Traumatic brain injury

Epidemiology
Undocumented immigrants are more likely to suffer TBI from assault than from a fall.

Pathophysiology
Early ischemia within the first 24 hours occurs in two-thirds of TBI patients. From days 1-3 there is a period of hyperemia, and vasospasm after day 3 that again may lead to ischemia. Ischemia and hyperemia can coexist, showing abnormalities in flow-metabolism coupling.

Marshall CT classification (1992)
This is a standard terminology initially published in 1992 to classify various degrees of mathological injury in TBI. However, analysis suggested it only had fair discriminative ability for predicting 6-month mortality, with AUROC 0.67 when applied to a database from clinical trials of 2,269 patients.

Rotterdam CT classification (2005)
The authors used a database of patients from trials of tirilazad to develop predictive models for mortality. The authors chose to add 1 point to all cases to make the grading numerically consistent with the motor score of the GCS and the 6 categories of the Marshall CT classification. This model had AUROC of 0.75 for prediction of 6-month mortality.

Pulse amplitude of ICP (AMP)
A loss of tension in the walls of cerebral arteries may increase the passive amount of blood pulsatility translating to the ICP waveform. The pulse amplitude, termed the AMP, is calculated as the difference between the beginning of diastolic and systolic pressure, or the amplitude of the first spectral harmonic component of the ICP waveform. The measure is less susceptible to noise and drift than the ICP waveform itself. A loss of vascular reactivity may be associated with a large ABP.

Brain tissue oximetry (PbtO2)
One study suggested that there was not really any correlation between brain tissue oximetry and volume of ischemic brain after TBI. This may be due to PbtO2 only being measured in one area of the brain.

Cerebral autoregulation
Autoregulation can be intact, partially impaired, or completely impaired in TBI patients, and this can change in an individual patient over their course.

Mean velocity index (Mx) and Systolic velocity index (Sx)
These indices represent correlation coefficients for the relationship between CPP and TCD measures of velocity, including mean velocity (Mx) and systolic velocity (Sx). In normal autoregulation they have a negative correlation (rising CPP leads to distal vasoconstriction which decreased large vessel flow velocity), while in impaired autoregulation they have a positive correlation. In a study of 82 patients with TBI, Mx became statistically positive for CPP < 55 mmHg on average, and Sx became statistically non-negative for CPP < 40 mmHg on average. These are the thresholds, on average, in these patients where they could no longer distally vasodilate in response to falls in CPP. Impairments in cerebral autoregulation (Mx or Sx became positive) were associated with unfavorable outcomes (GOS 1-3). . The difficulty with these indices is that they require constant focusing of transcranial doppler probes, which can limit their ability for long-term monitoring.

Pressure reactivity (PR)
This measure is calculated as the slope of the regression line relating MAP to ICP for an individual patient over one hour. One study of TBI patients showed that a value of ≥0.13 suggests pressure-passive (i.e. loss of autoregulation) and <0.13 suggests pressure active patients. In pressure active patients (autoregulation intact), targeting CPP of 70 mmHg led to better outcomes, while in pressure passive patients targeting ICP <20 mmHg led to better outcomes.

Pressure reactivity index (PRx) and variations (PRx55-15, L-PRx, LAx)
While pressure reactivity measures correlations over a 1 hour period, PRx is a moving correlation coefficient between 40 consecutive 5-second averages of ICP and mean arterial pressure (MAP), moving in a 5 second window. This measure reflects the ability of brain arteriole smooth muscle to react to changes in transmural pressure. A positive value means a positive slope (y-axis MAP, x-axis ICP) indicating loss of autoregulation, while negative reflects intact autoregulation. This is because when autoregulation is intact a rise in MAP will lead to cerebral vasoconstriction, which will lower cerebral blood volume and thus lower the amount of overall intracranial volume which lowers ICP. In a study of 82 TBI patients, a value < -0.2 correlated strongly with favorable outcome, while PRx >0.2 correlated strongly with unfavorable outcomes. In a study of 459 patients, a PRx threshold of >0.25 correlated strongly with mortality, while a PRx threshold of >0.05 correlated strongly with unfavorable outcome (GOS 1-3). . The percentage of time with PRx > 0.25 is associated with progression of pericontusional cerebral edema.

An optimal CPP can be defined as the point where PRx reaches its minimum value when plotted against CPP. Patients with a mean CPP close to optimal have been shown to be more likely to have a favorable outcome after TBI. Raising CPP higher than optimal does not yield improvement in brain tissue oxygenation (PBtO2), while having CPP lower than optimal was associated with lower PBtO2.

This may be more predictive of outcome when a filter is used to limit the analysis to oscillations with periods of 15-55 seconds, which has been dubbed PRx55-15. Any of these methods, however, requires special software to calculate these moving averages. Using a time window of 20 minutes, researchers created the low frequency PRx (L-PRx), which was shown in one study to correlate as well with outcome as PRx in patients with intracerebral hemorrhage (not TBI) and to improve the rate of identification of an optimal CPP range. In predicting mortality, one study of 307 patients showed a higher AUROC for the PRx itself compared with the L-PRx (0.61 vs 0.54, p=0.011).

Another low-frequency autoregulation index analyzes data in 1 minute averages and allowed for defining an optimal CPP range in 97% of cases. A low-frequency PRx has also been calculated using 6s intervals for measurements in 6 hr windows with an optimal CPP range being definable in 91% of cases.

Despite numerous positive studies, most are retrospective and there are limitations in all of these cohorts. As such, a meta-analysis in 2017 concluded that there was insufficient high-quality evidence to implement a CPP optimization strategy based on PRx compared with the usual CPP targets as advocated by the Brain Trauma Foundation. This measure may be more valid than PRx at levels of ICP <15 mmHg.

In TBI patients, the percentage of time with PAx > 0.25 is associated with progression of pericontusional cerebral edema.

Correlation of CPP and AMP (RAC)
A retrospective study of 358 TBI patients evaluated this correlation. Similar to PAx, the correlation should usually be negative but rises in the setting of severe ICP elevations and impaired cerebrovascular reactivity.

Correlation of AMP and ICP (RAP)
This measure is normally small and positive but rises with ICP until ICP reaches a 50-60 mmHg threshold at which point the RAP rapidly declines to zero and vascular transmural pressure approaches the critical closing pressure of the cerebral vessels.

Glycerol
Elevated cerebral glycerol levels (mean or max) correlate with worsened autoregulation as measured by pressure reactivity. This correlation is tempered or alleviated entirely with prostacyclin infusion, which may affect an imbalance in thromboxane-prostacyclin towards thromboxane occurring in TBI patients.

Jugular bulb oximetry
One study suggested a limited correlation between jugular bulb oximetry and volume of ischemic brain after TBI. This may be due to jugular bulb oximetry reflecting a very global measure, while there is considerable tissue heterogeneity in ischemic response after TBI.

Transcranial doppler (TCDs)
In patients with TBI, early TCD measurements with low cerebral blood flow (CBF) defined as mean velocity <40 cm/s had 100% positive predictive value for brain tissue hypoxia with PbtO2 ≤ 20 mmHg.

Anti-fibrinolytic therapy
TXA (1g IV over 10 minutes, then 1g IV over 8 hours) is likely of benefit in reducing head-injury related mortality in patients with TBI when administered within 3 hours in patients with GCS 9-15. In patients with GCS of ≤8 the benefit is less certain, but there is little evidence of harm. Overall NNT based on the post-CRASH-3 meta-analysis is ~63. The large CRASH-3 trial as well as several additional studies suggest a modest head injury-related mortality benefit for administration of TXA in patients with TBI. Click here for more details.

Blood pressure management
ACE-inhibitor use prior to TBI is associated with higher higher mortality after severe TBI (aOR 3.66, 95% CI 1.43-9.39, n=600), but this association was partially dampened by prior beta-blocker administration.

Cerebral autoregulation impairment
In one small study, prostacyclin seemed to limit impairment in autoregulation (as measured by PR, pressure reactivity). Hyperglycemia seems to worsen it.

Corticosteroids
High-dose corticosteroids should not be used in TBI, as the MRC CRASH trial showed an increase in mortality with their use.

Corticosteroid Randomization After Significant Head Injury (MRC CRASH, 2004, 2005)
This trial randomized 10,008 patients with TBI of all severities (GCS 3-14) to receive methylprednisolone (2 g IV over 1 hour, then 0.4 g per hour for 48 hours, vs. placebo. They found increased early mortality (within 2 weeks) in the corticosteroid group (21.1% vs 17.9%, RR 1.18, 99% CI 1.09-1.27, p=0.0001). No severity classes of TBI benefited from steroids.  The authors also performed a meta-analysis of MRC CRASH with prior studies and showed a RR of death of 1.12 (1.05-1.20). In 2005 they published their 6 month outcomes and showed a similar increased risk of death at 6 months in the corticosteroid group compared with placebo (25.7% vs 22.3%, RR 1.15, 99% CI 1.07-1.24, p=0.0001).

Osmotherapy
In pediatric patients, a 2019 systematic review suggests that both hypertonic saline and mannitol are effective to lower intracranial pressure and improve clinical outcomes, but data is not high quality.

Corticosteroids
In general corticosteroids should not be used for treatment of edema in moderate and severe TBI. This is a Neurocritical Care Society Clinical Performance Measure.

Cortical spreading depolarization
Cortical spreading depolarizations are a sustained collapse of electrochemical membrane gradients that propagate at 1 to 9 mm/min through gray matter affecting neurons and astrocytes on mass. They cause loss of electrical signaling, intracellular calcium influx, associated release of neurotransmitters with resultant excitotoxic injury, and cerebral edema.. In one study of 138 patients, 60% developed cortical spreading depolarization, and 37% of the 138 patients had clusters of spreading depolarization. If they occurred sporadically they were not associated with poor outcome, but if they occurred in clusters or isoelectric spreading depolarizations they correlated with worsened outcomes (proportional OR 2.29, 95% CI 1.13-4.65).

Early
The incidence of post-traumatic seizures varies depending on the studies from 0.4-10.5%. Many if these seizures occur immediately (8.9% incidence of immediate or impact post-traumatic seizures), but between days 1-7 the incidence was 1.9%. Many seizures, however, are nonconvulsive, with one study showing that a total of 22% of TBI patients had early seizures, but more than half had only nonconvulsive seizures that could not be detected without cEEG.

Their occurrence of early post-traumatic seizures is associated with longer length of stay, an increased risk of 24-month mortality (RR 2.14, 95% CI 1.32-3.46), an increased risk of 24 month poor neurological outcomes (RR 2.10 for severe disability, RR 3.97 for vegetative state) and development of post-traumatic epilepsy (RR 2.91, 95% CI 2.22-3.81).

Cardiac enzyme elevation
Cardiac enzymes are frequently elevated in TBI patients, including CK-MB and troponin, as well BNP levels. These have been shown to remain elevated up to to 8 days in a pig fluid-percussion injury model. Phenylephrine may increase enzyme release in males but may decrease enzyme release in females.

Hypotension
One study in pigs suggested that dopamine may be better at protecting cerebral autoregulation than phenylephrine.

Myocardial stunning
Traumatic brain injury is associated with acute cardiac dysfunction due to myocardial stunning, which may be one mechanism for neurogenic pulmonary edema. Patterns of injury seen on TTE include global hypokinesia, inverse takotsubo pattern (akinesia of basal portions with hyperkinesia of the apex), and regional wall motion abnormalities. Myocardial stunning is associated with increased in-hospital mortality (aOR 9.6, 95% CI 2.3-40.2). The association with overall TBI severity was unclear in one study.

Risk factors for myocardial stunning after TBI include: EKG findings in myocardial stunning after TBI include:
 * Age ≥ 65 (aOR 1.05, 95%CI 1.02-1.08)
 * Head Abbreviated Injury Scale 4-6 (reflects severe to critical to untreatable brain injury) (aOR 3.34, 95% CI 1.19-9.34)
 * Elevated CK-MB (aOR 3.63, 95% CI 1.28-10.31)
 * Elevated tropI (aOR 6.21, 95% CI 2.02-19.09)
 * Sinus tachycardia
 * Prolonged QTc
 * Morphological end-repolarization abnormalities

Pulmonary
In mouse models, TBI causes monocyte functional impairments that may alter the ability to combat respiratory infections. Mortality is increased in mice with TBI with early or late bacterial lung infection.

Early tracheostomy
In a meta-analysis of eight studies, early trachestomy (average 6 days after onset) vs. late tracheostomy (average 12 days after onset) was associated with shorter duration of mechanical ventilation (-4 days), ICU length of stay (-6 days), and hospital length of stay (-7 days). It also produced a lower risk of ventilator-associated pneumonia (risk difference 0.78, 95% CI 0.70-0.88) but had no effect on mortality.

Ventilator-associated pneumonia (VAP)
A meta-analysis found an incidence of 36% (95% CI 31-41%).

Risk factors include: VAP was associated with prolonged mechanical ventilation time (5.45 days longer, 95% CI 3.78-7.12), prolonged ICU length of stay (6.85 days longer, 95% CI 4.90-8.79), and prolonged hospital length of stay (10.92 days longer, 95% CI 9.12-12.72). However there was no significant difference in mortality.
 * Smoking (OR 2.13, 95% CI 1.16-3.92)
 * Tracheostomy (OR 9.55, 95% CI 3.24-28.17)
 * Blood transfusion on admission (OR 2.54, 95% CI 1.24-5.18)
 * Barbiturate infusion (OR 3.52, 95% CI 1.68-7.40)
 * Higher injury severity score (OR 4.65, 95% CI 1.96-7.34)
 * Higher head abbreviated injury scale (OR 2.99, 95% CI 1.66-5.37)

Hyperglycemia
Average daily blood glucose correlates with worsened autoregulation as measured by PRx55-15 values, and higher lactate-pyruvate ratio on cerebral microdialysis, without a clear effect on cerebral glucose concentration.

Patient factors associated with worsened mortality

 * ACE inhibitors (prior administration)
 * Age (older)
 * APACHE II score (worse)
 * Cerebral autoregulation impairment
 * Extracranial injury (major)
 * Female sex
 * Glasgow Coma Scale (worse)
 * Motor score specifically
 * Hypotension
 * Hypoxia
 * Injury Severity Score (higher)
 * Intracranial pressure (elevated)
 * Pupil reactivity (absent of one or both pupils)

Imaging factors associated with worsened mortality

 * Basal cisterns or ventricular compression
 * Epidural hematoma (ABSENCE OF)
 * Hematoma (non-evacuated)
 * Marshall CT classification
 * Midline shift
 * Petechial hemorrhage
 * Rotterdam CT classification
 * Traumatic SAH (tSAH)

Lab values associated with worsened mortality

 * Serum glucose (elevated)
 * Serum hemoglobin (lower)
 * Serum melatonin (higher)
 * Serum oxidized guanine species (higher). See full article on Pathophysiology of secondary brain injury in TBI - Oxidative injury.

Patient factors associated with worsened outcome

 * Age (older)
 * Blood pressure (lower)
 * Cerebral autoregulation impairment
 * Extracranial injury (major)
 * Glasgow Coma Scale (worse)
 * Motor score specifically
 * Hypotension
 * Hypoxia
 * Intracranial pressure (elevated)
 * Pupil reactivity (absent of one or both pupils)

Imaging factors associated with worsened outcome

 * Basal cisterns or ventricular compression
 * Epidural hematoma (ABSENCE OF)
 * Hematoma (non-evacuated)
 * Marshall CT classification
 * Midline shift
 * Petechial hemorrhage
 * Traumatic SAH (tSAH)

Lab values associated with worsened outcome

 * Serum glucose (elevated)
 * Serum hemoglobin (lower)

Marshall CT classification (1992)
This descriptive model is described above, and shows an AUROC of 0.67 for prediction of 6-month mortality.

Rotterdam CT Classification (2005)
This model is described above, and shows an AUROC of 0.75 for prediction of 6-month mortality.

CRASH Head Injury Prognosis Model (2008) (calculator available here)
This score was developed from the 10,008 cohort from the MRC CRASH trial (corticosteroids in TBI) and analyzed various aspects to develop a multivariable regression model to predict mortality and prognosis after TBI. Important to note is that these patients were originally enrolled in the trial from 1999-2004. The model worked well in both high-income and low-income countries. For mortality the AUROC was 0.84-0.86, and for death or severe disability (GOS 1-3) the AUROC was 0.81-0.84. Including CT parameters, for mortality the AUROC was 0.84-0.88, and for GOS 1-3 was 0.83-0.84. They subsequently validated this in the IMPACT database of 8,509 patients and showed good predictive value with AUROC 0.77 for both models, but the calibration performed better with the model including CT parameters. The included clinical variables are age, GCS, pupil reactivity, major extracranial injury, and the included CT variables are petechial hemorrhages, obliteration of the 3rd ventricle or basal cisterns, subarachnoid hemorrhage, midline shift, and non-evacuated hematoma.

As this is a logistic regression model, it is not simple to calculate manually, but a web-based calculator is available here.

Adding physiological parameters such as MAP, PRx, PAx, and RAC may improve the prediction of the CRASH model for unfavorable outcome (GOS 1-3).

IMPACT Prognostic Calculator (2008) (calculator available here)
This score was developed from the IMPACT database, which at the time included data from eight randomized trials and three observational studies conducted from 1984-1997, and included 8,509 patients over age 14 years. They developed a model using only clinical characteristics, another that also included radiographic characteristics, and another that also included both radiographic characteristics and laboratory values. AUROC for prediction of mortality in various subpopulations ranged from 0.70-0.84 for the clinical model, 0.71-0.87 for the model also including imaging, and 0.72-0.80 for the model also including both imaging lab values. For predicting unfavorable outcome (GOS 1-3), the AUROC was 0.70-0.82 for the clinical model, 0.73-0.84 for the model also including imaging, and 0.75-0.82 for the model also including both imaging and lab values. This study was externally validated in the MRC CRASH trial, in which for mortality the AUROC was 0.78 for the clinical model and 0.80 for the model including imaging. For predicting GOS 1-3, the AUROC was 0.78 for the clinical model and 0.80 for the model including imaging. The included clinical variables are age, GCS motor score, pupillary reactivity, hypoxia, and hypotension.. The radiographic variables are Marshall CT classification, tSAH, and epidural hematoma, and the lab values are glucose and hemoglobin.

As this is a logistic regression model, it is not simple to calculate manually, but a web-based calculator is available here.