Management of intracranial hypertension with and without invasive intracranial pressure monitoring

Abstract

Management of intracranial hypertension (IH) has improved in the last decades driven by advancements in monitoring technologies and a deeper understanding of its pathophysiology. Although intracranial pressure (ICP) catheters are still recommended by current guidelines for monitoring patients at risk of IH, these methods are not without limitations. Challenges include procedural complications, availability of these devices in many healthcare settings and technical issues. In this context, management in the absence of ICP monitoring is common and now it can be augmented by intensivist-led point-of-care ultrasound, which includes tools such as transcranial doppler, optic nerve sheath measurement and brain ultrasound. These methods offer anatomic information that can sometimes withhold repeated head computed tomography (CT) scans, but they are also a window into ICP dynamics without the associated risks of invasive monitoring and are reasonable alternatives for guiding treatment, provided an integration between neurological examination, head CT anatomical findings and noninvasive monitors is considered. This manuscript synthesizes the evidence for using invasive ICP monitoring and methods for non-invasive monitoring, more focused on the role of ultrasound, given its wider availability. We also propose a practical approach of how to integrate this information at bedside to avoid both under and overtreatment, by embracing a clinical epidemiology paradigm to guide management decisions.

Key Words: Intracranial hypertension; Intracranial pressure; Neurophysiological monitoring; Transcranial doppler ultrasonography; Neuroimaging

Core Tip: Intracranial hypertension (IH) is one of the most critical syndromes in critical care medicine. Its management is classically guided by invasive intracranial pressure (ICP) monitoring, although recent evidence has questioned the value of invasive ICP monitoring while advances in noninvasive assessment of cerebral dynamics have evolved. In this manuscript, we guide the readers through the recent evidence base, we critically discuss the literature and we provide suggested guidance to the management of IH with and without invasive ICP monitoring.

INTRODUCTION

Invasive intracranial pressure (ICP) monitoring is considered the standard of care for guiding the management of patients with suspected or confirmed intracranial hypertension (IH)[1], as high ICP values for prolonged periods of time are associated with worse outcomes in patients with acute brain injury (ABI)[2-4]. However, access to ICP monitoring technology is limited-especially in low and middle income countries (LMICs)-and the best way to use this tool remains uncertain as using ICP thresholds alone might lead to both over- and undertreatment[5,6].

Noninvasive monitoring modalities for ICP monitoring have increased in the last years and may help to guide therapy in a scenario where invasive ICP monitoring is inaccessible, or indication remains controversial[7]. We conducted a narrative review of the literature to provide a comprehensive and updated overview of ICP monitoring techniques, describing the physiological basis and limitations of invasive ICP monitoring and summarizing current non-invasive methods. We also propose a practical approach to clinical decision-making in scenarios with and without access to invasive monitoring, aiming to guide clinicians on how to avoid over- and undertreatment through a multimodal, individualized management strategy for IH.

PHYSIOLOGY

The Monro-Kellie doctrine states that the brain, cerebrospinal fluid (CSF) and intracranial blood are surrounded by the dura mater and enclosed in a rigid skull. When the volume of an intracranial component expands (e.g., brain edema, hydrocephalus, bleeding) it creates a pressure gradient that causes a compensatory displacement of the remaining components. When this mechanism is exhausted, linear increases in volume cause exponential rises in ICP[8].

The normal ICP varies from 7 to 15 mmHg and current brain trauma foundation (BTF) guidelines define IH as sustained (i.e., > 5 minutes) ICP levels above 22 mmHg[1,9]. However, time spent above the threshold and its intensity-ICP dose[10,11]-might be a better indicator of harm and using a single pressure point to orient treatment is rather simplistic.

The ICP wave corresponds to each cardiac beat and is formed by three peaks: P1 (percussion wave)-reflecting the systolic arterial pulsation transmitted from the aorta; P2 (tidal wave)-corresponding to cerebral compliance; and P3 (dicrotic wave)-associated to venous pressure and relates to the aortic valve closure during diastole[12]. The waveform analysis also provides helpful insights as an inverse P2/P1 ratio represents a decrease in cerebral compliance even in the presence of an ICP below 22 mmHg[13]. Another way of evaluating cerebral compliance is through the Lundberg waves, which represents different patterns of rhythmic or periodic fluctuations that reflect changes in cerebral dynamics[14].

Cerebral perfusion pressure (CPP) is defined as the difference between mean arterial pressure (MAP) and the ICP. CPP is directly related to autoregulation-the capability of the arteries to dilate or vasoconstrict in response to changes in blood pressure to maintain a constant cerebral blood flow-and may be impaired in ABI[15]. A target of 60-70 mmHg is currently recommended by the BTF[1]; higher values do not correlate with a linear improvement in brain oxygenation and the use of higher doses of vasopressors (necessary to achieve CPP > 70 mmHg) may increase complications, such as acute respiratory syndrome[16,17]. CPP and cerebral autoregulation can be estimated with invasive ICP monitoring. Importantly, when assessing CCP, the invasive arterial pressure monitor should be zeroed at the tragus level, and not at the flebostatic axis, to allow correct measurement of CPP[18,19]. The MAP challenge can be used as a bedside dynamic test to assess autoregulatory status in real time. This procedure involves transiently increasing MAP (typically by 10 mmHg for up to 20 minutes) while maintaining other physiological variables stable. If this elevation in MAP leads to a concomitant increase in ICP, it suggests impaired autoregulation (passive pressure transmission), whereas stable or decreasing ICP may reflect preserved autoregulatory capacity[20]. Another method is the pressure reactivity index, which correlates MAP and ICP values through a computer software. A result higher than 0.3 is associated with impaired autoregulation and worse outcomes in traumatic brain injury (TBI) patients[21].

INVASIVE ICP MONITORING

Indications

Invasive ICP monitoring is strongly recommended for patients with overt signs of acute IH for whom a repeat neurological examination is not reliable. ICP monitoring may be considered for patients without overt signs of IH, but who have a high suspicion, based on neurological examination and head computed tomography (CT) findings associated with a high likelihood of IH. A third possible indication would be patients who exhibit no neurological examination abnormalities nor head CT findings associated with IH, but who are at risk of later deteriorating (Table 1).

Table 1 Usual indications for intracranial pressure monitoring.

Indication	Likelihood of ICH
Overt signs of intracranial hypertension	Very high
Herniation syndromes (specially anisocoria)
Cushing's triad
Comment (1): These cases are usually identified upon presentation and usually undergo surgical treatment followed by ICP monitoring, except for diffuse brain edema, where isolated ICP monitoring may be indicated; Comment (2): During the ICU stay, this may present as a neuroworsening scenario that prompts treatment escalation, sometimes regardless of ICP monitoring
High risk of intracranial hypertension with unreliable neurological examination	High (approximately 50%)
Coma (i.e., GCS ≤ 8) with abnormal head CT scans (cistern compression, midline shift, contusions, hematomas)
Comment (1): These cases usually don't have overt signs of intracranial hypertension. Coma may be a manifestation of intracranial hypertension or not, but monitoring is advised both to diagnose elevated ICP and eventually to guide treatment. Monitoring may be beneficial both to avoid overtreatment and undertreatment; Comment (2): In scenarios where ICP monitoring is unavailable, these patients represent the greatest challenge to address treatment escalation and de-escalation
Low risk of intracranial hypertension	Low (approximately 10%-15%)
Coma (i.e., GCS ≤ 8) with normal head CT scans
Comment: These cases are unlikely to benefit from ICP monitoring. Repeat CT scans, frequent neurological examinations and noninvasive strategies may help identify the few patients who develop intracranial hypertension

CT: Computed tomography; ICP: Invasive intracranial pressure; GCS: Glasgow Coma scale; ICH: Intracranial hemorrhage.

TBI is the prototype, the most studied condition associated with IH, but invasive monitoring is also suggested for patients with intracranial bleeding secondary to hemorrhagic stroke or subarachnoid aneurysm rupture[22]. In these patients, hydrocephalus is a frequent cause of elevated ICP and a ventricular catheter that allows external drainage may be indicated[23,24].

The 4^th BTF guidelines conditionally recommends that invasive ICP should be monitored in patients with a Glasgow Coma scale (GCS) ≤ 8 after resuscitation and an abnormal CT scan, as many studies have demonstrated that these patients have an approximately 50% risk of developing IH. It also retained an ungraded, conditional recommendation (from the 3^rd guideline) to monitor ICP in patients with GCS ≤ 8 and a normal CT scan with at least two of the following features: Age > 40 years, unilateral or bilateral motor abnormal posturing (motor GCS ≤ 3) or a systolic blood pressure < 90 mmHg at TBI onset[1]. Importantly, the reference for this recommendation is dated more than 40 years ago and is based on 10 cases that fulfilled these criteria[25]. Out of 61 patients with normal CT scans, only 13% developed IH. This highlights the fragile evidence base behind this indication, derived from studies with small sample size and non-standardized ICP monitoring techniques or neuroimaging at the time[26,27]. Another consensus stated that ICP should be monitored in comatose TBI patients with cerebral contusions in whom clinical examination is not completely reliable, following decompressive craniectomy and after evacuation of an acute supratentorial intracranial hematoma if increased risk of IH[28]. These authors have recommended not to monitor ICP in patients with normal CT scans, regardless of any risk factors.

Despite any recommendations, clinical practice varies worldwide: A survey in 66 European trauma centers described that ICP monitor would be used specifically in patients with a GCS ≤ 8 and CT abnormalities, in case of inability to clinically assess a patient with CT abnormalities and in the presence of intraventricular hemorrhage[29].

Modalities of invasive ICP monitoring

ICP devices for pressure monitoring can be ventricular catheters, intraparenchymal transducers, subdural or epidural monitors. Intraventricular placement is the gold standard for ICP measure and presents reasonable availability, cost and accuracy. The ideal position is in the frontal horn of the lateral ventriculi near the foramen of Monro and the transducer needs to be zeroed to atmospheric pressure to ensure an adequate measurement[30,31]. As an alternative, intraparenchymal probes are usually positioned in the right frontal white-matter region and have significantly less risk of complications than ventricular catheters[30]. A meta-analysis comparing the two methods for guiding therapy did not show a difference in mortality or functional outcome between the groups[32]. Subdural and extradural devices underestimate ICP values compared to other monitors, are less accurate and are now rarely employed in clinical practice.

Limitations

Intraparenchymal transducers can only be calibrated before insertion and therefore are prone to drift over time, especially for fiber optic catheters. The average drift is less than 0.75 mmHg after position[33], but can significantly increase after some days, which affects the measure reliability[34,35]. Moreover, this type of catheter reflects a localized pressure, which does not necessarily represent the global ICP. Intraventricular devices may also record inaccurate values in case of diffuse cerebral edema and if the ventricles are collapsed[36].

ICP monitoring relies on the Monro-Kellie doctrine and the role of this measure after decompressive craniectomy is uncertain. Although the assumption of a rigid closed compartment is not maintained, the restriction imposed by the reconstructed dura mater and the scalp may be enough to raise ICP[37]. In a group that underwent decompressive hemicraniectomy after malignant middle cerebral artery infarction, all patients monitored with ICP had similar initial values, but those with a trend towards an increasing pattern fared worse, being more likely to die in the intensive care unit (ICU) or remain alive with an unfavorable outcome at three months of follow up[38]. Another small cohort suggested that the ICP absolute threshold for treatment in this population should be more aggressive. They described that an ICP over 10 mmHg within 72 hours after surgery was prognostically associated with mortality[39]. How to interpret these findings and translate into daily care to guide therapy is debatable and we advise caution in applying these thresholds to clinical practice[40].

Among possible complications of ICP monitoring, ventriculitis is a major concern with an average incidence of 8.8%, reaching 22% in some studies. It is more frequent in patients with intraventricular catheter than parenchymal ones. The risk of ventriculitis significantly increases over time, with most infections occurring after the fifth day of insertion[41]. Therefore, limiting the duration of ICP monitoring may decrease the incidence of infection. This concept has led to considerable variation in the SIBICC statement[20] regarding the timing of ICP monitoring withdrawal (represented as heatmaps), as different intensivists favor differently the potential benefits of ICP monitoring against the risks of infection-we favor an aggressive ICP monitor withdrawal. Other complications include postprocedural tract hemorrhage, misplacement or blockage of the catheter[41,42].

In addition to the catheter related infection and possible drift of the measure over time, ideal duration for aggressive treatments to control an isolated pressure number is unknown. In the BEST-trip study, the median time of ICP monitoring was four days[43]. In the SYNAPSE-ICU cohort, TBI patients were monitored for a median of six days, while patients with subarachnoid hemorrhage (SAH) were monitored for twelve days[44]. More prolonged ICP monitoring in SAH patients may more likely represent an ongoing indication for ventricular drainage instead of increased ICP, which makes any comparisons futile.

Evidence that favors use of ICP monitoring

BTF recommends that ICP monitoring should be used in patients with severe TBI and abnormal head CT to reduce in-hospital mortality and 2-week post injury mortality with a level of evidence IIB[1]. This suggestion derives from Farahvar et al[45] prospective cohort, which included 2134 patients with severe TBI (GCS ≤ 8) and described lower two-week mortality rate for patients that were treated with ICP monitoring. Alali et al[46] also reported a reduced mortality for patients submitted to ICP monitoring, even though variability in ICP monitoring rates contributed only partially to hospital mortality.

Considering there is no strong evidence to favor an ICP monitoring-driven approach, clinical practice varies among the countries. The SYNAPSE-ICU cohort, composed mostly by high income nations, reported that 56% of the patients with ABI (including TBI and hemorrhagic stroke) were managed with an ICP catheter[44]. The median therapy intensity level was higher in the invasive monitoring group. In addition, patients with unreactive pupils, judged as a marker of neurological severity, presented lower 6-month mortality (34% vs 49%, P < 0.0001) and 6-month unfavorable neurological outcome (60% vs 65%, P < 0.0001). Nevertheless, the authors observed worse outcomes (Glasgow Outcome Scale-Extended) at 6 months among patients with both reactive pupils (OR 1.34, 95%CI: 1.11-1.63). The Chinese extension of the CENTER-TBI[47] group showed a reduced in-hospital mortality for patients with ICP monitoring but with worse functional recovery for the patients discharged home. Shibahashi et al[48] analyzed 31660 patients with TBI in Japan of which 2165 used invasive ICP monitoring and also found lower in-hospital mortality but with no difference in functional outcomes at hospital discharge. A summary of recent studies evaluating the use of ICP monitoring is described at Table 2.

Table 2 Studies comparing invasive intracranial pressure monitoring vs no invasive intracranial pressure monitoring from 2020 to 2024.

Ref.	Country	Design, sample size	Population	Findings
Al Saiegh et al[88]	United States	Retrospective cohort, n = 36929	TBI	ICP monitored patients had a 25% reduction of in-hospital mortality compared to non-ICP monitored patients[79]
Che et al[89]	China	Retrospective cohort, n = 116	ICH	No difference in mortality; ICP monitoring predicted of 6-month favorable outcome (OR: 17, 95%CI: 3-95, P = 0.001)[80]
Ren et al[90]	China	Retrospective cohort, n = 196	ICH	ICP monitoring group presented higher rate of favorable GOS-E at six-months (OR: 0.54, 95%CI: 0.31-0.93, P = 0.027)
Li et al[91]	China	Retrospective cohort, n = 91	Moderate TBI	No difference in 6-month GOS or mortality between the groups
Robba et al[44]	42 countries, mostly in Europe	Prospective cohort, n = 2395	1287 TBI; 587 ICH; 521 SAH	ICP monitoring group with lower 6-month mortality (34% vs 49%, P < 0.0001)
Menacho et al[92]	United States	Retrospective cohort, n = 494	ICH	ICP monitor placement was associated with poor outcome (OR: 2.76, 95%CI: 1.30-5.85, P = 0.008), but not with death (P = 0.652)
Dallagiacoma et al[93]	42 countries, mostly in Europe	Prospective cohort, n = 587	ICH	ICP monitoring is associated with reduction of 6-month mortality (HR: 0.49, 95%CI: 0.35-0.71; P = 0.001)
Foote et al[94]	United States	Retrospective cohort, n = 123	Severe TBI	ICP monitored patients had longer length of hospital stay (12 vs 3, P < 0.001)
Yang et al[47]	China	Prospective cohort, n = 2029	Severe TBI	ICP monitoring patients had lower in-hospital mortality (19.82% vs 26.83%, P < 0.001)
Nattino et al[52]	7 countries in Europe	Prospective cohort, n = 1448	TBI	Worse 6-month GOS-E for ICP monitored patients (death/vegetative state: 39.2% vs 40.6%; severe disability: 33.2% vs 25.4%; moderate disability: 15.7% vs 14.9%; good recovery: 11.9% vs 19.1%, P = 0.005)
Shibahashi et al[48]	Japan	Retrospective cohort, n = 31660	Severe TBI	ICP monitoring associated with lower in-hospital mortality (31,9% vs 39.1%, P < 0.001) and no difference in patients with unfavorable outcomes at discharge (80.3% vs 77.8%, P = 0.127)
Lee et al[95]	Korea	Retrospective cohort, n = 912	TBI	No difference in in-hospital mortality (62% vs 58.9%, P = 0.59) or favorable outcome (25.3% vs 24.6%, P = 0.88)
Waack et al[96]	United States	Retrospective cohort, n = 1664	TBI	ICP monitoring associated with less mortality (35.1% vs 42.4%, P < 0.01) and discharge home (7.9%, 19.3%, P < 0.001)

TBI: Traumatic brain injury; ICH: Intracranial hemorrhage; SAH: Subarachnoid hemorrhage; ICP: Intracranial pressure.

Evidence against using invasive ICP monitoring

Despite the BTF recommendation, compliance with this recommendation is low even in some high income countries[49]. A few studies suggest that complying with previous BTF criteria for ICP monitoring could be associated with decreased survival[50].

There is increasing evidence that generalized ICP monitoring might not result in better neurological outcomes. In a Dutch retrospective cohort study, Cremer et al[51] found that patients with invasive ICP monitoring had increased use of sedatives, vasopressors and prolonged mechanical ventilation with no difference in long-term functional outcomes compared to patients not monitored. Moreover, in almost 25% of the patients, the treatment failed to control ICP below 20 mmHg.

The recent CREACTIVE study[52] evaluated prospective patients from Italy and Hungary that underwent ICP monitoring compared to those monitored clinically and described a worse 6-month recovery in those with invasive monitoring.

In low- and middle-income countries, ICP catheter is not as available as in other scenarios. When available, it may be sometimes reserved to the most severely ill patients, which does not necessarily translate to better outcomes[50]. To investigate if ICP monitoring would be beneficial in resource constrained settings, the BEST-TRIP randomized clinical trial[43], the only one published so far, was conducted in Ecuador and Bolivia. The investigators found similar functional outcomes in TBI patients that were managed on a guideline -based approach with or without an ICP catheter. Patients in the imaging-clinical examination group received a greater number of specific treatments for IH except for high-dose barbiturates which was more frequent in patients with ICP monitoring. The trial corroborates the idea that pursuing an isolated static threshold for managing a complex pathology is likely ineffective[53].

A survey showed that the major concerns for ICP monitoring in LMICs are the cost, availability and no previous experience with the device. Meanwhile, the reported reasons for not using an ICP catheter in high income settings are reservations regarding the real impact in clinical practice and potential complications[54]. There is not yet a real-world, multicenter cost-effectiveness evaluation directly comparing invasive and non-invasive ICP monitoring strategies. However, economic burden of ICP monitoring varies considerably depending on the modality used: External ventricular drains (EVD) cost approximately €100 per disposable catheter, with additional non-disposable circuit costs around €10, and require operating room resources; intraparenchymal fiberoptic probes cost approximately €600 per unit, and require ICP monitoring modules valued at around €2200. For an estimated 100 patients per year, the cost of ICP monitoring using EVDs is approximately €11000, while fiberoptic monitoring may reach €60000 annually. In contrast, non-invasive devices, while involving higher initial costs-such as €5000 for a pupillometer and €30000 for a multipurpose ultrasound machine-are non-disposable and can be used across multiple indications, making them more cost-effective over time[55].

Making sense of divergent data

To understand these divergent data, a few concepts are worthy of discussion (Table 3). First, in Medicine in general, to derive answerable research questions, we need to apply clinical epidemiology principles, and we need to consider biases and statistical issues related to research questions. The first issue is to consider whether ICP monitoring is a question of diagnosis, prognosis or treatment. Studies comparing noninvasive methods to invasive ICP monitoring are diagnostic studies, whereas studies evaluating the benefits of ICP monitoring are not. The next step is whether current studies are studying prognostic associations (i.e., prognosis questions) or causal effects (i.e., treatment questions). In this sense, the question of whether ICP monitoring benefits patients is not well-defined from a causal perspective. Most studies have demonstrated a prognostic association between high ICP levels and worse outcomes. However, since the current recommended treatments for high ICP define “high” and “low” ICP in different ways, the consistency assumption in causal inference remains unsatisfied[56]. If current studies primarily reflect prognostic associations rather than causal effects, clinicians could interpret elevated ICP not necessarily as a modifiable factor, but potentially as a marker of disease severity. This distinction can influence whether aggressive interventions are pursued or avoided. For instance, applying intensive therapies in patients with only modest elevations in ICP, based on the assumption that “lower is always better”, may result in overtreatment and associated complications. Achieving the target ICP with tier 1 therapies is different from achieving the target ICP with tier 2 therapies and tier 3 therapies. Therefore, a shift towards treatment intensity given different versions of baseline and time-dependent confounders is necessary to properly answer this research question. These methodological biases carry significant implications for clinical decision-making.

Table 3 Concepts that need to be considered for the interpretation of neutral and divergent results regarding intracranial pressure monitoring benefit.

Concept	Interpretation
Selection bias	Some observational studies include participants that would be excluded in a clinical trial (e.g., non salvageable patients) of ICP monitoring. They are less likely to receive ICP monitoring in clinical practice and very much more likely to die or have unfavorable outcomes. Confounding adjustment is not enough to address selection bias when the proposed treatment (or monitoring device) would not have been received in a clinical trial
Confounding	While many studies address confounding at baseline, the use of ICP monitoring may lead to differential treatment intensities, which may vary from place to place. Therefore, modern causal inference methods that properly address time-dependent confounding with G-methods is necessary to address not only treatment thresholds, but the benefits or not of ICP monitoring
Dichotomania	Dichotomization of continuous variables leads to loss of information. While there are thresholds that are associated with increases in mortality, the relationship between high ICP levels and worse outcomes is not linear and certainly not a two-level (higher or lower than 20 or 22 mmHg) relationship. This dichotomization leads to riskier treatments being proposed for patients with intermediate (20-25 mmHg) ICP elevations that may not be as harmful as very high (> 30 mmHg) ICP elevations
Risk-guided management	In Medicine, in general, management is guided by different prognostic risk categories. Riskier therapies should be reserved for high risk scenarios where no other alternatives exist. Using tier 3 therapies (barbiturates, hypothermia, craniectomy) for intermediate risk patients based on dichotomania (ICP threshold) may therefore lead to more harm than benefit. Even among the very high risk patients, it may not be beneficial whatsoever, unless as a bridge to a definitive treatment
Prognostic association vs causal effect	Increased intracranial pressure is essentially the consequence of a severe acute brain injury, whether traumatic or not, and is a potential mediator of worse outcomes. Its prognostic association with worse outcomes is well documented. However, whether high ICP has a causal effect on increased mortality or disability is likely dependent on the non-linear relationship of high ICP with worse outcomes. At very high ICP levels (i.e., > 30 mmHg), this is very likely true and high ICP likely mediates this relationship. However, at intermediate high ICP levels (20-30 mmHg), it's likely that there is a complex interplay between treatment intensity, underlying cause of high ICP (diffuse axonal injury, cerebral edema, contusions, etc.), physiological compensation (e.g., maintenance or not of cerebral flow autoregulation) and ICU-acquired complications potentially caused by treatment intensity. Hence, we cannot assume that treatment escalation above proposed ICP thresholds is always beneficial with the current evidence base

ICP: Intracranial pressure; ICU: Intensive care unit.

There is empirical evidence that corroborates this. In the BEST-TRIP trial[43], participants in the control group (no ICP monitoring) had higher treatment intensity of tier-1 therapies (i.e., hyperosmolar therapy), but lower treatment intensity of tier-3 therapies (i.e., barbiturates) compared to the intervention group. This has led to neutral results, although there are many criticisms to the setting where the study was conducted. Furthermore, in the SYNAPSE-ICU cohort[44], participants who were invasively monitored with both reactive pupils had worse outcomes at 6 months. These results are not unsurprising, as they had higher treatment intensity, and the adverse effects of more intense treatment have not been reported in the study. Other studies may present different stories, but essentially, we are comparing different versions of treatment that are likely not comparable.

Beyond inconsistency, absence of positivity (i.e., non-null likelihood of receiving treatment) may also play a role in divergent observations. This is most likely the consequence of selection bias, by including in observational studies participants who would not have received ICP monitoring in the first place given a very high likelihood of death (i.e., non-salvageable patients). The sensitivity analysis of the SYNAPSE-ICU cohort[44] (which excluded patients with a GCS score of 3 and unreactive pupils showed that patients with both reactive pupils had even worse outcomes) is compelling evidence of non-positivity. Finally, the third (and most popularly addressed) assumption for causal inference is exchangeability, which is typically addressed through confounding adjustment at baseline. However, as treatment intensity and thresholds may change throughout a patient's treatment course, time-dependent confounding plays a substantial role in observational studies[57]. Unfortunately, this has not been addressed in most observational studies. Taken together, studies defining treatment thresholds can only be interpreted as studies of prognostic associations rather than causal effects.

Another important issue in ICP monitoring discussion is dichotomization of ICP values. ICP is a continuous variable with a non-linear prognostic association with worse outcomes. Dichotomization may lead to substantial loss of information[56]. Empirical evidence for this can be seen in the different ICP thresholds applied to patients with and without a craniectomy, as mentioned earlier. Studies that define an ICP > 10 mmHg-rather than the traditional 20-22 mmHg threshold-as a marker of higher mortality risk in patients who underwent decompressive craniectomy provide further empirical support that it is not high ICP itself that causes worse outcomes, but rather the underlying disease process leading to increased ICP.

Intermediate levels of high ICP may not be enough to escalate treatment beyond tier-1 therapies, as treatment intensity may be worse than the values of high ICP itself. This may not be true for higher ICP levels (i.e., higher than 30 mmHg), where one may more confidently suppose that, if nothing is done, a spiral of neuroworsening and herniation will follow.

All these considerations lead to an important conclusion: Treatment intensity should be guided to a balanced risk assessment of dose of IH against the treatment intensity level required to achieve a given ICP level, and not only to ICP thresholds.

NON-INVASIVE METHODS FOR ICP MONITORING

Although intraventricular ICP monitoring is considered the gold standard for management of IH[1,22], it is not harmless and there is growing interest in developing and validating noninvasive alternatives to estimate ICP. These may both augment ICP monitoring management when available and, sometimes, partially substitute its utilization where not widely available or when risk profiles favor less invasive strategies.

Many different strategies exist, and we grouped them according to the mechanism used to detect ICP changes or its consequences: Physical examination; brain imaging CT, magnetic resonance imaging (MRI); optic nerve sheath diameter (ONSD); cerebral blood flow variations transcranial doppler (TCD) and monitoring of metabolic alterations [near-infrared spectroscopy (NIRS)]. Less common techniques include indirect ICP estimation [tympanic membrane displacement (TMD)] and neurophysiological studies [electroencephalogram (EEG), Visual evoked potentials (VEP), otoacoustic emissions (OE)]. While physical examination and imaging should always be part of management of patients with IH, other alternatives such as point-of-care brain ultrasound (including ONSD and TCD) are much more available as part of modern intensive care, even in resource constrained settings, given the high versatility of ultrasound machines for ICU, specialized or not. It is important to emphasize that effectiveness of ultrasound techniques relies on the examiner's experience; without training and bedside practice, differences in image acquisition and interpretation may lead to inconsistent findings. Other non-invasive devices may have limited availability, especially in non-specialized ICU that may lack the volume of severe patients to make it cost-effective in practice.

Physical examination

In 1983, Marshal et al[58] described a case series of patients who developed oval pupils in the setting of IH. They hypothesized that IH leads to transtentorial herniation and compression of the third nerve-this abnormality disappeared after ICP was lowered to levels ≤ 20 mmHg. Since then, much attention has been devoted to studying physical examination findings in neurocritical patients.

A meta-analysis with data from over 5000 patients[59] compared physical examination with the gold standard and found that only three findings had a sufficient number of studies: Pupillary dilation, abnormal posturing (motor GCS ≤ 3), and decreased level of consciousness (total GCS ≤ 8). This study showed that pupillary dilation is the most specific finding and decreased level of consciousness the most sensitive, but in isolation physical examination lacks sufficient predictive validity to either detect or exclude IH.

Fundoscopy is a time-honored ancillary exam; it is well known that optic nerve swelling is an indicator of IH and if present should raise concern. However, there’s considerable variability in normal morphology and accurate assessment demands both time and adequate training. Even more, optic disc swelling is a late finding[60], hence this technique is not suitable for emergency conditions.

Pupillometry

It is possible to refine pupillary evaluation with an automated pupillometer, a device capable of tracking minor variations in pupils’ size to allow early detection of IH. Chen et al[61] described the neurological pupil index (NPI) and demonstrated that when pupil reactivity was normal, the average ICP was 19 mmHg, whereas when it was abnormal the average ICP was 30.5 mmHg. Also, patients with nonreactive pupils had the highest pressures with an average ICP of 33 mmHg. Pupillary reactivity changes detected by NPI preceded ICP elevations in an average of 16 hours, indicating a possible role as an early warning tool. However, automated pupillometry relies on the same neural arc as the standard pupillary light reflex and requires that the afferent and efferent pathways remain intact. NPI is usually unaffected by medications such as opioids and anaesthetics[62].

The ORANGE prospective cohort assessed six-month functional prognosis based on the NPI in patients with ABI. They found an association of at least one recording of abnormal NPI measure and poor neurological prognosis[63]. However, a subsequent secondary analysis of the same population did not find a correlation between abnormal NPI and ICP values[64]. Therefore, the impact of pupillometry in monitoring ABI patients remains subject to discussion. Furthermore, for resource-constrained settings, as it needs disposables to perform measurements, it may increase costs without added benefits compared to imaging and clinical examination (ICE) augmented by point-of-care ultrasound.

Brain imaging techniques

CT is widely used to evaluate morphological changes in patients with ABI with various findings considered to predict increased ICP: Marshall score, presence of contusions, subarachnoid blood, effacement of sulci and gyri, absent or compressed cisterns, ventricles size, midline shift and signs of herniation[59,65]. However, these findings have good specificity but lack sufficient sensitivity and may present high false-negative rates[66]. Therefore, a normal CT scan does not necessarily exclude the risk of early-stage IH or late development of increased ICP[67].

MRI provides a more reliable assessment of ICP. Alperin et al[68] developed in an animal model a technique to derive brain elastance from MRI scanning. It takes advantage of the concept of intracranial elastance, derived from the exponential relationship between ICP and volume. The pulsatility of blood flow at each cardiac cycle induces small changes in cerebral blood volume that can be perceived by the MRI machine. Intracranial elastance can then be derived from the variation of these changes during the cardiac cycle making it possible to estimate ICP. In that study this technique had a good correlation with invasive ICP measurements. On the other hand, MRI is not easily accessible, patients need to be stable enough to endure transportation and a long exam duration, and it does not provide continuous assessment[69].

ONSD

The optic nerve is surrounded by a dural sheath, so ICP variations travel through the subarachnoid space and cause an enlargement of the dura in patients with IH. This can be measured with ultrasound, CT, or MRI with good intra- and inter-observer variability. ONSD should be measured 3 mm beneath the retina and the cutoff for IH ranges from 4.8 to 6 mm[70]. This variation occurs because studies have used different measuring techniques (axial vs coronal) and there is uncertainty of which measure to use-some studies chose an average of the measurements between eyes while others used the highest value[66]. A metanalysis evaluating the diagnostic accuracy of sonographic ONSD measurement showed good accuracy in both predicting and excluding elevated ICP, with a diagnostic odds ratio of 67.5 (95%CI: 29-135)[70]. Nonetheless, interpretation must account for limitations of this method. The duration of ONSD dilatation after resolution of IH remains uncertain, and sudden increases of ICP, such as aneurysm rupture, can alter elastic properties of the optic nerve even in the absence of IH[71]. Inter-observer variability and the need for proper training are also challenges for effective implementation[72].

TCD

TCD is a valuable tool for monitoring neurocritical patients with the advantage of being a point-of-care test that can be repeated as needed. A probe of 2MHz in average is placed in one of the possible insonation windows. The most used is the transtemporal window, located at the thinnest portion of the temporal bone, between the lateral edge of the orbit and the auricular area. It allows the visualization of the anterior cerebral circulation. Transforaminal, transorbital and transcervical windows are less evaluated in daily practice[73]. Transcranial color-coded duplex (TCCD) is a suitable alternative to conventional TCD measurements.

TCD/TCCD provides a way of evaluating flow velocity on the major cerebral arteries and understanding cerebral perfusion dynamics. Nevertheless, we stress it does not measure flow itself, but flow velocity. Flow velocity is related to blood flow and to resistance to blood flow. Hence, as ICP increases, flow changes are expected, and they can be studied with doppler. The Pulsatility index (PI)[74] is derived from the TCD waveform and is defined by a mathematical relation between flow velocities (PI = systolic flow velocity-diastolic flow velocity/mean flow velocity). Values above 1.4-especially when accompanied by diastolic flow velocity ≤ 20 cm-suggest IH[75]. Other TCD derived formulas estimate CPP (nCPP) and with basic arithmetic it is possible to calculate non-invasive ICP (nICP = MAP – nCPP)[76].

Brain ultrasound

The ultrasound can be useful to detect extra-axial and intra-axial intracranial hematomas, hydrocephalus, and midline-shift. Measurement of acute cerebral hemorrhage, which appears as a hyperechoic mass, and has a good accuracy compared to brain CT, with the volume estimated by the formula “Volume = longitudinal diameter X sagittal diameter X coronal diameter /2”[77]. Midline shift is evaluated in the diencephalic plane at the transtemporal window and is calculated by the difference between the measurement from the probe to the ipsilateral center of third ventricle minus the same distance in the contralateral side divided by two. It has an area under the ROC curve of 0.85 to detect a significant midline shift (> 5 mm) compared to CT[78]. The assessment of hydrocephalus can be performed with ultrasound and is a valuable tool particularly to compare changes in the diameter of the ventricular system over time. Other applications include identification of aneurysms and arteriovenous malformations and cranial tumors[73].

Skull elasticity

The Monro-Kellie doctrine establishes that the cranium is an inelastic box but that is not entirely accurate. Animal studies[79] revealed that the skull suffers subtle changes in volume according to internal pressure variations. This deformation follows ICP in a linear fashion as no bone hysteresis occurs (i.e. the tendency of a system to preserve a deformation generated by a stimulus). Brain4care^® is a new device that can track this deformation and provides continuous, real-time information about the ICP with a waveform resembling an invasive ICP monitor. It identifies P1, P2, P3, and calculates mean ICP as a time-average of the ICP waveform[80,81]. Small cohorts show reasonable correlation with invasive ICP measure, but additional validation is needed before applying it to clinical practice[82].

Other non-invasive modalities in ABI patients

All the following methods may be used complementarily but are not widely available. Additionally, some are difficult to interpret and translate into clinical practice.

Metabolic alterations: Tissues absorb light in different ways according to the oxygen concentration in them. NIRS takes advantage of that and measures local tissue oxygenation (much the same as a pulse oximeter) providing insights about the brain’s metabolism. It does not allow for estimating the ICP, but it provides an indirect way of evaluating the effects of IH as elevated ICP decreases cerebral blood flow with a consequent reduction in brain metabolism and oxygenation. These neurophysiological changes precede increases in ICP[65].

Functional changes (EEG, VEP, OE): An alternative method of detecting metabolic derangements secondary to IH is through functional assessment of the brain. EEG burst patterns can vary with increasing ICP and can be used as a marker of neurological injury. Once again, these methods do not provide a quantitative assessment of ICP[82].

Indirect transmission of ICP (TMD): The perilymph and the CSF are connected via the perilymphatic ducts. This means that changes in ICP are transmitted to the perilymph and affect the excursion of the tympanic membrane. This can be measured and correlates well with the ICP[69].

Overall acuracy for diagnosis of IH

In a observational study, Robba et al[83] evaluated the diagnostic performance of four non-invasive methods to ICP in ICU patients with ABI. All techniques showed significant correlation with invasively measured ICP. The ONSD had an area under the curve (AUC) of 0.78 (95%CI: 0.68-0.88), with 70% sensitivity and 75% specificity using a cut-off value of > 5.3 mm. The PI showed an AUC of 0.85 (95%CI: 0.77-0.93), with 81% sensitivity and 78% specificity for values above 0.97. Estimated ICP based on TCD and blood pressure yielded an AUC of 0.86 (95%CI: 0.77-0.93), with 73% sensitivity and 84% specificity for values > 20 mmHg. The NPI had the lowest performance, with an AUC of 0.71 (95%CI: 0.60-0.82), 65% sensitivity and 70% specificity for values < 4.1. The best accuracy was obtained by combining ONSD and estimated ICP, reaching an AUC of 0.91 (95%CI: 0.84-0.97), highlighting the potential of a multimodal non-invasive approach for identifying IH.

TREATMENT PROTOCOLS

Since the 4^th version of the BTF guidelines, in which no specific treatment protocol has been proposed to guide management decisions, a few consensus reports have proposed treatment algorithms. Here we briefly review the ICE protocol[84], the CREVICE (Consensus REVised ICE) protocol[85], the SIBICC (Seattle International Severe TBI Consensus Conference) protocol[20] and the most recent B-ICONIC[55] (Brussels consensus for non-invasive ICP monitoring when invasive systems are not available in the care of TBI patients). All these treatment protocols are suggestions for clinical practice based on consensus and Delphi methods, with their inherent criticisms. Furthermore, in all these suggested protocols, the authors were careful enough to explicitly state the importance of individual clinical judgment at the bedside, and to recognize where residual uncertainty exists.

ICE[84] is the treatment protocol proposed by the BEST-TRIP[53] clinical trial for the management of patients without ICP monitoring. The protocol is based on observed evidence of cerebral edema on head CT, with frequent neurological re-assessments and repeated CT at scheduled intervals. The main characteristic of the ICE protocol is to aggressively treat brain edema with hyperosmolar therapy and possibly mild hypocapnia (30-35 mmHg). When clinical neuroworsening is observed or repeat head CT demonstrate worsening cerebral edema, it suggests escalating therapy for even more aggressive hyperosmolar therapy, increased hyperventilation and even barbiturates or decompressive craniectomy. Importantly, the ICE protocol has been tested in a very severely ill population, as demonstrated by baseline characteristics of participants, many of which had CT signs of IH (89%). Its translation to current practice is hampered by a few issues, including the aggressive treatment protocol that may not be suitable to less severely ill patients, but most importantly its reliance on repeat CT scans that may delay treatment de-escalation for patients who improve earlier. With current knowledge about noninvasive ICP estimation, especially with brain ultrasound, this treatment protocol could be adapted to avoid overtreatment.

The same authors of the ICE protocol further developed the CREVICE protocol[85]. Although it retains many characteristics of the ICE protocol, it revised a few issues. First, it defines suspected IH (SICH) and it recommends treatment should be started based on one major or at least two minor criteria (Table 4). After initial treatment (tier 1 therapies) is commenced, treatment escalation is recommended if there is neuroworsening or unacceptable improvement in repeat head CT. Treatment de-escalation is recommended following a period of acceptable stability and following a decision matrix that considers repeat head CT findings and neurological examination findings from pupils and motor scores in the GCS. Finally, the most recently added treatment should be de-escalated first.

Table 4 Criteria for suspected intracranial hypertension from the CREVICE protocol.

Major criteria (one criteria should indicate SICH treatment)

CT classification of Marshall III or worse

Compressed cisterns (Marshall diffuse injury III)

Midline shift > 5 mm (Marshall diffuse injury IV)

Non-evacuated mass lesion > 25 cc

Minor criteria (at least two should be observed to indicate SICH treatment)

Glasgow coma scale (motor) ≤ 4

Pupillary asymmetry

Abnormal pupillary reactivity

Marshall diffuse injury II

CT: Computed tomography; SICH: Suspected intracranial hypertension.

The SIBICC treatment protocol[20] is very similar to the CREVICE protocol in many aspects. It provides a heatmap as a matrix to guide treatment de-escalation and ICP monitoring withdrawal. It also addressed treatments that are not recommended for patients with ICP monitoring (Table 5). The main difference here is that, while ICE and CREVICE recommend scheduled boluses of hyperosmolar therapy in the absence of ICP monitoring, when ICP monitoring is available it is not recommended.

Table 5 Treatments that are not recommended for patients with severe traumatic brain injury.

Mannitol by non-bolus continuous intravenous infusion
Scheduled infusion of hyperosmolar therapy
Lumbar cerebrospinal fluid drainage
Furosemide
Routine use of steroids
Routine use of therapeutic hypothermia to temperatures below 35 ℃
High-dose propofol to achieve burst suppression
Routinely decreasing PaCO2 below 30 mmHg
Routinely raising cerebral perfusion pressure above 90 mmHg

In both CREVICE and SIBICC, the authors used a definition of neuroworsening (Table 6) to better describe situations where treatment escalation and repeated head CT evaluation should be prompted.

Table 6 Neuroworsening definition.

Spontaneous decrease in the GCS motor score ≥ 1 points

New decrease in pupillary reactivity

New pupillary asymmetry or bilateral mydriasis

New focal motor deficit

Herniation syndrome or Cushing's triad requiring immediate physician response

GCS: Glasgow Coma scale.

Other treatments follow a somewhat similar recommendation based on tiers of therapy (Table 7). Importantly, these tiers are considered of increased risk of adverse outcomes with lower risk-benefit profiles as they increase. Hence, treatment escalation should be considered judiciously and patients who don't respond to Tier 1 therapies will not necessarily benefit from Tier 3 therapies, which require discussion and shared decision-making between intensivists, neurosurgeons and next-of-kin. More specifically, we believe that specially barbiturate coma should be considered only as a bridge to additional therapy, but not routinely used for patients with Tier 1 and 2 refractory IH. Additionally, Tier 3 therapies should usually not be considered for patients with mild IH (below 25 mmHg). The RESCUE-ICP[86] trial itself, which showed reduced mortality with a shift in functional status (without improvement in traditional favorable outcome dichotomization), only included patients with refractory hypertension defined by the 25 mmHg threshold.

Table 7 Tiers of therapy for intracranial hypertension.

Tier 0: Basic neurocritical care management for ventilated patients at risk of intracranial hypertension

ICU admission with proper monitoring, including: (1) Invasive arterial pressure monitoring; (2) End-tidal CO2 monitoring; and (3) Core temperature measurement

Venous return optimization with: (1) Head-of-bed elevation to 30-45 ℃; (2) Midline head positioning; and (3) Avoidance of tight cervical collars when possible

Avoidance of ICP spikes with: (1) Analgesia and (2) Mild sedation (not ICP directed) to prevent pain, agitation and ventilator asynchrony

Avoidance of secondary insults, including:

(1) Hypoxemia: Target SpO2 94%-98%

(2) Hypotension: Avoid hypotension by targeting a minimal systolic arterial pressure of 100-110 mmHg or to CPP 60-70 mmHg

(3) Hypocapnia: Target PaCO2 to normal levels (35-40 mmHg)

(4) Hyponatremia: Target serum Na+ to 140-145 mmol/L

(5) Hyperthermia: Target core temperature to below 38 ℃

(6) Hypoglycemia: Target glucose levels to 110-180 mg/dL

Avoid anemia (i.e., Hb < 7.0 g/dL)

Consider anti-seizure prophylaxis for up to 1 week

Tier 1: Deep sedation, CPP optimization, EVD drainage and hyperosmolar therapy

Revise Tier 0 treatment:

(1) Target CPP of 60-70 mmHg or MAP 80-90 mmHg in absence of invasive ICP monitoring

(2) Maintain PaCO2 at lower end of normal (35-38 mmHg)

Intermittent bolus hyperosmolar treatment: (1) Hypertonic saline; and (2) Mannitol

Increase sedation beyond mild sedation to lower ICP (if measured) or to a target RASS of -4/-5 (if ICP not measured), but not to target burst-suppression

Cerebrospinal fluid drainage if external ventricular drain in situ or consider the placement of an EVD

Consider EEG monitoring (if available)

Tier 2: Additional measures with controversial effect

Mild hypocapnia in a lower range (32-35 mmHg)

Trial of neuromuscular paralysis (among ICP monitored patients)

If ICP decreases with a bolus, consider a continuous infusion

Trial of hemodynamic augmentation beyond usual CPP targets:

Perform MAP challenge to assess cerebral autoregulation: If ICP decreases with increased MAP, consider sustaining higher MAP, but no more than a CPP greater than 90 mmHg

Avoid any other adjustments during MAP challenges

In patients without ICP monitors, this trial may be considered with TCCD/TCD measurements

Tier 3: Highly efficacious therapies to reduce ICP, but with demonstrated increased risk of complications (i.e., should not be routinely used except under specific circumstances)

Barbiturate coma with pentobarbital or thiopentone to ICP control (if efficacious) or to pupil abnormality correction (when ICP is not measured)

Secondary decompressive craniectomy

Mild hypothermia (35-36 ℃) with active cooling measures

ICP: Intracranial pressure; MAP: Mean arterial pressure; CPP: Cerebral perfusion pressure; TCCD: Transcranial color-coded duplex scan; TCD: Transcranial doppler; EVD: External ventricular drain; ICU: Intensive care unit.

The most recent B-ICONIC[55] consensus addressed the use of non-invasive ICP monitoring for the management of TBI patients when ICP monitoring is not available. The authors relied on four noninvasive parameters to make their consensus recommendations: (1) ONSD evaluation; (2) TCD/TCCD estimated PI and FVd > 20 cm/s; (3) TCD/TCCD estimated ICP; and (4) NPI abnormalities with pupillometry[55]. Importantly, they used the major and minor criteria proposed by the CREVICE protocol to guide initiation of the protocol to consider SICH, but included the non-invasive estimates as additional minor criteria to escalate treatment. For the B-ICONIC protocol, the authors considered that these non-invasive monitors are imperfect and may have false-positives for high ICP. Therefore, they consider non-invasive ICP assessment as abnormal when at least two of the measurements are considered out of range. The ranges proposed by the authors are described in Table 8. Comparison and main features of the four protocols are summarized in Table 9.

Table 8 Altered findings from non-invasive estimates of intracranial pressure.

(1) Optic nerve sheath diameter (ONSD)

ONSD > 6 mm (any side)

Increase in ONSD > 0.5 from baseline value (any side)

(2) Noninvasive TCD/TCCD ICP estimation (nICP)

nICP > 20-22 mmHg (any side)

(3) Pulsatility index (PI)

PI > 1.4 with FVd < 20 cm/s (any side)

Increase in PI > 0.5 from baseline value (any side)

(4) Pupillometer derived neurological pupillary index (NPi)

NPi < 3 (any side)

NPi reduction > 1 from baseline value (any side)

Table 9 Summary of intracranial pressure managing protocols.

	ICE protocol	CREVICE protocol	SIBICC protocol	B-ICONIC protocol
Setting	BEST-TRIP study (Latin America, limited resources)	Post-BEST-TRIP refinement (Latin America, limited resources)	High-resource ICU with access to invasive multimodal monitoring	Low-resource settings without access to invasive monitoring
Type of monitoring	Non-invasive ICP monitoring	Non-invasive ICP monitoring	ICP ± brain oxygen monitoring (PbtO2)	Non-invasive ICP monitoring
Trigger for protocol activation	GCS ≤ 8 with abnormal CT scan	Suspected intracranial hypertension (SICH): 1 major or 2 minor criteria (e.g., imaging, age, motor response, hypotension)	Elevated ICP and/or low PbtO2	SICH plus ≥ 2 abnormal non-invasive findings (see below)
Basis for monitoring	Serial clinical exams and CT scans	Clinical exams and CT; revised with definitions of neuroworsening and structured de-escalation	Continuous ICP and/or PbtO2 values with tiered thresholds	Four non-invasive markers: (1) ONSD; (2) pulsatility index (3) estimated ICP via TCD; and (4) NPi on pupillometry
Threshold for escalation	Clinical or radiological neuroworsening	Neuroworsening or failure of Tier 1 therapy	ICP ≥ 22 mmHg, PbtO2 < 20 mmHg, or clinical deterioration	Two or more non-invasive abnormalities, or clinical worsening
Treatment Strategy (Tiers)	Aggressive from outset; includes hyperosmolar therapy, hyperventilation, barbiturates, decompressive craniectomy	Tiered escalation (Tier 1 to Tier 3); last treatment in should be first out	Tiered algorithm with increasing intensity and risk (Tier 1 to 3), individualized per ICP and PbtO2 status	Uses CREVICE framework with adapted triggers; relies on availability of non-invasive modalities
De-escalation guidance	Not clearly structured; based on improvement or repeat imaging	Defined matrix combining imaging and neuro exam (pupils, motor score); last treatment added is first to be withdrawn	Heatmap and structured matrix based on ICP/PbtO2; includes neuroworsening and “tier zero” baseline care	Structured de-escalation using improvement of non-invasive markers and clinical status
Limitations	Aggressive, CT-dependent, lacks flexibility for mild cases	Still CT-dependent; subjective criteria for SICH; limited data validation	Based on expert consensus (Delphi method), lacks RCT validation	Dependent on accuracy of non-invasive markers; potential for false-positives or over-treatment
Unique features	First consensus protocol for ICP-unmonitored patients; used in RCT	Adds clarity on neuroworsening, introduces structured de-escalation	Integrates ICP and brain oxygenation; includes “MAP challenge” and not recommended interventions list	First consensus using only non-invasive data; bridges evidence gap in LMICs

CT: Computed tomography; GCS: Glasgow Coma scale; ICP: Intracranial pressure; LMICs: Low- and middle-income countries; MAP: Mean arterial pressure; NPi: Neurological pupil index; ONSD: Optic nerve sheath diameter; PbtO₂: Brain tissue oxygen pressure; RCT: Randomized controlled trial; SICH: Suspected intracranial hypertension; TCD: Transcranial doppler; ICU: Intensive care unit.

BRINGING CONCEPTS TOGETHER FOR IMPROVED PATIENT CARE

A patient-tailored approach is essential to avoid both undertreatment and overtreatment and to optimize the use of resources, especially in constrained settings. Specifically, we believe the main issue here is to try to identify patients with ABI and separate them into three discrete groups. Importantly, we need to recognize the adverse events of treatment intensity, which as we discussed earlier may be the main reason why so many discrepant findings have been observed in cohort studies. A summary of management of IH without invasive ICU monitoring is available in Figure 1.

Figure 1 Patients with clinical signs of intracranial hypertension must be submitted to a head computed tomography and ultrasound (neuro POCUS). An optic nerve sheath diameter larger than 6 mm and an elevated Pulsatility index at transcranial doppler are indicative of elevated pressure. “Tier zero” measures are initiated. In case of persistent intracranial hypertension, treatment must be escalated following the steps described in the text. CT: Computed tomography; IH: Intracranial hypertension; TCD: Transcranial doppler.

First, patients with coma who have a low likelihood of IH should be identified to avoid overtreatment (i.e., unnecessary Tier 1 therapies). Nevertheless, these patients should usually undergo optimized Tier 0 therapies, aiming to avoid secondary neurological damage and to avoid unnecessary complications of higher Tier therapies, which might compromise their outcome, especially prolonged unnecessary sedation. These patients are those with normal head CTs or with mild alterations compatible with diffuse axonal injury among TBI patients. Repeat head CTs and noninvasive ICP evaluations with brain ultrasound derived variables are all that is needed to identify the few that might evolve with IH.

Second, the patients with evidence of clinical neuroworsening (here not considering the noninvasive measurements) not attributed to extracranial causes will usually be the candidates to more aggressive treatment escalation. Optimized Tier 1 therapies should be accomplished, Tier 2 therapies might be considered, and Tier 3 therapies should be carefully discussed. As we mentioned previously, barbiturate coma may be an option as a bridge to definitive treatment for patients with initially conservative management who deteriorate and may benefit from secondary surgical approaches (e.g., cerebral contusion with secondary expansion). However, barbiturate coma for diffuse brain edema is unlikely to be a bridge to good recovery when alternative treatments don't exist and should likely be avoided in most cases.

Third, there are patients in the middle ground, where treatment decisions are more likely to fall in a grey zone. These are patients without overt pupillary abnormalities, who are in coma, but who have concerning head CT scans. For patients who have ICP monitored, treatment escalation should be careful for ICP values of 20-25 mmHg, as treatment escalation may be more harmful than mild ICP elevations. Especially if treatment intensity is continued over many days, increased risks of EVD infection, ventilator-associated pneumonia (VAP) and others should be weighed against mildly elevated ICP values, as ICU-acquired infection may lead to secondary neuroworsening. The risks of ICU-acquired weakness, which could delay recovery and predispose patients to secondary complications, may also counterbalance any benefits. For patients without ICP monitoring, every effort should be made to sustain optimized Tier 1 therapies. Treatment escalation beyond Tier 2 should only be considered if clinical evidence of herniation is evident, if head CT shows confirmation of worsening or if noninvasive measurements are concordant in demonstrating very abnormal ICP values.

FUTURE DIRECTIONS

Whether ICP monitoring is beneficial or not in neurocritical care practice will continue to be a matter of discussion for much time. As for any monitoring device, it's unlikely that its utilization itself may be beneficial. However, if used for prolonged periods of time, it may harm patients by itself when the risk of ICP-monitoring associated infection is not low. The discussion should be, therefore, the treatment intensity that patients with ABI should receive. Sedation, hemodynamic augmentation and neuromuscular blockade should be used judiciously in neurocritical care too, just as they should be in non-neurocritical care patients. Sedation is not good for the brain, although it may be helpful in decreasing ICP. Neuromuscular blockade in ABI poses the patients at very high risk of developing VAP and other complications of prolonged sedation and immobility. Hemodynamic augmentation with higher MAP targets have been consistently associated with adverse events and even worse outcomes among patients with septic shock[88-96] and other conditions, which provides indirect evidence of harm if we consider the very sick population of patients with ABI.

The neurocritical care community should recognize the risks of unnecessary treatment intensity, which could be the consequence (or not) of more intense monitoring. Further research should therefore aim to evaluate different treatment intensities, regardless of monitoring devices. Pragmatic randomized controlled trials comparing a higher treatment intensity level aiming at traditional ICP levels compared to a lower treatment intensity level, which allows for permissive IH to avoid overtreatment, should be designed to finally answer this question. This is especially relevant given the inherent trade-offs of more aggressive ICP management strategies, which frequently necessitate prolonged sedation, increase the risk of ICU-acquired infections and critical illness polyneuropathy, and may significantly delay or impair neurorehabilitation. Both survival and long-term functional outcomes should be assessed. Until then, monitoring should be reserved to where treatment uncertainty exists. While patients with overt signs of IH will need more treatment intensity and patients with coma, but normal head CT scans, should avoid completely higher treatment intensities, uncertainty will rely on patients without reliable neurological examinations with probable IH.

CONCLUSION

Intraventricular pressure monitoring is the gold standard for ICP measurement as it provides continuous and accurate data that reflect global ICP. But it comes at a high price: Infections and bleeding are some of the main risks we inflict our patient. ICP monitoring should be guided by the estimated likelihood of IH. While patients with herniation signs or abnormal CT scans in coma have a high risk and may benefit from monitoring, those with normal imaging and preserved neurological exams are unlikely to benefit and may be safely managed with clinical observation and non-invasive tools. Even though non-invasive methods cannot yet fully replace invasive monitoring-due to limitations such as observer variability, the need for specialized training, limited availability, and insufficient accuracy-they have a role in settings where invasive ICP monitoring is unavailable or its indication remains uncertain. For the future, a multimodal strategy integrating current technologies with advances in artificial intelligence and machine learning algorithms will potentially improve diagnosis and treatment precision, guiding individualized interventions. Also, noninvasive monitoring could be used to identify patients more likely to benefit from invasive monitoring.