The goal of neurocritical care (NCC) is to improve patient outcomes following an acute brain injury or the progression or exacerbation of a chronic neurologic disease.1,2 This includes the treatment and management of the primary brain injury and factors causing secondary brain injury: hypoxemia, ischemia, inflammation, or edema, in order to prevent irreversible neurologic dysfunction. "Real time” monitoring is necessary to detect early brain injury and intervene in a timely manner to prevent irreversible injury. The phrase “To detect and protect,” referring to the utility of continuous EEG (CEEG) monitoring3 can be applied to any of the modalities used in the Intensive Care Unit (ICU) for continuous or frequent bedside monitoring. For example, the change in a specific physiologic modality, such as tachycardia, could indicate decreased brain oxygenation. The subsequent evaluation of this tachycardia could result in a specific treatment to reverse the underlying process. Monitoring also permits individualized patient management and the ability to assess the response to treatment and modify subsequent treatment to maintain certain parameters to prevent injury.4 This chapter reviews the non-invasive and invasive brain monitoring techniques used in combination in the ICU.

Various techniques are available for brain monitoring in the ICU, ranging from the neurologic examination, various physiologic parameters, neurophysiologic tests, and neuroimaging.5, 6, 7, 8 These all have certain advantages and limitations: ability to perform at the bedside, spatial and temporal resolution, scope of use, and ease of interpretation.9 Historically, brain monitoring in the ICU starts with the neurologic examination. This has temporal limitations as it is not done continuously, the clinical change may occur late, lagging behind neuropathophysiological deterioration with the result that the neurologic insult becomes irreversible despite subsequent treatment, or the neurologic examination itself is frequently altered by coma, sedation, or pharmacologic neuromuscular blockade.

MMM refers to using a combination of physiological measurements, including hemodynamic parameters and specific non-invasive and invasive neuromonitoring techniques.1,2,4, 5, 6, 7, 8 The term multimodal monitoring, now referred to as multimodal neurologic monitoring (MNM), refers to the integration of data from these multiple intensive care unit devices to monitor the physiologic changes occurring with ongoing brain injury.10,11 The combination of multiple modalities may deliver a more tailored care to the individual patient12 referred to as precision medicine.13

A recent Delphi consensus survey on MMM consisting of both adult and pediatric neurocritical care (NCC) clinicians came to an agreement that frequent bedside neurologic examinations are insufficient to prevent secondary brain injury. This same survey also demonstrated consensus that there are specific key physiologic processes important to NCC: the clinical examination, systemic hemodynamics, intracranial pressure and cerebral perfusion pressure, cerebrovascular autoregulation, systemic and brain oxygenation, cerebral blood flow and ischemia, electrophysiology, cerebral metabolism, glucose and nutrition, hemostasis and hemoglobin, temperature and inflammation, and biomarkers of cellular damage and degeneration.14

MNM uses invasive and non-invasive monitoring techniques.1,2,4, 5, 6, 7, 8 The non-invasive techniques include vital signs (systemic hemodynamics and temperature), CEEG, near-infrared spectroscopy (NIRS), pupillometry, and transcranial doppler (TCD) ultrasonography. Invasive monitoring techniques include intracranial pressure (ICP), cerebral perfusion pressure (CPP), cerebral blood flow (CBF), brain tissue oxygenation (PbtO2), microdialysis, and intracranial electroencephalography (iEEG). MNM was introduced in adult NCC but is now utilized in many pediatric NCC programs.15

The application of computer technology and the development of biomedical engineering has greatly improved brain monitoring. CEEG monitoring markedly increased with the switch from analog to digital electroencephalography (EEG) technology. With analogue recording and paper EEG, a “routine” EEG was done intermittently to identify seizures or assess background, especially in TBI to identify a burst-suppression pattern as the treatment goal for pentobarbital therapy. In 1972, the protocol for monitoring the patient with Reye Syndrome required an EEG done every 8 h or at the time of a clinical change.16 Subsequently, CEEG monitoring could be done at the bedside, with good spatial resolution and excellent temporal resolution due continuous monitoring that eliminated the “sampling error” inherently present in the intermittent EEG. CEEG permitted the identification of seizures, subclinical seizures, or acute EEG background changes in high-risk situations, such as status epilepticus, stroke, aneurysmal subarachnoid hemorrhage (aSAH), intracranial hemorrhage (ICH), or CNS infections (meningitis, encephalitis). However, its ease of interpretation is poor, requiring specific training. CEEG highlighted the importance of waveforms constituting the ictal-interictal continuum (IIC);17,18 these had been identified on “routine EEG” but CEEG demonstrated their dynamic nature.

The addition of quantitative EEG (QEEG), also called digital trending analysis (DTA)19 to CEEG monitoring not only has greatly enhanced the ability to identify seizures (Fig. 1) but also the early changes indicating ischemia, such as the alpha-delta ratio used to detect vasospasm in aSAH.20,21 QEEG is also easier to interpret at the bedside than standard EEG, which is a great advantage. DTA is derived from, and time-locked to, the CEEG. The simultaneous raw EEG must be available to exclude artifacts. Certain software packages can simultaneously integrate and display EEG, digital trending, and multiple physiologic parameters. CEEG is discussed at length as it is the mainstay of MMM in many neurological intensive care units. Specifically for the following non-invasive and invasive techniques, the reviews used for the basic information are referenced1,2,4, 5, 6, 7, 8,10 with additional specific information added.

Electrophysiology: CEEG with QEEG, is now preferred over routine EEG for brain monitoring, because of its superior temporal resolution. Evoked potentials (EPs), including auditory evoked potentials (AEPs), visual evoked potentials (VEPs) and somatosensory evoked potentials (SSEPs) are also used for MNM. EPs can be done at the bedside, in the sedated or comatose patient, but have only fair spatial and temporal resolution. EPs are useful for prognosis, such as the SSEP after cardiac arrest and may be especially useful to test the somatosensory spinal pathways, in the posterior cord, in the comatose patient. Common data elements for electrophysiology in disorders of consciousness have recently been recommended.22

Systemic hemodynamics: Heart rate, blood pressure, mean arterial pressure, pulse oximetry. These are used for monitoring cardio-respiratory function but also have important relationships to intracranial hemodynamics, including ICP and CPP.23,24 Common data elements for physiology and big data in disorders of consciousness have recently been recommended.25

Temperature: it is important to prevent hyperthermia to reduce metabolic demand and promote brain homeostasis in the setting of critical illness. The known relationship between hyperthermia and worsening outcomes in adult NCC26 also applies to children with acute brain insults. Persistent hyperthermia in the first 24 h after cardiac arrest in children has been associated with an unfavorable neurologic outcome.27 Targeted temperature management (TTM) to either 32 to 34 degrees centigrade or 36 to 37.5 degrees centigrade is recommended for pediatric post-cardiac arrest care at this time.28 A prospective multicenter study, Pediatric Influence of Cooling Duration on Efficacy in Cardiac Arrest Patients (P-ICECAP), is ongoing to establish the efficacy of cooling and optimal duration of induced hypothermia for neuroprotection in pediatric comatose survivors of cardiac arrest (NCT05376267).

NIRS: NIRS uses infrared light to estimate brain tissue oxygenation using the relative absorption of oxygenated and unoxygenated hemoglobin, producing cerebral oximetry values. The ratio of oxygenated Hb to total Hb, the tissue oxygenation index, is related to regional CBF. NIRS has a long history of use in the neonatal and cardiac intensive care units and during cardiac surgery. NIRS has been used an indirect measurement of CBF and as such has been used in assesments of cerebral autoregulation (CA) for children after cardiac arrest, extracorporeal membrane oxygenation (ECMO), and neonatal hypoxic ischemic encephalopathy.29,30,31 Higher cerebral oximetry values during cardioplmonary resuscitation after pediatric cardiac arrest has recently been linked with higher rates of return to spontaneous circulation, but not with survival to discharge.32 The use of NIRS and cerebral oximetry values as an indirect measure of CBF carries limitations, as this assumes that changes in cerebral oxygenation are directly related to CBF, whereas oxygenation changes can also be the result of fluctuations in metabolic demand or worsening cerebral edema. Concerns exist that increased skull thickness, evolving scalp edema, and variations in skin pigmentation can affect NIRS results. At present, there are no established thresholds for NIRS values that are actionable or representative of worsening pathology. Nevertheless, the value of NIRS is thought to come in terms of monitoring the trend of its values over time.

TCD: TCD indirectly measures the mean blood flow velocity of red blood cells in major cerebral vessels.33, 34, 35 TCD is used in children with TBI, intracranial hypertension, vasospasm, stroke, cerebrovascular disorders, CNS infections36 and ECMO.37,38 The Pediatric Neurocritical Care Research Group (PNCRG) is a multicenter collaborative research group dedicated to promoting clinical and preclinical research to advance pediatric NCC. The PNCRG conducted a survey of the use of TCD, for which it is considered clinically useful in intracranial hemorrhage, arterial ischemic stroke, traumatic brain injury, cerebral vascular malformation, mechanical circulatory support, cardiac arrest, hepatic encephalopathy, and cerebral venous malformation (all used at or equal to 5 centers).39 Commonly used metrics asssessed using TCD in neurologic monitoring include flow velocities (FVs) to assess CBF, as well as the pulsatility index (PI). Normative values of FVs have been developed for critically ill, sedated children.40 The PI represents the difference between systolic and diastolic FVs, divided by mean FVs. The PI have been considered as an indirect measure of cerebrovascular resistance, and some consider elevated values to be potentially a marker of raised intracranial pressure.41 In our ICUs, we use it to assess for vasospasm, determining the Lindegaard ratio, especially in patients with subarachnoid hemorrhage from a ruptured vascular malformation. The two main changes in clinical management using the results of TCD are to perform head imaging, and manipulation of cerebral perfusion pressure with fluids or vasopressors.39 A recent prospective observational study identified that asymmetric TCD waveforms of the middle cerebral arteries, assessed using the Thrombolysis in Brain Ischemia (TIBI) grading system, is independently associated with acute brain injury in children undergoing ECMO support.37 A Delphi process on TCD in children recommended it for any ICU patient in whom pathophysiological changes in cerebral hemodynamics were considered.42 A recent study evaluating the use of TCD with MMM in pediatric TBI patients demonstrated that increased critical closing pressure and reduced diastolic closing margin and cerebrospinal space compliance were associated with unfavorable outcomes.43

Pupilometer: The pupilometer is a hand-held device that provides objective measures of pupil size and reactivity before and after a light stimulus. In the management of patients with suspected disorders of ICP, abnormalities of pupil reactivity are often associated with neurologic deterioration and poor outcome.44 Abnormal pupil reactivity is associated with cerebral herniation, third nerve compression and brainstem perfusion.45 Traditional means of detecting changes in pupil size and reactivity by gross inspection are highly subjective46 leading to the need for a more objective tool. Aside from an objective measurement of pupil size, the pupilometer can calculate pupil reactivity (NPi), latency time and constriction and latency velocities.47 The changes in these variables may be helpful to detect pathological ICP increases in selected patients. Attempts have been made to establish pediatric normative data.46

Optic Nerve Sheath Diameter: The measurement of the optic nerve sheath diameter (ONSD) by point of care ultrasound is now used as a non-invasive measurement of ICP. Increased ICP is indicated by a greater than 5 mm measurement, the crescent sign or optic disc elevation.48 However, it is dependent on proper technique and its ability to correlate with elevated ICP is controversial.49 Currently, ONSD can't be recommended to replace ICP monitoring.

Biomarkers: A brain insult may cause structural damage in neurons and glial cells leading to extravasation of specific proteins into blood, cerebrospinal fluid, and extracellular matrix.50 Neuron-specific enolase (NSE), an intracellular neuronal enzyme, is specific for neurons, S100B is from glial cells, and myelin basic protein (MBP) is from oligodendrocytes and these may have prognostic significance.51 NSE and S100B have been measured after pediatric cardiac arrest, with serum NSE at 48 h or later associated with neurologic outcome and S100B at 48 h associated with survival. A recent analysis from the Personalizing Outcomes After Child Cardiac Arrest multicenter prospective cohort study demonstrated that a variety of blood-based brain injury biomarkers, especially neurofilament light (NfL), were associated with an unfavorable outcome at 1 year after pediatric cardiac arrest.52

In MMM, neuroimaging is a specific data point in time that assesses brain anatomy and function. Neuroimaging has poor temporal resolution but great spatial resolution.9 Cranial computed tomography (CT) scans, including CT angiography (CTA), and magnetic resonance imaging (MRI), magnetic resonance angiography (MRA), magnetic resonance venography (MRV), and magnetic resonance spectroscopy (MRS) are structural neuroimaging studies that have excellent spatial resolution but poor temporal resolution, having worse temporal limitations than the neurologic examination. Portable cranial CT scans are available but CTA is not. Portable MRI is not readily available at the bedside in most centers, as the patient needs to be transported to the scanner, which may be impossible in many critically ill patients, such as those on extracorporeal membrane oxygenation (ECMO). However, portable low-field MRI is now available.

Functional neuroimaging measures cerebral blood flow, the single-photon emitted CT (SPECT) or MRI arterial spin labelling (ASL) or cerebral metabolism, the positron emitted tomography (PET) scan. These functional neuroimaging studies may aid in determining the significance of some MMM parameters. SPECT measures CBF and PET scan measures metabolic activity, which are increased in the ictal state and may aid in assessing the significance of some of the IIC waveforms. Arterial spin labelling (ASL) is a new MRI technique that identifies CBF. Vessel wall imaging is an emerging MRI technique that allows for characterization of pathologic features involving the cerebrovascular arterial wall, and may be helpful in assessment of cerebrovascular vasculitis.53

The major invasive MMM techniques include ICP, PbtO2, cerebral microdialysis, and invasive EEG monitoring. Several parameters, such as ICP, CPP, PbtO2, pressure reactivity, and microdialysis (chemical concentrations of dialysate) require an invasive device.

Intracranial pressure: ICP is measured from the brain parenchyma or intraventricular space through an intraparenchymal probe or external ventricular drain, respectively. The ICP is generated by the major contents of the intracranial space: brain tissue, ventricular fluid, arterial and venous blood volume. Intracranial pathology, such as a space-occupying lesion, increases intracranial volume contents and may result in increased ICP. The skull is a fixed space, so when the volume of one component increases, there must be a compensatory decrease in the other components, or else, the ICP will increase.

The ICP waveform is time-synced with each beat of the cardiac cycle and is analyzed for its normal and abnormal components: the P1 wave, the percussion wave, represents arterial input at peak systole; the P2, the tidal wave, represents the venous pulsation just prior to the dichrotic notch; and the P3, the dichrotic wave, represents venous drainage immediately after the dichrotic notch. The origin of the ICP waveform is primarily arterial, with retrograde venous pulsations contributing to the latter components. As intracranial pressure increases with decreased brain compliance, the pulse amplitude of the ICP waveform increases and P2 increases above P1, leading to a gradual loss of waveform morphology (Fig. 2). Transient increases in ICP with a broader time scale are the Lundberg waves, or the A wave. These represent ICP increase to the 50 mm Hg range for 5–20 min with then a spontaneous reduction. The A wave occurs when autoregulation is intact but when there is decreased compliance. B waves are transient sharply peaked waves to pressure levels of 30 mm Hg, which frequently recur and are related to the respiratory pattern; B waves also signify abnormal compliance. C waves, rhythmic elevations every 10 min, are associated with fluctuations in intracranial arterial blood volume.

In adult guidelines, the recommended threshold for treating an elevated ICP is set as 22 mm Hg.54 In children, the Brain Trauma Foundation Guidelines support management strategies to maintain ICP below 20 mmHg.55

CBF is determined, in part, by the intracranial pressure along with the MAP. The venous pressure also contributes to the ICP. Brain perfusion reaches zero when the ICP exceeds the MAP.

CPP: The cerebral perfusion pressure is the perfusion pressure across the brain, related to mean arterial blood pressure (MAP) and ICP. The CPP = MAP – ICP. The recommended CPP is between 50- and 70 mm Hg in adults. A CPP of 45 to 60 mm Hg may be targeted in children, with lower values in younger children.22 In the pediatric TBI guidelines, CPP is recommended to be managed above an absolute minimum value of 40 mmHg55 and suggested CPP targets for a variety of pediatric age groups have been proposed that are above such a minimum value, and significantly so for older children.56

Cerebral blood flow (CBF): Thermal diffusion flowmetry (TD), an invasive measure of cerebral blood flow (CBF), provides continuous quantitative brain perfusion measurements. This technique uses two thermistors within the probe, a proximal source set at the temperature of surrounding tissue and a distal censor heated 2 degrees Celsius higher. TD takes advantage of the capacity of blood to dissipate heat to quantify CBF in units of mL / 100 g / minute.57 TD values below 15 mL / 100 g / min have been associated with cerebral vasospasms in adults with aneurysmal subarachnoid hemorrhage.58 Laser Doppler (LD) flowmetry is another technique in which a fiberoptic probe is placed onto brain parenchyma and detects light reflected by red blood cells to derive flow velocity. CBF is largely driven by the CPP.

Cerebral autoregulation (CA) refers to the maintenance of CBF across a range of the MAP. This is mediated by sympathetic innervation and local tissue factors, such as ischemia. When CA is intact, a gradual increase in blood pressure may cause cerebral small vessel arterioles to vasoconstrict, leading to reduced intracranial arterial blood volume, and subsequently a reduction in ICP. When there is poor CA, cerebral small vessel arterioles lack such a response, and an increase in blood pressure may lead to cerebral small vessel arteriole vasodilation, a net increase in intracranial arterial blood volume, and a subsequent rise in ICP. (Fig. 3). A common method of determining CA is by use of the pressure-reactivity index (PRx). The PRx is a model-based index derived from the relationship of ICP with MAP and assumes that changes in intracranial blood flow are being driven fundamentally by changes in cerebrovascular motor responses.59 Because the PRx is calculated from a Pearson linear coefficient between ICP and MAP, it is reported with values between -1 and +1 (Fig. 4). A positive result is postulated to represent a passive non-reactive cerebral vascular bed; a negative result is postulated to represent a normally reactive vascular bed. When PRx is plotted using an error bar over a range of CPP, an ‘optimal cerebral perfusion pressure’ can be formulated, representing the lower threshold at which CPP may be maintained above to maintain a patient in an optimal autoregulatory state (Fig. 4).60 There is an increasing body of evidence suggesting that for pediatric TBI patients, elevated PRx values, suggestive of inefficient cerebral autoregulation, are associated with unfavorable outcomes. These same studies also demonstrate that when determining optimal autoregulatory states using PRx and other similar indices of CA, increased time with CPP values below the lower limit of CA appears associated with unfavorable outcomes.61, 62, 63 This raises the possibility that targeting CPP toward the optimal autoregulatory state may improve outcomes, but prospective clinical interventional studies are needed to assess such a therapeutic strategy. The use of TCD for clinical assessments of CA in critically ill pediatric TBI patients have been described for two separate pediatric neurocritical care centers, either as part of an integrated MMM analysis with concordant ABP and ICP monitoring64 or through dynamic and static testing performed through a cerebrovascular laboratory.65 In both studies, clinical assessments of cerebral autoregulation was feasible, safe and led to changes in hemodynamic targeting for specific patients.

PbtO2: PbtO2 is measured as the partial pressure of O2 in brain tissue, obtained by either a polarographic method or the oxygen diffusion across a semi-permeable membrane. PbtO2 is most used in trauma as it requires an invasive catheter. In adults, it is typically placed along with ICP and cerebral microdialysis catheters. The recent BOOST-II trial showed that in adult TBI patients, there was a significant reduction in cerebral hypoxia burden with therapy guided by PbtO2- and ICP monitoring as compared to ICP-based therapy alone as well as a trend towards improved functional outcome.66 In pediatric TBI guidelines, there are recommendations to maintain PbtO2 levels > 10 mmHg when PbtO2 monitoring is used.67 There has been an increased mortality reported when there were lower PbtO2 values on Day 1 and Days 3–5, and lower PbtO2/PaO2 ratios on days 1, 2, and 5 in those that died.68 For MMM, PbtO2 has been used along with the alpha-delta ratio (ADR) as an indicator of cerebral hypoxia.69 When the PbtO2 decreased below 10 mm Hg, there was a reduction in the ADR. Just as with using ICP monitoring, CA can be measured using PbtO2 monitoring using the oxygen reactivity index, a model-based index evaluating the relationship of PbtO2 with MAP.70 Evidence exists that PbtO2 values can relate differently to carbon dioxide reactivity pediatric TBI patients, increasing the value of monitoring for such changes using MMM.71

Microdialysis: cerebral microdialysis (CMD) is a real-time measurement of cerebral interstitial space extracellular fluid analyte concentrations, using a dialysis catheter with a semi-permeable membrane typically placed in the white matter. The usual analytes measured are glucose, pyruvate, lactate, glycerol, and glutamate. The term “metabolic crisis” has been applied to elevations in the lactate/pyruvate ratio (LPR). There are few CMD studies in pediatric TBI that show associations of anaerobic metabolites with poor outcome.72, 73, 74 Even in adult NCC, the grade level for using glucose, glutamate, glycerol, the LPR and lactate in NMN is C (low).75

Intracranial EEG (iEEG): iEEG, using a depth electrode (DE) has been used in acute brain injury, including trauma and aSAH. Electrographic seizures are seen on IEEG that do not show up on scalp recording.76 The DE should be placed into the tissue at maximal risk.

Waziri placed an 8 contact DE in 16 patients undergoing invasive brain monitoring along with conventional CEEG monitoring in 14 patients with brain trauma76 Clinically important findings were detected in 12/14, including electrographic seizures in 10 and acute changes related to secondary neurologic injury in 2 patients, one with ischemia and 1 with hemorrhage. In the 10 with electrographic seizures detected on DE, no surface EEG correlate was seen in 6, rhythmic delta activity (RDA) in 2, and there was intermittently correlated scalp electrographic seizure activity in 2 patients. In the 2 patients with secondary injury, EEG attenuation was seen 2–6 h before changes in other parameters and 8 h before clinical deterioration. This attenuation mirrors the change in the alpha-delta ratio detected by CEEG monitoring that is seen hours prior to clinical vasospasm in patients with subarachnoid hemorrhage.20,21,77 In a follow-up multicenter study, Vespa at al reported seizures in 62% of patients with severe TBI; seizures were seen only on the intracortical or invasive electrodes in 43%.78 A metabolic crisis detected by microdialysis was more severe when epileptiform activity occurred.

In children, intracranial EEG was used in 11 children with severe TBI, using a 6 contact DE in conjunction with surface CEEG. Comparative QEEG analysis between intracranial and scalp EEG was done including total power, alpha percentage, and alpha-delta power ration to CPP. Three patients had epileptiform activity on DE, two had epileptiform abnormalities noted exclusively on DE, and one patient had seizures originating on DE before spreading to surface EEG. Epileptiform abnormalities were associated with stroke or malignant cerebral edema.79 There were significant positive associations between the ADR to CPP in 9/11 patients with an increased strength of association on the IEEG.

Cortical spreading depressions (CSDs): represent transient and terminal cortical depolarizations.80 CSD can be seen on conventional surface EEG or invasive EEG monitoring, including subdural grids, strips, or DE. So far, CSDs remain a research tool.81 CSDs are waves indicating electrophysiological hyperactivity, followed by slowly spreading cortical inhibition at a rate of 2 to 6 mm/minute. CSDs correlate with decreased scalp EEG amplitudes.82,83 CSDs have been associated with worse outcomes in TBI, increased stroke volume, and with delayed cerebral ischemia in aSAH.81

Pediatric MNM started with severe traumatic brain injury (TBI),84,85 but its use has now expanded. The PNCRG conducted a survey on the clinical use of neuromonitoring devices, integrative MMM capabilities, and neuromonitoring infrastructure among its 70 participating institutions with 52 responding. The indications for use, devices and infrastructure vary by institution and non-invasive modalities are used more often than invasive ones. Invasive ICP devices, electrophysiology, and intermittent TCD were done in all 52 institutions; CEEG was done in 50 (90%) whereas QEEG was done in only 20 (38%). Only eight institutions used a MMM system that integrated and synchronized data.15

The use of MMM has undergone a Delphi Consensus Process to develop practice standards.11 Consensus was determined in several areas. Clinical considerations for the need of MMM : level of consciousness, underlying disease or diagnosis, potential risk for secondary brain injury or deterioration, structural imaging findings, confounding factors clouding the neurologic examination, desire to understand pathophysiology underlying brain dysfunction, guiding individualized management decisions, informing goals or thresholds for targeted management, abstaining from or de-escalating a therapy or treatment that might cause harm. Minimum necessary monitors and devices: for the use of ICP, CPP, Pbto2, CEEG, pupillometry, arterial blood (ABP), electrocardiography or cardiac telemetry, continuous core body temperature, end-tidal carbon dioxide (EtCO2), and plethysmography. Monitors and devices for specific contexts of use: brain tissue hypoxia, cerebral ischemia, autoregulatory dysfunction, acute coma and disorders of consciousness, post-cardiac arrest hypoxic-ischemic brain injury, metabolic crisis or mitochondrial dysfunction, seizures or IIC patterns, spreading depolarizations, and intracranial hypertension. Educational Formats: There is also the need for the training and expertise needed to understand and interpret NMN is acquired through hands on workshops or seminars, clinical practice or bedside teaching, development of a core curriculum, supervised performance, and demonstration of procedural competency.11

The EEG is a composite of various frequencies with resultant waveforms relating to CBF.86 Normal CBF is approximately 50 ml/100 g/min. With ischemia, as CBF decreases, EEG changes occur: as CBF decreases to 25–35 ml/100 g/min, there is a loss of the faster frequencies; at 18–25 ml/100 g/min, increased slowing occurs (4–7 Hz), that may be rhythmic; at 12–18 ml/100 g/min, there is a further increase in slower frequencies (1–4 Hz); and < 10–12 ml/100 g/min, there is EEG suppression with loss of electrocerebral activity (Fig. 5). Therefore, sequential changes in EEG may indirectly indicate decreased CBF. However, these EEG changes, such as slowing, are non-specific and independent of etiology, and may also be related to underlying brain injury, medication effect or treatment. Slowing can be focal, multifocal, or diffuse, like the analysis of the neurologic examination.

EEG changes can be used to identify neurologic deterioration and permit earlier intervention. New-onset seizures in a critically ill patient could signify a change, such as stroke, especially in the newborn.87 The occurrence of new onset spikes or sharp waves, or periodic features, or a sudden change in EEG background, especially if lateralized or focal, may indicate neurologic deterioration, such as new focal lesion. As mentioned, a loss of faster frequencies is the first change occurring in ischemia.86 Generalized EEG discontinuity can indicate increased intracranial pressure and can occur prior to clinical deterioration88 or herniation.89,90 This is especially so in the patient with known increased ICP, in whom EEG changes may indicate increasing ICP, such as EEG suppression during a plateau wave91,92 or with increasing intracranial pressure.93 Generalized rhythmic delta activity may also occur with a plateau wave.94 These EEG changes may occur before a clinical change.89,92 Rhythmic slowing on EEG associated with alteration of awareness with decreased responsiveness, activity mimicking seizures, may also occur with increased ICP.95 The disappearance of a pattern, such as triphasic waves, may occur with elevated ICP and reappear when ICP is treated.96 The worsening of the EEG background, such as the development of delayed EEG suppression after cardiac arrest, carries a poor prognosis.97 For monitoring purposes at Texas Children's Hospital, EEG changes that may herald neurologic deterioration are in Table 1. We refer to these “actionable” changes set “notification criteria for each individual patient: if a specified actionable change is detected by our monitoring team, the clinical teams are immediately notified.

EEG waveforms (patterns) may be ictal, indicating an overt electrographic seizure, or inter-ictal, indicating the non-ictal state. The American Clinical Neurophysiology Society (ACNS) has developed standard Critical Care EEG terminology for the interpretation of CEEGs in the ICU, now in its third version.18,98,99 Electrographic seizures are defined as having at least a 10 s duration with clear evolution in frequency, morphology, or location or continuous epileptiform discharges > 2.5 Hz, periodic discharges, and RDA.18 The IIC, identified best by CEEG monitoring, are predominantly rhythmic or periodic EEG waveforms100, 101, 102, 103 representing activity with a less certain significance than overt electrographic SE but clearly between the ictal and inter-ictal states. The IIC is a dynamic pathophysiological state suggesting an increased risk of evolution to the ictal state, with a greater potential for neuronal injury.18 The clinical question regarding IIC waveforms is whether they indicate a potential for secondary neuronal injury, which mandates aggressive therapy, or are IIC waveforms an epiphenomenon, occurring secondary to the underlying insult without causing ongoing neuronal injury. Or could they be both? Aggressive therapy is indicated if there is potential neurologic injury.

Convulsive SE may be easy to detect in the ICU because of the associated motor movements but NCSE is not. However, not all paroxysmal motor movements or autonomic changes in ICU patients may result from electrographic seizure activity and CEEG may identify the non-epileptiform nature of a suspicious clinical event. Young et al. used CEEG monitoring to identify the primary and secondary criteria needed for NCSE103; these were subsequently modified into the “The Salzburg Criteria”: 1) epileptiform discharges (EDs) > 2.5 Hz, 2) EDs < 2.5 Hz or rhythmic delta/theta activity (> 0.5 Hz) and one of the following: EEG and clinical improvement after IV AEDs, 2) subtle clinical ictal phenomenon during the suspicious EEG patterns, or a typical spatiotemporal evolution.104 Treatment decisions in patients with a known chronic epilepsy and epileptic encephalopathy using these criteria should be made with caution.105

Table 2 lists these IIC EEG patterns, ranked from those with the greatest epileptic potential to the lowest. A major criterion for identifying an ictal pattern is the frequency of the rhythmic or periodic waveform. The higher the frequency, the more likely that the pattern represents an ictal pattern: 3 Hz for Young103 and 2.5 Hz for the Salzburg criteria.104 Periodic discharges (PDs) are stereotyped, repetitive discharges occurring at regular intervals.18,106 PDs are classified as either generalized periodic discharges (GPDs) when they are generalized, synchronous, and have relatively equal amplitudes across homologous brain regions18 or lateralized periodic discharges (LPDs) when maximally involving one hemisphere. These were previously called generalized periodic epileptiform discharges (GPEDs) or periodic lateralized epileptiform discharges (PLEDs), with the descriptor “epileptiform” removed because these patterns may not be associated with seizures. There are PD subtypes: LPDs plus (+) is a modified term describing LPDs with superimposed rhythmic fast activity and/or sharp waves; bilateral independent periodic discharges (BIPDs) represent asynchronous periodic discharges occurring over both hemispheres.18 It is beyond the scope of this review to discuss these in more detail.

The relationship among etiology, association with seizures, and prognostic utility of PDs is a major question in NCC. GPDs occur in diverse settings, whereas LPDs are most associated with acute structural lesions.107,108 GPDs may be associated with seizures.106 In children with refractory SE, GPDs occurring after the control of SE are still considered in an active epileptic state, and have a lower mortality compared to adults.109 LPDs also correlate with clinical seizures107,108 and LPDs plus are more likely associated with clinical or electrographic seizures than LPDs alone.110 A recent multicenter cohort study found that LPDs were associated with seizures at all frequencies with and without a plus modifier.111 Recent exploratory work evaluating early post-traumatic seizures in children with MMM suggests that certain characteristics of seizures, such as spectral edge frequency and peak value frequency, may have a relationship with changes in cerebral hemodynamics such as intracranial hypertension and respiratory rate.112

RDA represents activity in the delta frequency with a relatively uniform morphology, referred to as “monomorphic delta,” and a frequency of 4 Hz or less, without an interval between consecutive waveforms.17 RDA can appear as either generalized (GRDA) or lateralized (LRDA). Like PDs, RDA may use modifiers including fast activity (RDA +F), sharply contoured waveforms (RDA +S), or both (RDA, +FS).18 Intermittent rhythmic delta activity (IRDA) occurs when this activity is repetitive but not constant.

The addition of QEEG and DTA to CEEG has greatly added to our monitoring capabilities.19 DTA is especially useful in identifying ictal versus interictal periods in the EEG. The use of QEEG is rapidly increasing. The 2022 PNCRG survey showed that QEEG was used in 38 % of responding institutions (20/38).15 A subsequent survey by a new organization, the Pediatric Quantitative EEG strategic taskforce (PedQuEST) in 2023 showed that now 22 of 39 (56%) institutions now use QEEG. The use of QEEG is related to having a dedicated NCC service, > 200 NCC consults per year, > 1500 ICU admission per year and > 4 ICU EEGs per day.113

QEEG has been instrumental in the identification of cyclic patterns, either the cyclic alternating pattern of encephalopathy, called CAPE18 (Fig. 6), or cyclic seizures114 (Fig. 7), a sub-type of SE. Cyclic seizures, occurring in both adults and children, is detected easily by DTA displays whereas the cycling is difficult to appreciate when scanning conventional CEEG. Cyclic seizures represent a discrete EEG pattern seen during SE, constituting NCSE, with seizures recurring in a cyclic pattern, typically with a regular interval, usually associated with an acute symptomatic etiology.114 Although not typically included in the IIC, cyclic seizures should be considered a variant of SE. We have seen cyclic seizures in 28 of 279 (10%) of pediatric SE.115 Like cyclic seizures, CAPE occurs in adults and children, with its significance not yet known.18,116 QEEG has also been helpful in identifying features reflective or predictive of acute brain injury in children. Using MMM, changes in EEG background activity have been linked to brain tissue hypoxia and predictive of intracranial hypertension, inefficient cerebral autoregulation, and poor functional outcomes in pediatric TBI patients.64,117 QEEG amplitude asymmetry across the cerebral hemispheres have been demonstrated to be independently associated with acute brain injury in a prospectively collected cohort of pediatric ECMO patients.37

A key component to the clinical use of MMM is to improve clinical care on an individualized basis based upon a patient's personalized physiologic profile. For pediatric TBI patients, a single-center cohort study demonstrated that standardizing the reporting process resulted in improved clinical decision-making, using it for timing of neuroimaging, ICP monitoring discontinuation, timing of extubation trials, body repositioning, paralytic therapy, hyperosmolar therapy, pentobarbital therapy, cerebral autoregulation testing, CPP threshold adjustment, PaCO2 threshold, neurosurgical interventions, and neurologic prognostic decisions.64 Current TBI practice recommendations support the evaluation of PRx in assessing management strategies during intracranial hypertension, although specific pathways to act on such information have not been established.

MMM tools, including functional neuroimaging, may help to differentiate the ictal from the interictal state (Fig. 8). With IIC patterns, such as PDs, PET scans obtained in 18 patients demonstrated that hypermetabolism was common and predicted SE.118 SPECT has shown increased CBF in regions in which PDs were actively occurring, suggesting an ictal pattern.119,120 A high correlation with the location of elevated rCBF to PDs was found in patients who underwent SPECT following convulsive seizures; regions with decreased rCBF had little correlation with PDs.121 MR perfusion was done in two patients with LPDs; one had hyperperfusion and clinical improvement with aggressive treatment, and the other without perfusion asymmetries had spontaneous resolution of LPDs over several days without aggressive treatment.122

Witsch et al used MMM, including CEEG and iEEG, to study the relationship between PD frequency and CBF and PbtO2 in order to determine if there is a threshold frequency for tissue hypoxia after aSAH.123 The median PbtO2 was 23mm Hg without PDs, compared with 16 mm HG at 2.0 Hz and 14 mm Hg at 2.5 Hz. The rCBF also increased with increasing frequency of PDs. This study suggests that rCBF increases with PDs to compensate for an increased metabolic demand, but higher-frequency PDs may be inadequately compensated and lead to brain tissue hypoxia. Subramaniam et al correlated LPD frequency with glucose metabolism using CEEG and PET scanning. LPDs were stratified into those < 1 Hz, 1 Hz, and > 1 Hz. Metabolism increased by 100% for LPD frequencies at 1 Hz and by 309% for LPD frequencies > 1 Hz.124 In SAH, the time spent in a “supra-optimal” CPP range was higher in patients with seizures or in the IIC, and the supra-optimal CPP occurred an hour before the seizures or the IIC patterns and increased during these.125 A prospective study in a cerebral metabolic crisis of the relationship between PDs and cerebral metabolism in patients with severe TBI assessed time-locked neurochemical responses to PDs found that elevated LPRs were more likely to occur in seizures and PDs as compared to interictal epochs.78 These studies all suggest that in acute brain injury, increases in CBF or metabolism may be associated with a more pathological relationship with PDs.

Through standardized expert recommendations involving MMM, there is consensus that the work involved in MMM is beyond the scope of an intensivist working at the bedside and would require expert personnel and services beyond the scope of the bedside clinician.11 Furthermore, the training of both intensivist and electrophysiologists may be insufficient toward closing the gaps of how real-time cerebral hemodynamics relate to electrophysiology. Ultimately for MMM to mature as a service line of care in pediatric NCC, increased education regarding neuromonitoring techniques and future research to identify secondary brain insults and management against them are needed. Like the evolution of ICU EEG monitoring, methods of developing standard processes for visual analysis, exclusion of artifact, and recognition of both normal and pathologic patterns will need to be described and validated to increase adoption of MMM strategies toward optimizing clinical care, and further outcome-based studies may also assist in appraising the value of this line of work.