Best motor response
Extensor response (decerebrate): elbow extension with pronation and adduction (to local painful stimulus, sometimes spontaneously)
Abnormal flexion (decorticate): slow withdrawal with wrist pronation, shoulder adduction (to local painful stimulus, sometimes spontaneously)
Withdraws: normal flexion of elbow or knee to local painful stimulus
Localizes: other limb moves to site of painful stimulation
Obeys commands (exclude grasp reflex or postural adjustments)
Incomprehensible speech (moans and groans only – no words)
Inappropriate words (intelligible words but mostly expletives or random)
Confused conversation (attends and responds but answers muddled/wrong)
Orientated (knows who, where, when; year, season, month)
To pain applied to limbs (not face, where grimacing can cause closure)
To speech when spoken to (not necessarily the command to open eyes)
Spontaneous (indicates arousal, not necessarily awareness)
Mild 12< GCS
Moderate 9≤ GCS ≤12
Severe GCS <9
Patients may be found unconscious from the time of injury, a typical occurrence in DAI, or may present a “lucid interval” between trauma and loss of consciousness (patients that “talk and die”).
This lucid interval is poorly defined and does not seem to be related to prognosis. It typically occurs in epidural or subdural hematomas.
Patients with severe TBI require intensive care unit (ICU) admission.
4.1.3 Diagnostic Markers
Glasgow Coma Scale
CT scan (on admission and then repeated according to guidelines)
ICP monitoring (if indicated)
Brain oxygen monitoring (if indicated)
Jugular vein O2 saturation
Brian tissue O2 saturation
4.1.4 Top Differential Diagnosis
The traumatic event and its temporal correlation with typical posttraumatic lesions usually excludes any differential diagnosis.
Nevertheless, sometimes the traumatic event may be secondary to other acute diseases. Patients suffering acute impairment of consciousness from neurological (e.g., stroke or epileptic seizure), cardiac (e.g., myocardial infarction or arrhythmias), or toxic/metabolic causes may fall or be involved in various types of accidents leading to TBI.
This issue is not only significant on a legal basis, but deserves clinical attention, since concomitant diseases may exert an important role in patient management.
In dealing with TBI patients, suspicion of possible causative (or at least favoring) acute diseases should arise on the basis of:
Patient history (e.g., presence of epilepsy or heart disease)
TBI history from witnesses
Neuroimaging (e.g., intraparenchymal or subarachnoid hemorrhage suggesting intracranial spontaneous bleeding)
The differential diagnosis between accidental and postaggression TBI is a legal issue that goes beyond the scope of this review.
However, clinicians should be trained to recognize child abuse as a particular postaggression TBI.
It should be noted that magnetic resonance evidence of recurrent subdural hematomas in children are highly suggestive of the “shaken baby syndrome.”
Patients affected by cognitive or sensory-motor diseases may undergo TBI with increased frequency. Differentiating previous and ongoing disease may be challenging.
Major efforts have been made to establish prognostic criteria for severe TBI.
This is motivated not only by the high incidence of severe TBI in young people but also by the compelling need to streamline resource allocation.
Although it is not always formally justified, conclusions about severe TBI prognosis are commonly extended to moderate TBI, while mild TBI is considered separately.
Assessment of TBI outcome is usually quantified by means of the Glasgow Outcome Scale (GOS) (Table 4.2) .
Glasgow Outcome Scale (GOS)
Death within a specified time from trauma
Death within a specified time from trauma
Unawareness with only reflex responses but with periods of spontaneous eye opening
Unawareness with only reflex responses but with periods of spontaneous eye opening
Conscious but dependent for daily support from another person by reason of mental or physical disability
Low severe disability
Cannot be left alone for more than 8 h at home
Upper severe disability
Can be left alone for more than 8 h at home
Some disability (e.g., dysphasia, hemiparesis, epilepsy, deficits of memory or personality) but able to look after themselves, do shopping and travel by public transport
Low moderate disability
Unable to return to work even with special arrangement
Upper moderate disability
Able to return to work even with special arrangement
Resumption of normal life with the capacity to work (preinjury status not necessarily achieved) – neurological or psychological deficits possible
Low good recovery
With disabling deficits
Upper good recovery
Without disabling deficits
In order to simplify the outcome assessment, GOS is often dichotomized into two classes of outcome:
“Good outcome” (or “favorable outcome”) encompassing good recovery and moderate disability
“Poor outcome” (or “unfavorable outcome”) encompassing death, vegetative state, and severe disability
Although statistically inefficient , this dichotomization simplifies statistical analysis and clinical information and is the most widespread tool for TBI outcome assessment.
As an opposite trend, the Extended GOS (GOS-E: Table 4.2), with its 8-level score ranging from 1 to 8 (Dead to Upper Good Recovery), offers an improvement in sensitivity and is increasingly used in the research setting .
Patients may theoretically be scored with GOS at any time after trauma, but GOS is specifically validated on a time span of 6 months after trauma and this is currently considered the most correct timing of GOS scoring.
126.96.36.199 Moderate and Severe TBI Prognosis
Several single TBI prognostic variables have been outlined and their predictivity has been extensively scrutinized .
A major resource for a comprehensive literature review on single predictors is the Brain Trauma Foundation (BTF) guideline collection, available online and updated on a regular basis .
Past attempts to define TBI prognosis examined single variables with a univariate approach.
Recently, a new approach of multivariate logistic predictive models combining various predictors of outcome has crossed over from scientific research into the clinical setting.
This approach has radically changed the attitude towards prognostic variables.
As evidence grew that the severity of TBI outcome is actually multifactorial, attempts were made to build multivariate predictive models of outcome after TBI.
Two prediction models are particularly important with regard to this: the International Mission for Prognosis and Clinical Trials in Traumatic Brain Injury (IMPACT) prognosis calculator [6, 7] and the Corticosteroid Randomization after Significant Head Injury (CRASH) prognosis calculator . These models underwent both accurate internal calibration and extensive external validation, and also discriminate outcomes with areas under the receiver operating characteristic curve (AUROC) in the range of 0.6–0.8. In particular, the IMPACT calculator represents the most evolved prognostic tool for moderate and severe TBI available at present.
It combines covariate adjustment together with ordinal logistic regression and explores the possibility of nonlinearity for regression variables.
IMPACT is validated for patients with a GCS ≤12 on admission (after stabilization of systemic conditions) and yields the probability of death and of poor outcome 6 months after trauma, taking into account ten predictive variables recorded on admission. These variables generate three models with increasing complexity.
The core model includes age, GCS motor score, and pupil reactivity to light. It yields most of the prognostic information of IMPACT.
The extended model also includes the occurrence of hypoxia, hypotension, posttraumatic subarachnoid hemorrhage, epidural hematoma, and Marshall’s Traumatic Coma Data Bank CT scan classification.
The laboratory model also includes serum glucose and blood hemoglobin.
A major interest of the IMPACT enterprise lies in the modern methodology it exploits  in analyzing the heterogeneity among patients regarding causes, pathophysiology, treatment, and outcome.
Neither the CRASH nor the IMPACT prognosis calculator takes the treatment protocol into account. This could be a critical issue and some evidence suggests that both calculators actually overestimate the frequency of death and poor outcome in patients treated according to the Lund concept [9, 10].
A second limitation of CRASH and IMPACT is that they address predictors that were suggested by previous studies: it is possible that other relevant predictors need to be included in multivariate models.
A third limitation of these models is that they produce baseline prognosis and cannot be employed for dynamic predictions that take into account conditions and events occurring over the course of the disease process.
Mathematical interpretation of single predictive variables has to be put into perspective and no single predictor can now be considered outside the framework of the proposed multivariate logistic predictive models.
Nevertheless, an account of relevant or potential predictors of TBI prognosis is reported below.
Age is a strong independent predictive factor for poor outcome after severe TBI, with a minimum 70 % positive predictive value for poor outcome when considered in univariate analysis.
The probability of poor outcome increases with increasing age. This relationship is nonlinear and there is some debate about the relevant coefficients, especially in the pediatric age.
Some characteristics of increased age may account for this relationship, particularly the decline in baseline health conditions with a higher incidence of preexisting disease, and a general decreased capacity for functional repair of the brain. These factors have not yet been clearly identified.
However, it has been demonstrated that the age effect on outcome is not explained by the increased frequency of systemic complications or intracerebral hematomas with age.
Impairment of Consciousness
Impairment of consciousness is conceivably the most impressive effect of trauma on the brain.
The severity of this impairment intuitively marks the severity of TBI and the clinical evaluation of it provides important clues to the prognosis after TBI.
The Glasgow Coma Scale (GCS)
The GCS (Table 4.1) was originally developed  for the repeated intrahospital assessment of consciousness over time after TBI, but its use has gone far beyond this scope and it is frequently applied to evaluate impaired consciousness regardless of its causes.
This extension of the GCS beyond its original scope has not always been adequately validated but it stemmed from the good performance of GCS in assessing consciousness after TBI.
To this end, its most important result is the continuous relationship found between mortality and the score across the entire range of GCS score . It yields a more than 70 % positive predictive value for poor outcome when considered in univariate analysis.
Criticism of the GCS arose early and has increased sharply in the last few years:
The temptation to add up the three GCS items in a single “total” GCS score leads to the widespread use of a “total” GCS. This produces loss of information and is mathematically unsound, since the items differ in their respective prognostic importance and different weights should be attributed to them. In fact, different permutations of item scores with a same “total” GCS have actually different prognostic significance.
Although the “total” GCS is deeply rooted in clinical practice, GCS should always be reported with its three separate items.
Inter- and intraobserver reproducibility may be low, particularly in differentiating GCS motor score 3 and 4.
GCS does not provide a complete clinical assessment, since it does not address some important features (e.g., brain stem impairment), so it is evident that prognosis cannot rely exclusively upon GCS evaluation.
GCS is affected by several conditions other than the severity of injury. Toxic and metabolic causes of coma, suboptimal respiratory or hemodynamic conditions, sedation, and muscle paralysis lead to underestimation of GCS. These factors should be evaluated and possibly removed before scoring patients with the GCS. The very evaluation of some GCS items is impaired by several conditions, such as tracheal intubation, facial trauma, spinal injury, and preexisting disease.
Due to the aforementioned criticism about GCS, other scales gained some degree of popularity after demonstration that they are at least as predictive of TBI outcome as the GCS. They are usually designed to evaluate impaired consciousness of any cause.
Some scales are more complex than the GCS, since they aim at providing a more thorough evaluation of TBI. The FOUR (Full Outline of UnResponsiveness) scale is a typical example : it consists of four components (eye, motor, brainstem, and respiration evaluation).
Other scales are simpler than the GCS and aim at providing easier-to-use instruments that avoid some of the GCS pitfalls. Distinctive examples of these are the AVPU scale , the ACDU scale , and the Simplified Motor Scale  (Table 4.3).
Three commonly proposed simpler alternatives to the Glasgow Coma Scale (GCS)
Simplified Motor Scale
Withdrawal to pain or less response
Responds to vocal stimuli
Responds to painful stimuli
Unresponsive to any stimulus
Although theoretically interesting, these scales are not extensively used.
Damage to the midbrain third nucleus or to the efferent third cranial nerve by temporal lobe compression produces “fixed” pupils, i.e., the pupillary light reflex is absent (pupillary diameter reduces <1 mm after exposure to bright light). Usually pupils are also dilated (>4 mm).
A bilaterally absent pupillary light reflex is a strong predictor of poor outcome, with a more than 70 % positive predictive value of poor outcome in univariate analysis.
Pupil evaluation has good intra- and interobserver reliability, but direct third nerve injury can also lead to dilated or fixed pupils and confound neurological evaluation.
Moreover, suboptimal respiratory and hemodynamic conditions may affect pupil evaluation and hypotensive or hypoxemic patients may even exhibit fixed dilated pupils that return to normal after proper resuscitation.
Other pupillary signs, namely asymmetry in pupillary size (>1 mm) or light reflex, are not independent predictors of outcome, although they are very important in guiding therapy.
Arterial Hypotension and Hypoxia
The occurrence of a systolic blood pressure of less than 90 mmHg has a 67 % positive predictive value for poor outcome, and increases to 79 % if hypotension is associated with hypoxia, when analyzed with univariate analysis.
This variable has been studied particularly during the prehospital and early phases after TBI.
In fact, hypotension is a typical example of the usefulness of nonlinear regression evaluation employed in modern multivariate prognostic models. The relationship between systolic blood pressure and outcome is U-shaped, so that very low and very high systolic blood pressure on admission yield a similar probability of poor outcome.
The effect of hypotension and hypoxia on the outcome of severe TBI is unique, since they are causes of poor outcome and not merely severity indicators. This makes them particularly interesting for two reasons.
From a pragmatic point of view, hypotension and hypoxia are strong predictors of outcome that are treatable and sometimes preventable.
From a theoretical point of view, their importance is related to the concept of secondary brain injury.
The occurrence of any TBI-related abnormality on CT scan is predictive of poor outcome.
Significant posttraumatic brain lesions may develop over time and early CT scans may be unreliable. CT scans obtained less than 6 h after trauma mandate a new CT examination after 6–8 h as 40 % of patients with a negative CT scan at admission develop CT abnormalities thereafter.
Various classifications of post-TBI CT scan imaging have been proposed, but Marshall’s Traumatic Coma Data Bank CT scan classification is the most widely used due to its prognostic value (Table 4.4).
Marshall’s Traumatic Coma Data Bank classification of post-TBI CT scan imaging
Normal CT scan for age and preinjury status
0–5 mm midline shift present
No high- or mixed-density lesion > 25 mL (May include bone fragments and foreign bodies)
Cisterns compressed or absent
0–5 mm midline shift
No high- or mixed-density lesion >25 mL
Midline shift >5 mm
No high- or mixed-density lesion >25 mL
Evacuated mass lesion
Any lesion surgically evacuated
Nonevacuated mass lesion
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical TreeGet Clinical Tree app for offline access