrillary Acidic Protein

Glial fibrillary acidic protein (GFAP) is a monomeric intermediate filament protein of astrocytes that may be measured in serum or CSF samples. Median serum GFAP has been shown to be high in patients who died following severe TBI compared with patients who were alive at 6 months post-injury. Patients with a poor outcome (GOS 1–3) had higher serum concentrations of GFAP than those patients with good outcome (GOS 4 or 5) at 6 months post-injury [97]. Persistent elevation of GFAP on day 2 following TBI is predictive of increased mortality [104].

Ubiquitin C-Terminal Hydrolase L1

Ubiquitin C-terminal hydrolase L1 (UCH-L1) is a deubiquitinating enzyme that is selectively expressed in the brain and is required for synaptic function. Elevated CSF levels of UHC-L1 in the first week post severe TBI were associated with increased mortality at 6 weeks post-injury and poor outcome (GOS 1–3) at 6 months post-injury [105]. Serum UCH-L1 levels in the first 7 days have been shown to independently predict mortality in the severe TBI population [106].

Additional study has shown that serum levels of S100B, GFAP, UCH-L1, and SBDP150 biomarkers predict unfavorable clinical outcomes (GOSE 1–4) assessed 6 months after moderate-to-severe acute TBI. When all four biomarkers were above cutoff threshold, 77% of subjects experienced a poor outcome (GOSE 1–4), while 22% experienced a poor outcome when all four biomarkers were below the cutoff threshold. Elevations of S100B, GFAP, UCH-L1, and SBDP150 levels early following moderate-to-severe TBI independently predicted outcome. A predictive model combining S100B and GFAP levels with patient variables including age, sex, GCS, and CT findings provides a sensitivity of 67% and specificity of 83% in predicting a poor outcome (GOSE 1–4) [107••].

Somatosensory Evoked Potentials

Somatosensory evoked potentials (SSEPs) may serve as a prognostic tool in the severe TBI population. The median nerve is stimulated at the wrist, and recordings are obtained at key locations from the neck at the level of C-2 (N13 potential) and the scalp (N20 potential). The peak latencies of N13 and N20 potentials may be used to calculate the peak-to-peak central conduction time (CCT), and using Judson’s protocol, CCT may be classified as normal, increased latency (> 7 ms in patients < 50 years of age or > 7.3 ms in patients > 50 years of age or difference between sides > 0.8 ms), unilaterally absent, or bilaterally absent [108]. Since 1979, there have been several studies documenting the use of SSEP monitoring to help predict the outcome of patients who have suffered severe TBI [109,110,111,112,113,114]. Judson et al. studied 100 adults with severe TBI and found that if the SSEPs were absent bilaterally, the outcome was always death or a VS [108]. Blinded review SSEPs for 105 TBI patients revealed that SSEP monitoring in the ICU provides a reliable predictor of outcome following severe TBI. It has been shown that 60% of patients can expect a good outcome if CCT is normal bilaterally and the chances of death are less than 10% [113]. In contrast to the anoxic brain injury population, the presence of normal SSEPs is an indicator of awakening from a coma and good outcome [115]. Unilateral or bilateral increased CCT latency decreases the chance of a good outcome to less than 30%, and if an N20 is absent, the likely outcome is severe disability, VS, or death [113]. Bilateral loss of N20s has been thought to be a reliable marker of poor prognosis. Performing SSEPs early following injury and performing only one SSEP examination in cases of bilateral loss of N20s may lead to an underestimate of long-term functional recovery, and repeat SSEPs should be considered in the early acute phase of severe TBI. Confounding factors that should be considered when interpreting SSEPs include (1) cervical cord injury, (2) the presence of a scalp hematoma that would reduce the amplitude of the cortical potential, (3) brainstem hemorrhage, and (4) muscle activity [113]. As a consequence, caution is warranted in predicting a poor prognosis based predominantly on SSEPs in patients with TBI and should be considered in the overall clinical picture [116,117,118,119].

Overall, studies have indicated that (1) bilaterally normal SSEPs usually imply a good outcome, (2) bilaterally absent SSEPs are strongly predictive of poor outcome, and (3) SSEPs provide valuable complementary information to assist with the prediction of outcome in patients with severe TBI.

Predictor Variables in the Pediatric Population

Severe pediatric TBI is associated with estimated 16–22% mortality and 50.6% rate of unfavorable outcomes at 6 months [20, 120]. Healthcare-related quality of life across multiple dimensions is diminished in up to 40% of children 12 months following moderate or severe TBI [121]. In addition to persistent physical impairments, traumatic injury to the developing brain is of particular concern and reflected in neuropsychological and academic impairments following TBI in young children [122, 123].

Outcome prognostication in children with TBI is challenging and influenced by multiple factors to include injury severity, mechanism, age, the presence of concomitant injuries, physiological factors, social and family characteristics, and trajectory of early recovery [19•, 20, 121, 124]. Injury severity represented by the GCS score, particularly that obtained post-resuscitation, has been shown to have reliable predictive power [124]. A retrospective cohort study of 888 patients with severe TBI found mortality rates decreased as GCS scores increased (GCS 3, 53% mortality; GCS 5, ∼ 9% mortality; GCS 8, ∼ 3% mortality) in children aged ≤ 15 years [19•]. Hypothermia (temperature less than 36 °C) at presentation in combination with bilateral fixed and dilated pupils and a GCS of 3 confers the greatest TBI mortality risk [125]. Though children with severe injuries (GCS 3–4) have high rates of death and severe disability, 14.9% had good long-term global outcomes defined by the achievement of independent living, employment, or academic participation with minimal neurologic or cognitive deficits [125].

As has been well defined in adults, longer PTA duration is associated with less favorable outcomes in pediatric TBI. The validated scale specifically developed for assessing PTA in children is the Children’s Orientation and Amnesia Test (COAT) [126]. An additional tool often utilized for the assessment of PTA in adolescents is the O-Log [74]. A child is deemed to have emerged from PTA on the first of two consecutive days on which the COAT score falls within the average range for the child’s age [126]. A systematic review evaluating school-aged children (6–18 years) with TBI found that longer PTA duration is associated with worse outcomes in global functioning, memory, problem-solving, executive functioning, social functioning, academic performance, independent living skills, and self-care [126]. In most studies, PTA duration was a stronger predictor of outcomes than GCS or length of coma [126]. An additional recovery milestone with prognostic utility is the time from injury to following commands (TFC). TFC often occurs during the acute phase of recovery and is associated with poorer outcomes if extending beyond 26 days [124, 127, 128].

The health and function of the family unit are instrumental in child development and have been shown to play a significant role in TBI recovery. Lower socioeconomic status and increased parental stress not only are risk factors for sustaining TBI in childhood but also influence recovery as essential environmental factors [129]. Restricted family access to financial and social resources is associated with impaired intellectual function in children after TBI [129, 130]. Positive family function, including parent psychological functioning and communication skills, is associated with improved psychosocial and behavioral outcomes [17, 21, 22•, 129]. Neuroimaging in pediatric TBI provides useful diagnostic information and can also be utilized for outcome prediction. CT is the most frequently utilized and appropriate modality for initial evaluation of TBI but may only reveal 30–60% of lesions identified by standard MRI [131]. Studies have examined the impact of injury burden with respect to total lesion volume as well as anatomic location and depth of brain regions affected on outcomes in pediatric TBI. Greater fluid-attenuated inversion recovery (FLAIR) hyperintensity total lesion volume correlated with worse GOSE Pediatric scores and IQ at 12 months, as did the presence of lesions in all three anatomic zones of the brain (cortical, middle, and brainstem) compared to lesions isolated to higher zones [131]. Deeper lesions are also associated with lower GCS scores, and lesion depth is most predictive when assessed at the time of discharge from rehabilitation compared to a 1-year follow-up [132].

Conclusions

The role of physiatry in evaluating patients following moderate-to-severe TBI is a valuable but difficult proposition. Appropriate distribution of acute treatment and rehabilitation resources is of utmost importance. For prognostication to be clinically useful, outcomes must provide a reasonable impression of what life will be like for the patient in the longer term. Dichotomizing outcomes into “good” or “poor” may not be accurate enough to provide granularity to counsel patients or their families sufficiently. The needs of families should be considered in future research and it is important that researchers consider the types of prognostic information that families desire. In the future, additional work will be needed to better combine predictor variables tailored with precision to the patient to provide a clearer picture of individual outcome following moderate-to-severe TBI.