Introduction

Traumatic brain injury (TBI) is a common, burdensome health condition disproportionately affecting young adults. In 2010, the CDC estimated that 1.7 million Americans sustain a TBI annually, with TBI contributing to as much as one-third of all injury-related US deaths [1]. In the USA, falls are the leading cause of TBI-related emergency department (ED) visits and hospitalizations, with over half a million visits and 60,000 hospitalizations per year [1]. Motor vehicle accidents are responsible for most of the TBI-related deaths among young adults [1]. Similarly, in European countries, falls and motor vehicle accidents are the first and second most prevalent causes of TBI [2]; the inverse is true in Asia, with motor vehicle accidents leading over falls [3].

Classically, clinical classification of TBI severity has been based on a patient’s level of consciousness, as assessed by motor, verbal, and eye-opening examination findings that comprise the ubiquitous 15-point Glasgow Coma Scale (GCS) [4]. The American Congress of Rehabilitation Medicine defined mild TBI as GCS ≥ 13 and moderate TBI as a GCS of 9–12 [5]. The contemporary clinical definition of severe TBI (sTBI), adopted from the 1991 Traumatic Coma Data Bank [6], is GCS ≤ 8 after resuscitation, within 48 h of injury. Limitations of using GCS alone to define TBI severity include possible confounding by drug intoxication, sedation, and intubation, as well as the inability of the GCS to account for the pathophysiologic heterogeneity of TBI. To address these limitations, the Department of Veterans Affairs has released a clinical practice guideline to further define severity categories of TBI based on criteria including structural imaging, duration of loss/alteration of consciousness, and post-traumatic amnesia, in addition to GCS [7].

Two major radiographic scales have been developed to classify TBI based on morphologic characteristics on head computed tomography (CT). Most notable is the Marshall Classification [8], which subcategorizes injury based on the presence of a mass lesion with or without evacuation, basal cistern effacement, and midline shift. The Rotterdam Score also uses basal cistern compression and midline shift in its classification scheme, but with regard to mass lesions only incorporates epidural hematomas into scoring and also includes intraventricular and/or subarachnoid blood as a variable as well [9].

TBI has been defined as “an alteration in brain function, or other evidence of brain pathology, caused by external force [10].” While injury mechanisms are heterogeneous, the pathophysiology of sTBI involves both primary injury (occurring at impact) and secondary injury (occurring after impact). Primary brain injury includes traumatic shearing and tearing of axons, leading to diffuse axonal injury, and focal injuries, such as intra- and extra-axial hematomas. Secondary brain injury can result from ischemia, cerebral edema, seizures, and oxidative stress, leading to subsequent neuronal, axonal, and glial injury [11]. The aim of intensive care unit (ICU) management of the sTBI patient is to minimize and mitigate all secondary brain injury. To achieve this, an intensivist must balance multiple considerations for an sTBI patient, including the possibilities of increased intracranial pressure (ICP), decreased cerebral perfusion pressure (CPP), hypoxemia, hypotension and hypertension, impaired cerebral autoregulation, a heightened systemic inflammatory response, and both convulsive and non-convulsive seizures [12]. Close to 90% of sTBI patients also suffer from non-neurologic injuries and organ dysfunction that can influence treatment goals and strategies in the ICU [13].

In an effort to standardize care and guide treatment, the Brain Trauma Foundation (BTF; https://www.braintrauma.org) has created evidence-based guidelines for the management of severe TBI. This review focuses on aspects of practical management of sTBI for ICU clinicians and highlights recommendation updates contained in most recent BTF guidelines, now in their fourth iteration [14]. A notable change to this edition was a more stringent approach to the level of evidence necessary for a recommendation to be made, an approach which resulted in downgrading and even omitting some prior recommendations. Moving forward, the BTF is adopting a “living guideline” model, with continual review of the literature and more frequent updates to recommendations.

Prehospital Management

The BTF’s most recently published guidelines for prehospital management of TBI were updated in 2007 [15]. Recommendations include assessments of GCS and pupillary size in the field, targets of systolic blood pressure (SBP) >90 mmHg and oxygen saturation (SpO2) >90%, and rapid triage to a trauma center (Fig. 1).

Fig. 1
figure 1

Initial severe traumatic brain injury management based in part on Brain Trauma Foundation guidelines. ATLS Advanced Trauma Life Support, BPM breaths per minute, CBC complete blood count, CT computed tomography, ETCO 2 end-tidal CO2, GCS Glasgow Coma Scale, ICP intracranial pressure, O 2 oxygen, PT prothrombin time, PTT partial thromboplastin time, SBP systolic blood pressure

Full size image

Airway Management and Blood Pressure

The BTF prehospital guidelines call for establishment of a secure airway “by the most appropriate means available” for patients with a GCS < 9, an inability to protect their airway, and/or an SpO2 < 90% that is not correctable by supplemental oxygen. Of note, studies examining the impact of prehospital intubation of TBI patients on outcomes have yielded mixed results, with uncertainty regarding “appropriate means.” In a retrospective study of over 1000 TBI patients, 75% of patients with GCS < 9 who underwent prehospital intubation survived, compared to 64% of those who did not [16]. In a prospective randomized controlled trial of 312 sTBI patients, prehospital rapid sequence intubation by paramedics resulted in more favorable neurologic outcomes at 6 months compared to intubation upon hospital arrival [17]. Conversely, in a systematic review of over 15,000 TBI patients, no evidence was found in support of prehospital intubation [18]. An expert panel commenting on conflicting findings from prehospital intubation studies noted the inconclusive evidence for rapid sequence intubation, in particular [19]. These inconsistent findings may be attributable to the limitations of using the GCS alone to identify intubation candidates, variability in intubation protocols and medications administered, and suboptimal intubation and ventilation techniques with subsequent hypo- or hyperventilation.

With regard to prehospital blood pressure goals, a prospective study of 717 patients from the Traumatic Coma Databank suggested that SBP < 90 mmHg is an independent predictor of mortality, with even a single episode of hypotension from the time of injury to resuscitation doubling mortality and increasing morbidity [20]. The BTF prehospital guidelines recommend treating hypotensive patients with isotonic fluids, with the option of hypertonic fluids for patients with GCS < 8 [15]. However, it should be noted that a more recent multicenter cohort study analysis incorporating 5057 patients with significant TBI has suggested that having an admission SBP even below 120 mmHg may increase mortality rates [21].

Triage

The BTF prehospital guidelines recommend that sTBI patients be transported to a hospital with CT scanning capability and neurosurgical care with ICP monitoring available [15]. In a retrospective study of over 5000 TBI patients, after adjustment for the case mix, risk of death was significantly lower for patients cared for in a major trauma center [22]. Moreover, several studies have observed that care led by dedicated neurointensivists improves outcomes for the general population of brain-injured patients [23, 24].

In an analysis of over 51,000 patients from the National Trauma Data Bank National Sample Program from 2007 to 2009, patients with sTBI who were transferred to a Level I or II trauma center had lower injury severity and lower adjusted mortality rates compared to those who were directly admitted to higher-level trauma centers [25], a finding suggesting that clinical severity has indeed been influencing prehospital triage decisions.

Impact of Guideline Adherence on Outcome

While the evidence cited in the BTF prehospital guidelines is limited in by the inclusion of cohort, case control, and database registry studies, guideline adherence has been correlated with higher quality of care and improved outcomes. In a before–after study testing BTF guideline education among emergency providers, data from over 1000 patients revealed that patient rates of hypoxia, hypotension, and mortality were lower in the post-training group and patients treated post-training had improved 14-day Glasgow Outcome Scores (GOS) [26].

Emergency Department Considerations

In addition to summarizing prehospital management principles, Fig. 1 outlines the standard ED management for sTBI patients. In addition to these general principles, there are special considerations for the prevention of intracranial hemorrhage expansion among sTBI patients.

Tranexamic Acid

Tranexamic acid (TXA) is an antifibrinolytic that has been studied in major trauma to prevent excessive blood loss, weighed against the risk of vascular occlusive events such as myocardial infarction, stroke, pulmonary embolism, and deep vein thrombosis. The CRASH-2 trial [27] concluded that all-cause mortality was reduced in trauma patients who received TXA, without an increased risk of vascular occlusive events. The nested CRASH-2 intracranial bleeding study suggested that “neither moderate benefits nor moderate harmful effects [of TXA in TBI] can be excluded [28].” Another randomized controlled trial did not find a benefit of TXA in reducing intracranial hemorrhage expansion [29]. In one systematic review and meta-analysis of TBI patients who received TXA, pooled results demonstrated a statistically significant reduction in intracranial hemorrhage progression and a trend toward improved clinical outcome [30]. Finally, a Cochrane review concluded that TXA reduces mortality in trauma patients, but its use later than 3 h post-injury may be harmful, with its efficacy in TBI remaining uncertain [31]. The CRASH-3 trial, still underway, plans to enroll 10,000 adult sTBI patients in order to assess the effects of TXA on in-hospital mortality [32].

Coagulopathy Management

When coagulopathy is present, rapid reversal should be a high priority in the ED to prevent hemorrhage expansion leading to secondary injury. In a review of 253 TBI patients requiring serial head imaging, 85% of patients who had at least one abnormal coagulation study on admission developed new or worsening lesions (swelling, intracranial hemorrhage, infarction) on follow-up CT scans within 72 h of admission, compared to 31% of patients with a normal coagulation profile on admission [33]. Another study found progression of traumatic intracranial hemorrhage (i.e., intraparenchymal, epidural, subdural, subarachnoid) in 80% of coagulopathic sTBI patients (INR < 1.3, PTT > 35, platelets <100,000) compared to 36% of non-coagulopathic sTBI patients, with the intracerebral hemorrhage (ICH) progression group having a fivefold increase in mortality [34]. New oral anticoagulants have made the reversal of coagulopathy more complex. While dabigatran, a direct thrombin inhibitor, now has an available antidote [35], direct factor Xa inhibitors still present a challenge. Of note, in a retrospective study of 18 TBI patients with either ICH or subarachnoid hemorrhage who were taking either rivaroxaban or apixaban, 4-factor prothrombin complex concentrate was safely used to potentially reduce hematoma expansion [36]. The Neurocritical Care Society has recently released guidelines for the rapid reversal of antithrombotics in ICH [37], the principles of which may serve as a guide for sTBI management as well.

Neuromonitoring in the Intensive Care Unit

Once an sTBI patient has been stabilized and transferred from the ED to the ICU, various forms of neuromonitoring may inform treatment thresholds to mitigate secondary injury. The BTF Guidelines for the Management of sTBI mention that a multimodal monitoring approach, in addition to traditional ICP and CPP monitoring, may influence medical decision-making in the ICU; however, no firm recommendation is ultimately made regarding this approach. Of note, in a consensus statement on neuromonitoring, the Neurocritical Care Society and the European Society of Intensive Care Medicine assert that the use of multiple monitoring techniques may help supplement the clinical examination, especially in those patients whose examination is confounded by medication or who are in a comatose state [38].

Utility of Intracranial Pressure and Cerebral Perfusion Pressure Monitoring

The 2007 BTF Guidelines for Management of sTBI recommended an ICP monitor to be placed specifically in those sTBI patients considered salvageable with an abnormal CT and in those with a normal CT and 2 or more of the following: age >40, unilateral or bilateral posturing, and SBP < 90 [39]. While the new guidelines state that management of sTBI patients using ICP monitoring information is recommended, the specific criteria on exactly which sTBI patients to monitor were not formally carried over, on the basis that their supporting evidence did not meet “current standards.” However, these criteria were explicitly restated in the new publication by the authors to “maintain sufficient recognition” of those patients at risk of elevated ICP. While there is otherwise no new guidance on exactly which population of sTBI patients to monitor in the 2016 guidelines, several studies released between the two most recent editions of the guidelines warrant attention, as assumptions regarding traditional ICP-driven management protocols and patient outcomes have been debated [40].

The BEST:TRIP randomized controlled trial compared a treatment protocol for sTBI patients where ICP monitoring was used versus a protocol in which treatment was based solely on clinical examination and imaging. A composite of survival time, follow-up level of consciousness, and follow-up functional status was the primary outcome [41]. The treatment protocol that included an ICP monitor, with the goal of maintaining ICP either below or at 20 mmHg, was not found to be superior to the protocol reliant on imaging and clinical examination.

Several key limitations of BEST:TRIP warrant mention. A common criticism of this study has been the use of 20 mmHg in the ICP monitoring arm as a strict threshold for escalating management, without in-depth consideration of ICP trends among monitored patients. Furthermore, the protocol called for a strict sequence of medical therapy escalation, with intermittent cerebrospinal fluid (CSF) drainage for those in the monitoring group with extraventricular drains (EVDs) only allowed for short periods of time. The study was conducted in Ecuador and Bolivia, where fewer sTBI patients may survive to hospitalization. The overall patient cohort in BEST:TRIP may thus be less severely injured than a similar patient cohort in a country with more resources. Furthermore, none of the study participants received rehabilitation after hospital discharge, which may have caused any protocol-related effects on outcome to be more difficult to detect.

In contrast to BEST:TRIP, evidence in support of an ICP-monitoring-driven treatment protocol being associated with improved outcomes has recently been published. In a retrospective analysis of 2134 patients with TBI who were tracked as part of a New York state initiative to assess compliance with BTF guidelines, those with sTBI whose treatment involved ICP monitoring had significantly lower mortality compared to unmonitored sTBI patients [42]. Furthermore, an observational study that extracted data from the American College of Surgeons Trauma Quality Improvement Program suggested that hospitals with higher rates of ICP monitoring had significantly lower mortality rates [43]. Given this mixed evidence, the new BTF continues to support the use of ICP monitoring information in the management of sTBI patients in the ICU but backs away from formal criteria on exactly which sTBI patients should have a monitor placed.

In addition to monitoring ICP, CPP monitoring may provide complementary data to be used for clinical decision-making. One study found significantly decreased mortality associated with a guideline-driven protocol based upon ICP and CPP monitoring data [44]. Based on this study, the new BTF guidelines recommend CPP-based management if an ICP monitor is placed, suggesting that this approach likely decreases mortality.

Mode of ICP and CPP Monitoring

The 2016 BTF guidelines do not specify a preferred mode of ICP and CPP monitoring—e.g., an EVD, an intraparenchymal fiberoptic pressure monitor, or other method. In a retrospective study, EVD use was associated with higher in-hospital mortality for those patients with GCS ≥ 6, but comatose patients with a lower GCS had a trend toward lower mortality with EVD use [45]. Furthermore, in a separate retrospective study of 62 patients, those patients managed with continual CSF drainage had more effective ICP control as compared to those sTBI patients with intermittent drainage [46]. Based on this evidence, the new guidelines recommend that CSF drainage to lower ICP could be used in patients with a low GCS (<6) within 12 h of injury and that an EVD may be more effective at lowering ICP burden if it drains continually as opposed to intermittently. These new recommendations are both based on Level III evidence.

Treatment Thresholds

Whether the care of patients with sTBI should focus primarily on treatment of ICP or optimization of CPP has been debated in the literature [47, 48], with some arguing that priorities should be assigned based on the cerebral autoregulatory status of the patient [49]. As a practical approach, the new guidelines simply discuss treatment thresholds for both parameters.

If an ICP monitor is placed, the BTF guidelines recommend treatment of ICP sustained above 22 mmHg [14]. This recommendation was mainly derived from a single retrospective analysis of a database with 459 TBI patients, with the aim at identifying proper management thresholds for ICP, as well as CPP and pressure reactivity index (PRx) [50]. Of note, PRx is an index derived from changes in ICP in response to arterial blood pressure over time and is meant to approximate cerebral autoregulation; higher values suggest impaired autoregulation [51]. Analysis of the sTBI patients in the aforementioned database revealed reduced mortality associated with an ICP treatment threshold of ICP > 22, a CPP maintenance threshold of >60, and PRx maintenance threshold of <0.3 [50]. Of note, another study of 327 patients with sTBI suggested that PRx values can help guide proper individualized ICP thresholds and that these individualized ICP thresholds may be stronger predictors of death as compared to universally accepted thresholds [52].

While optimizing capillary hydrostatic pressure via blood pressure control has long been suggested as a theoretical mechanism for reducing cerebral edema in sTBI [53, 54], the relationship of cerebral autoregulation and precise CPP thresholds on outcomes in practice has now been further studied [55]. In one study of 58 TBI patients (90% of whom had sTBI), CPP > 70 was indeed associated with unfavorable outcomes among patients thought to have impaired cerebral autoregulation based on analysis of ICP and MAP data; interestingly, CPP < 60 was associated with favorable outcomes in these same impaired patients [55]. As a result of these limited data, the BTF now gives Level IIB recommendations to maintain CPPs between 60 and 70, but with a note that the minimum threshold may depend on the autoregulatory function of the patient. The guidelines maintain prior Level III recommendations that aggressive treatment to maintain CPP > 70 should be avoided.

Advanced Neuromonitoring

As mentioned earlier, the BTF provides no new recommendations on multimodal monitoring, which includes extracellular microdialysis and brain tissue oxygenation (PbtO2). The low-level recommendation that use of information from jugular bulb monitoring of arteriovenous oxygen content difference may be considered to improve mortality and outcomes has been carried over to the new BTF guidelines from the prior edition. However, research is underway to address the utility of these advanced multimodal neuromonitoring techniques. A microdialysis study recently demonstrated that metabolic crises—characterized by simultaneously low glucose and high lactate/pyruvate ratio in the brain extracellular fluid—occur frequently after TBI despite controlled ICP and may be predictors of poor outcome [56]. Such crises may also be associated with seizures and periodic discharges [57]. With regard to implications of these findings on current clinical practice, it remains to be proven whether intervening on these values might improve outcomes.

With regard to PbtO2-guided therapy, the phase 2 Brain Tissue Oxygen Monitoring in Traumatic Brain Injury (BOOST-2) trial aimed to show that placement of a PbtO2 monitor was safe and that a monitor-guided protocol is feasible [58]. The study enrolled over 100 patients, with preliminary data showing non-futility of a protocol incorporating PbO2-directed interventions versus a standard-of-care algorithm incorporating ICP alone; a trend was seen toward lower mortality and overall improved 6-month functional neurologic outcomes in the intervention arm using a protocolled approach to ICP and PbtO2 treatments compared to the control group with treatment guided by ICP alone [59]. A separate systematic review of the literature has suggested that a combined paradigm using PbtO2, ICP, and CPP values to guide treatment is associated with better outcomes, when compared to traditional ICP- and CPP-guided therapy alone [60]. Despite these studies, the current BTF guidelines no longer include specific PbtO2 thresholds to be used in management. The 2007 BTF guidelines suggested a PbtO2 value less than 15 lasting greater than 30 min was associated with higher mortality and used this threshold to recommend treatment [61]. However, the recommendations were removed, as literature inconsistency exists with respect to neurologic outcomes and mortality among the above-mentioned systematic review and 8 other studies reviewed [14].

A full discussion of the extent of continuous electroencephalography (cEEG) as a neuromonitor in sTBI is beyond the scope of this review. cEEG monitoring can be used to “detect [non-convulsive seizures] and protect [with goal directed treatment]” the already injured and vulnerable brain [62]. An ongoing area of research is effect of non-convulsive seizures (NCS) on outcomes, as there has been an association made between NCS in TBI to hippocampal atrophy [63]. It should be noted that the use of electrocorticography is an active area of research in TBI. Electrocorticography is invasive cEEG used to investigate cortical spreading depressions, or propagating waves of astrocyte depolarization, which have been associated with secondary brain injury [56, 57]. A study that analyzed 53 TBI patients with electrocorticography concluded that spreading depolarizations are associated with worse clinical outcome [64]. The study identified depolarizations with isoelectricity or periodic epileptiform discharges, with prolonged depression of spontaneous activity, and with occurrence in temporal clusters as those in particular that portended worse outcomes.

Specific Strategies for Medical Management

Figure 2 highlights recommendations for sTBI medical management in the 2016 BTF guidelines that are unchanged from 2007, while Fig. 3 highlights the new recommendations in 2016 [14, 39]. Of note, in addition to revised ICP and CPP thresholds and recommendations on EVD management as discussed, the scope of the new recommendations also covers blood pressure goals, seizure and infection prophylaxis, and proper nutrition. This section of the review highlights additional aspects of clinical management that may be of particular interest to intensivists.

Fig. 2
figure 2

Unchanged BTF guideline recommendations. AVDO 2 arteriovenous oxygen difference, BtpO 2 cerebral tissue oxygenation, CPP cerebral perfusion pressure, CT computed tomography, DVT deep venous thrombosis, ICH intracerebral hemorrhage, ICP intracranial pressure, O 2 oxygen, PHT phenytoin, ppx prophylaxis, PTS post-traumatic seizures, SBP systolic blood pressure, SjO 2 jugular bulb venous oxygen saturation, VPA valproate. *Recommendations from the prior (third) BTF guidelines not supported by evidence meeting current standards

Full size image
Fig. 3
figure 3

New BTF guideline recommendations. CPP cerebral perfusion pressure, CSF cerebrospinal fluid, DC decompressive craniectomy, EVD external ventricular drain, GCS glasgow coma scale, ICP intracranial pressure, PHT phenytoin, PI povidone–iodine, ppx prophylaxis, PTS post-traumatic seizures, SBP systolic blood pressure, VAP ventilator-associated pneumonia. *Has been shown to reduce ICP and minimize ICU days

Full size image

Osmotic Therapy

Osmotic agents are useful for reducing both intracranial pressure and cerebral edema [65, 66]. Though the evidence that osmotic agents improve outcomes is limited, the most recent BTF committee maintains that osmotic agents are useful for ICP reduction for sTBI patients. The 2016 guidelines do not recommend a specific osmotic agent but suggest judicious use of hypertonic saline in patients with chronic hyponatremia and avoidance of mannitol in patients with hypotension. In a retrospective study of matched patients, hypertonic saline administered as bolus therapy was more effective than mannitol at lowering ICP burden, with shorter ICU length of stay but similar 2-week mortality [66]. Two additional studies have suggested that boluses of hypertonic saline are more effective at reducing ICP compared to mannitol [67, 68]. A meta-analysis has also concluded that hypertonic saline is more effective than mannitol at treating elevated ICP, although the study was limited by small sample size among its component studies [68]. A recent review of hyperosmolar therapy for intracranial hypertension highlighted the current lack of strong evidence to favor continuous, scheduled bolus, and/or as-needed bolus dosing of hypertonic agents [69].

Hypothermia

Hypothermia has been observed to reduce ICP in clinical practice and in studies, but definitive proof that either prophylactic hypothermia or hypothermia as a treatment for refractory ICP elevation may improve clinical outcomes for sTBI patients is still lacking.

Clifton et al. examined the role of prophylactic mild hypothermia (≤35.0 °C) by following patients who were hypothermic on admission and either kept hypothermic or allowed to passively rewarm. Outcomes were compared to those of patients who were normothermic on admission [70]. Six-month outcomes appeared to be improved when hypothermia was maintained, but it was unclear whether this benefit was attributable to neuroprotective effects of early hypothermia versus the adverse effects of early rewarming. In a follow-up trial, sTBI patients ≤45 years old were randomized to either prophylactic mild hypothermia for 48 h or normothermia [71]. The study was stopped early due to futility after enrolling 232 patients, as an interim analysis showed a nonsignificant trend toward worse 6-month outcomes and increased mortality in the hypothermia group. Based on these results, the current BTF guidelines do not recommend prophylactic hypothermia. This recommendation is now Level II B, a stronger evidence level than in the prior edition. Of note, the planned POLAR study will further address the effect of prophylactic moderate hypothermia (33 °C) on sTBI outcomes, by enrolling older patients (up to age 60) and maintaining hypothermia for a longer duration (72 h) [72].

The recently published Eurotherm trial examined the use of therapeutic hypothermia for refractory ICP elevations in sTBI patients [73]. Patients were randomized to either standard critical care or standard critical care plus hypothermia (32–35 °C). Patients in the hypothermia group were allowed to receive osmotherapy only if hypothermia did not control refractory elevations in ICP, defined as ICP > 20 mmHg. After enrolling 347 patients, the trial was suspended early due to futility with worse outcomes and higher mortality among patients in the treatment arm, despite hypothermia effectively reducing ICP.

Decompressive Craniectomy

A full discussion of the range of intensivist and neurosurgical opinions surrounding decompressive craniectomy for diffuse cerebral edema in sTBI is beyond the scope of this medical management review. Nevertheless, a brief exploration is warranted in response to the BTF statement that bifrontal decompressive craniectomy is not recommended to improve outcomes at 6 months in sTBI patients without a mass lesion (e.g., epidural hemorrhages) having refractory ICPs.

Data that influenced this recommendation were derived from the Decompressive Craniectomy in Diffuse Traumatic Brain Injury (DECRA) trial, which randomized 155 patients with sTBI and ICP elevations refractory to first-tier interventions to bifrontal surgical decompression versus continued medical management. The study found unfavorable outcomes in patients with bifrontal decompression, even though surgery effectively reduced ICP [74]. Criticisms were raised over inequity of patient population, with more bilaterally unreactive pupils in the surgical arm, and minimal elevations in ICP leading up to surgery, with the mean only around 20 mmHg. Bifrontal craniectomy also may not have been the optimal surgical approach to reduce ICP, especially with hemispheric lesions. Furthermore, there was an 18% crossover rate to the surgical arm from those randomized to the medical arm.

Importantly, the more recent RESCUEicp trial was published shortly after the release of the new BTF guidelines [75]. Patients in RESCUEicp were randomized to either a surgical or medical arm if, after aggressive medical management, ICP was sustained >25 mmHg for 1–12 h. This study found that patients in the decompressive craniectomy arm had decreased mortality but were more likely to survive in a vegetative state or with severe disability [75]. Notably, a subset of patients with severe disability regained functional independence by 6 months. Unlike the DECRA study, the RESCUEicp trial defined refractory elevated ICP as that over 25 mmHg (lasting for 1–12 h) as compared to 20 mmHg (for 15 min within a 1-h period). RESCUEicp also allowed for both bifrontal craniectomy (63% of craniectomies) and hemicraniectomy (37% of craniectomies), while DECRA only allowed for bifrontal craniectomy. How the RESCUEicp results might affect clinical practice and future revisions of the BTF guidelines remains unknown, but the study provides important data regarding surgical outcomes that can be used to inform ICU decision-making [76].

If a decision is made to proceed with decompressive craniectomy for an sTBI patient, BTF guidelines favor a large versus a small frontotemporoparietal decompressive craniectomy to reduce mortality and outcomes, based on a prospective study of 486 patients [77].

Paroxysmal Sympathetic Hyperactivity

A complication of sTBI that received little attention in the BTF guidelines is paroxysmal sympathetic hyperactivity (PSH), also commonly referred to as “dysautonomia” or “sympathetic storming.” This syndrome is characterized by a constellation of adrenergic symptoms, including agitation, diaphoresis, hyperthermia, hypertension, tachycardia, and tachypnea accompanied by motor symptoms such as hypertonia and extensor posturing [78]. A consensus statement sought to simplify >30 forms of nomenclature for this condition and define diagnostic criteria [79]. The writing group for this statement ultimately arrived at PSH as consensus terminology and also developed a diagnostic tool, the paroxysmal sympathetic hyperactivity measure (PSH-AM). This diagnostic tool was created as a probabilistic system which assigns likelihood to the diagnosis, emphasizing the difficulty in excluding alternative diagnoses such as sepsis and drug withdrawal, which can similarly present as overactive sympathetic activity.

PSH occurs in about 10% of sTBI patients [80,81,82] and is associated with younger age, more severe injury, early fever [83] and diffuse axonal injury [82, 84]. While recovery from the complication is possible, those with PSH generally have longer hospital and ICU stays, greater costs, and possibly worse outcomes [80, 85]. The exact pathophysiology of PSH remains unclear, but current mechanistic hypotheses involve a loss of inhibitory control, producing unopposed sympathetic activity [86, 87].

Treatment of PSH may shorten ICU stays and reduce complications, but this has not been assessed prospectively. The pharmacotherapy of PSH targets the inhibition of sensory afferents, central sympathetic outflow, and end-organ responses to the sympathetic system [88], including medications such as gabapentin, bromocriptine, oxycodone, clonidine, and propranolol to blunt the effects on those respective targets. Figure 4 illustrates an original approach to the management of PSH, with features adapted from Baguley et al. [79]. Recommendations are based upon expert opinion but not yet evidence based.

Fig. 4
figure 4

Suggested approach to paroxysmal sympathetic hyperactivity. Recommendations are based upon expert opinion but not yet evidence based. *Clinical features adopted from the “Diagnosis Likelihood Tool” and “Clinical Feature Severity Scale (CFSS)” from Baguley et al. [79]

Full size image

Seizure Prophylaxis

Early post-traumatic seizures (PTS), defined as those that occur within one week of injury, can occur in as high as 16.9% of TBI patients [89]. Ten percent of those patients are at risk of status epilepticus [90]. In a double-blinded randomized controlled trial of 404 patients randomized to either drug or placebo, prophylactic phenytoin was shown to significantly decrease the rate of early PTS [91], prompting a Level II A recommendation on its use for 7 days post-injury in the third edition of the BTF guidelines. This recommendation has been carried over to the current iteration, given the strength of the data. In a recent prospective study of 813 patients, there were no differences in early PTS rates, adverse drug reactions, or mortality when comparing levetiracetam to phenytoin prophylaxis [92].

Pharmacologic Neuroprotective Strategies

Multiple trials of neuroprotective agents have been conducted with hopes of reducing secondary neuronal necrosis and apoptosis in the injured brain [93]. Pharmacologic therapies that have been tested in sTBI clinical trials or that are currently in development are summarized below.

Corticosteroids

The most notable study regarding corticosteroids for sTBI patients is the CRASH trial, which randomized patients to 48 h of methylprednisolone treatment versus placebo, with the primary outcomes being death from all causes at 2 weeks and death or severe disability at 6 months [94]. The study was stopped early after 5 years, based on higher risk of death at 2 weeks as well as increased death or severe disability at 6 months. Based largely on this trial, the BTF has adopted a strong Level I recommendation that corticosteroid use among sTBI patients for neuroprotective purposes is contraindicated.

Progesterone

Animal models of TBI suggested a benefit of progesterone administration on the regulation of cytokines, excitotoxicity, apoptosis, and vasogenic edema [95]. Several small, early-stage clinical trials [96, 97] also suggested that acute progesterone administration may improve neurologic outcomes patients with sTBI [98]. Based on these findings, the efficacy of progesterone was recently studied in two Phase III trials. The industry-funded SYNAPSE trial enrolled 1195 patients with sTBI and randomized administration of progesterone versus placebo. There was no benefit of progesterone on the primary outcome of GOS at 6 months or secondary outcome of death. There was also no difference in the rate of adverse events between groups [99]. The similar NIH-funded PROTECT III trial was stopped early due to futility, after randomizing 882 of the planned 1140 patients to progesterone versus placebo [100]. There were no benefits of progesterone over placebo either regarding 6-month favorable outcome or secondary outcomes of death or disability.

Other Neuroprotective Strategies

Several other pharmacologic agents have been studied as possible neuroprotectants for sTBI patients, with most of these trials not powered adequately to assess benefits on neurologic outcome and mortality.

In experimental models of sTBI, cyclosporine-A has been shown to preserve mitochondrial integrity by preventing calcium influx, thus preventing secondary neuronal injury [101]. Several studies have shown that cyclosporine-A is safe and tolerable in humans, with at least no negative impact on outcomes or mortality [102, 103]. Lulic et al. have reviewed the shortcomings of the cyclosporine-A preclinical and clinical trials, especially with regard to the heterogeneity of timing of administration of the drug, and have offered guidance for future studies, to clarify the role of cyclosporine in sTBI [104].

Several recent studies have examined the neurocytoprotective effects of erythropoietin (EPO) on outcomes. In one study, 200 TBI patients, who were not able to follow commands, were randomized to one of the four groups: EPO, placebo, a hemoglobin transfusion threshold of 7 g/dL, or a hemoglobin threshold 10 g/dL, respectively [105]. Neither administration of EPO nor a transfusion threshold of 10 g/dL improved outcomes at 6 months. The higher transfusion threshold of 10 g/dL group had a higher incidence of thromboembolic events. EPO treatment in the recent EPO-TBI trial, which randomized 606 patients to either EPO or placebo, did not reduce severe neurologic dysfunction or mortality at 6 months nor increase the incidence of deep venous thrombosis [106].

In a phase II study of 86 sTBI patients, administration of dexanabinol, a cannabis derivative, was shown to help achieve better ICP/CPP control without jeopardizing blood pressure, with a trend toward better neurologic outcome [107]. However, a subsequent phase III trial of 846 patients concluded that the drug is safe but not efficacious in improving 6-month neurologic outcome, as compared to placebo [108].

The use of preinjury statin has been shown to be associated with reduced risk of death and improved functional recovery at 12 months in moderate to sTBI patients over 65 [109]. However, another study of preinjury statin use showed no detectable difference regarding disability at 3 months [110]. Whether starting a statin after TBI confers neuroprotective benefit is currently unknown.

Continuous magnesium infusion has been studied as a neuroprotective agent in sTBI. However, a randomized controlled trial not only failed to show efficacy but reported trends toward increased mortality, pulmonary edema, and respiratory failure in the magnesium treatment groups [111].

In animal models, pharmacologic modulation of TBI-induced nitric oxide production has been linked to decreased cerebral edema and improved outcomes [112, 113]. NOSTRA, a phase III, randomized, double-blinded multicenter trial, is currently assessing the efficacy of a nitric oxide synthase inhibitor on 6-month neurologic outcomes [114].

Neuroprognostication

For sTBI patients, there is significant variability among experts in perceptions of neurologic prognosis, on approaches to neuroprognostication, and on practice of recommendations made to families regarding continuing life-sustaining therapies [115]. Attempts to develop accurate, reliable models for outcome prognostication for sTBI patients date back as far as 1975, with the development of the GCS [4, 116]. The admission of GCS remains a key predictor in more recently developed models. Such models have utilized large collaborative databases and multivariable regression to attempt to adjust for confounders, but pitfalls pertaining to generalizability and suboptimal fit of observed data to the regression models limit their use in clinical care. There is still ongoing work to refine prognostication tools, and consideration of adding ICU complications, multimodal data, and “ICP therapy intensity levels” as variables in prognostication have been proposed [117,118,119].

Clinical Scales

The IMPACT study analyzed data from 11 studies, in 8509 patients, with severe or moderate TBI [120]. In this cohort, the strongest predictors associated with 6-month outcome included age, GCS motor score, pupillary reactivity, hypotension and hypoxemia in the field or ED, and the Marshall CT score for stratifying head CT findings. A publicly available IMPACT score calculator is available at http://www.tbi-impact.org/?p=impact/calc. Limitations of the IMPACT model include the fact that most patients from which the model was derived were treated in the 1980s and 1990s. Moreover, the accuracy of the GCS motor scores in the model may have been confounded by sedation and paralytics, incomplete data, and missing variables. Only variables from admission are included in the model, which does not factor in complications during the commonly lengthy ICU course of these patients.

Models based on trial data from the CRASH study involved over 10,000 enrolled patients. Two prognostic models (basic and CT) have been created, which include patients’ characteristics of age, GCS, pupillary reactivity, and extracranial injury with or without CT characteristics [121]. The inclusion of patients from high- and low- to middle-income countries in the study has the advantage of making the results generalizable to resource-poor settings, but the large discrepancies in outcome based on country and income status are limitations of the model. The publicly available CRASH model risk calculator (http://www.crash2.lshtm.ac.uk/Risk%20calculator/) includes “country” as a variable.

For patients with penetrating brain injury, a recent prognostic score, the SPIN score, was developed to identify predictors associated with survival [122]. Derived from a cohort of 413 penetrating TBI patients from two centers in the USA, the SPIN score identified motor GCS, pupillary reactivity, self-inflicted injury, transfer from another hospital, sex, injury severity score, and admission INR as independent predictors of survival. This score has not yet been validated.

Limitations of Clinical Scales

Both the IMPACT and CRASH models have undergone external validation. In one study with 300 consecutively admitted sTBI patients, the two models were shown to be “satisfactory” at predicting outcomes, with good discrimination based on area under the curve analysis [123]. However, as with all outcome prediction scales, one should caution their use in individual patients. The self-fulfilling prophecy of early withdrawal of care may be present in observational data on which predictive models are based. One small study of 47 patients formally concluded an overestimation of unfavorable outcome by the CRASH model [124].

These caveats are reflected in the limited use of such scales in clinical practice. A recent qualitative study of 20 attending physicians caring for sTBI patients in the USA from all 5 geographic regions in the USA and from neurocritical care, neurosurgery, trauma, and palliative care revealed a highly variable knowledge of the IMPACT model and cautionary attitudes toward its use [125]. Of all participants, 75% knew of the IMPACT model, but only 42% used it, and only with great caution. Reasons for hesitation included distrust in the data from which the IMPACT model was derived, heterogeneity of sTBI patients, and the general concern that outcome scales are derived from populations and are not reliable on individual patients.

Neuroimaging and Prognostication

Head CT is the preferred technique for acute neuroimaging of patients with TBI because of its accessibility, speed of acquisition, and ability to detect lesions that require urgent neurosurgical intervention. From the standpoint of prognostication, CT grading systems such as the Marshall CT classification [8] and Rotterdam CT score [9] have been incorporated into the IMPACT and CRASH prediction models [120, 121]. Yet despite the widespread use of head CT in the diagnostic and prognostic evaluation of patients with sTBI, multiple studies have shown that magnetic resonance imaging (MRI) provides higher sensitivity for detecting prognostically relevant intracerebral lesions, particularly traumatic axonal injury [126,127,128,129]. Accordingly, MRI provides greater prognostic utility than does CT [126, 127, 130]. However, MRI is not always feasible in sTBI, given possible metallic injury, inability to lie supine due to elevated ICP, and limited availability of MRI scanners at some hospitals. Moreover, unexpected recovery of functional independence has been reported in MRI studies of patients with severe intracerebral injuries, including brainstem traumatic axonal injury [131, 132]. Thus, it is important for intensivists to be aware even though MRI has limited specificity for predicting poor outcomes.

Given increasing appreciation for the limitations associated with CT- and MRI-based prognostication, a growing number of studies are testing the prognostic utility of advanced structural and functional imaging techniques [133]. One such advanced technique is diffusion tensor imaging (DTI), which measures the directional diffusion of water along axon bundles [134] and is therefore a potential biomarker of traumatic axonal injury [135]. In the largest DTI prognostic study performed to date in patients with sTBI, DTI measures of white matter injury outperformed the IMPACT score with respect to prognostic accuracy [136]. Diffusor tensor tractography (DTT) can further assess white matter connectivity by creation of 3-dimensional reconstructions of neural tracts using DTI data. Wang and colleagues have shown that DTT data acquired in the acute stage of sTBI may have utility at predicting long-term cognitive function [137].

Functional neuroimaging techniques such as functional MRI (fMRI) and positron emission tomography (PET) have also recently been tested for their prognostic utility, although mostly in the setting of chronic disorders of consciousness (DOC). fMRI uses the blood oxygen level-dependent signal as a marker for cerebral blood flow and, in turn, neuronal activity. In 2010, Monti et al. [138] used stimulus-based fMRI to examine 54 patients with chronic DOC and observed that 5 patients demonstrated fMRI evidence of volitional brain function beyond that detectable on bedside examination. Coleman et al. [139] showed that brain responses detected by fMRI may possibly predict whether a patient with a chronic DOC will go on to recover further behavioral evidence of conscious awareness. These observations raise the possibility that stimulus-based fMRI could have prognostic utility in the acute setting, but no acute studies of stimulus-based fMRI in the acute sTBI population have been performed to date. Stender et al. [140] recently reported that fluorodeoxyglucose (FDG)-PET can be used as a complement to the bedside examination to predict recovery from DOC and may provide greater prognostic accuracy than does fMRI. However, similar to fMRI, FDG-PET has yet to be tested as a prognostic tool in patients with acute sTBI.

Family Support

The lack of awareness of the general public regarding the impact of sTBI [141] coupled with the shock and stress of an unexpected hospital admission means that many families of sTBI patients are often overwhelmed in the ICU. The unique needs of families of comatose, brain-injured patients in the ICU have been explored in several studies [142,143,144]. Qualitative investigations have demonstrated that families of sTBI patients in particular specifically emphasize their “need to know”—their need for consistent information, their need for involvement, and their need to make sense of the unexpected ICU experience [142]. The importance of compassionate yet accurate communication of neurologic prognosis and acknowledgment of uncertainty where it may exist have also been reported to be first and foremost in the minds of families of sTBI patients [143]. Family needs for sTBI patients also change significantly over the course of a patient’s hospitalization [144]. At the time of survivors’ discharge from the hospital, families transition to the challenging task of finding the professional and community support necessary to facilitate ongoing rehabilitative care. Finding such care is especially challenging for patients with persistent DOC, whose rehabilitation options may be limited.

Several organizations provide resources for families of sTBI patients. The Neurocritical Care Society published a guidebook specifically geared toward families of acute sTBI patients who are attempting to make sense of the ICU environment while making decisions on behalf of incapacitated patients [145]. The book, available at https://www.pathlms.com/ncs-ondemand/courses/1282, was spearheaded by a family member of an sTBI patient who had an extended ICU experience. The Brain Injury Association of America (http://www.biausa.org/), whose mission is to progress the field of TBI while improving the quality of life for those injured, has local chapters in many states. Its website contains links to patient and family support groups that may be helpful as families prepare to transition out of the ICU.

Approaches to shared decision-making with surrogate decision makers in neuroscience ICUs have been reviewed recently [146, 147]. The aforementioned variability in approaches to neuroprognostication for sTBI patients has called attention to a need for evidence-based methods to assist with shared decision-making in goals-of-care discussions. An essential goal of these discussions is to focus on perceived patient preferences in the context of available prognostic data [147]. A formal decision aid to assist both ICU clinicians and sTBI patients’ families with prognostic and goals-of-care discussions is currently being developed by a team at the University of Massachusetts and will undergo validation testing over the next few years.

Summary

Severe TBI remains a burdensome public health concern worldwide. Preventing secondary injury continues to be the aim of ICU treatments. While the strength of evidence on which the most recent BTF guidelines are based is a work in progress and presents a challenge to the neurocritical care community, guideline-driven management nevertheless has been shown to improve outcomes in multiple patient cohorts [44, 148]. Attention to recommendation changes in the most recent BTF guidelines, awareness of clinical trial results, use of clinical judgment combined with evidence-based tools for prognostication, compassionate attention to families’ needs, and insight into patient values are all vital for optimizing sTBI patient care.