EEG is readily available in most centers and frequently used to evaluate the etiology of altered mental status in critically ill children. In critically ill children with altered mental status, there is a high incidence of nonconvulsive seizures (NCS) or nonconvulsive status epilepticus (NCSE) ranging from 7 to 47 %, depending on the study population [1]. In general, there is some disagreement whether NCSE and NCS, regardless of etiology, are both associated with worse clinical outcomes. Some studies evaluating short-term outcomes have found higher mortality and neurologic morbidity in patients with NCS or NCSE. There are fewer long-term outcome studies, but those performed report that there may be a lower quality of life or neurocognitive outcome in patients with NCSE, but not NCS alone. About 1/3 of critically ill patients who have seizures on cEEG will later develop epilepsy regardless of etiology. This is more commonly seen in patients with NCSE than NCS recorded during cEEG monitoring.

The challenge in many of these studies is to determine the effect size of NCSE or NCS on outcome. The underlying etiology likely plays the largest role in outcome, while NCS and NCSE likely play a smaller but potentially important role. Given the uncertainty about the impact of NCS and NCSE, it would be difficult to ignore these if encountered during cEEG. However, there is no data to support that aggressively treating NCS or NCSE leads to improvement in outcome. In fact, recent studies have demonstrated an increased relative risk for death in refractory status epilepticus patients receiving anesthetic therapy compared with those that do not. This has raised concerns that, at some point, the treatment may be more harmful than the continued presence of electrographic seizures. A randomized study to evaluate the impact of NCS and NCSE would be challenging to perform since equipoise between treatment and nontreatment arms may not exist based on observational studies.

cEEG background features can evolve over time and are dependent on factors such as sedation or other co-administered medications. cEEG features that were reported in early studies that had poor prognosis include electrocerebral silence, severe background attenuation, excessive background discontinuity, lack of reactivity, and periodic or multifocal epileptiform discharges.

Hypoxic-ischemic encephalopathy (HIE) is caused by a diffuse sustained hypoxic or ischemic brain injury in a variety of conditions, including perinatal asphyxia, cardiac arrest (CA), nonfatal drowning, severe hypotension, smoke inhalation, and carbon monoxide poisoning among others. HIE is one of the most common indications for a neurology consultation and cEEG monitoring in neonatal and pediatric intensive care units. HIE severity can vary from a mild case of a clinically transient postischemic confusional state with complete recovery and minimal or no irreversible brain tissue damage to a far more severe case of ischemic brain injury that clinically presents with multiple brain infarcts, deep comatose state, and cortical brain damage. Severe HIE is often fatal, but survivors of severe HIE often have lifelong chronic neurologic deficits. NCS and NCSE are commonly associated with HIE [1]. In the limited studies available, about 1/3 of patients with HIE who undergo cEEG will be found to have NCS.

The presence of epileptiform discharges and background activity was prospectively studied in children with HIE in one study using daily EEGs for 3 days. The background was classified as isoelectric, low voltage, slow, burst suppression, or discontinuous. Reactivity to sensory stimulation was also assessed. The study found that the presence of either discontinuous activity or epileptiform discharges had a positive predicted value (PPV) for poor outcome of 100 % (95 % CI of 56–100 %, P = 0.05 and sensitivity 27 % and 54 %, respectively), while lack of reactivity, defined as no change in frequency or amplitude of the background in response to external stimuli, or high-voltage slow waves less than 2 Hz had a PPV of 96 % (95 % CI of 76–100 %, P = <10⁻⁵) [2].

The background EEG features could be influenced by body temperature, for which discontinuity in such a setting has an unreliable significance. During deep therapeutic hypothermia, the EEG can demonstrate discontinuity followed by an isoelectric pattern; however, these EEG abnormalities are not reported with moderate hypothermia at 32–34 degrees centigrade.

In a prospective study of 35 children managed with a standardized clinical therapeutic hypothermia (TH) protocol after cardiac arrest, two samples of continuous EEG recordings were scored and categorized using a simple standardized method at onset of hypothermia and then after 24 hrs. EEG category 1 consisted of continuous and reactive tracings. EEG category 2 consisted of continuous but unreactive tracings. EEG category 3 included those with any degree of discontinuity, burst suppression, or lack of cerebral activity. During hypothermia, patients scoring a 2 or 3 on cEEG had an odds ratio of 10.7 and 35, respectively, for a poor outcome compared those scoring a category of 1. During normothermia, patients with a cEEG category of 2 had an odds ratio of 27 for a poor outcome compared to patients of category 1, and patients with a cEEG category of 3 had an odds ratio of 18 for a poor outcome compared with category 1 patients. During hypothermia, a score of 2 or 3 had a PPV for a poor outcome of 88 %, and during normothermia, the PPV was comparable for a poor outcome at 91 % with scores of 2 or 3. Given the similarity between normothermia and hypothermia, it is likely that the presence of a discontinuous or unreactive EEG cannot be explained by the decrease in body temperature alone and the presence of the EEG patterns during hypothermia can be useful in predicting poor outcome [3].

Since EEG is such a dynamic test, an EEG demonstrating prompt recovery over hours in comatose children has a better clinical outcome in terms of morbidity and mortality. The development of reactivity or normal background structures, such as sleep spindles, in comatose patients may suggest a lower degree of injury and a better prognosis.

Alpha coma (AC) is defined as an EEG pattern with an unreactive alpha (8–13 Hz) frequency as the primary EEG background feature in comatose patients (Fig. 1). AC is well recognized in adults and is associated with a poor prognosis following cardiac arrest. In children, there are limited studies regarding this pattern. The underlying pathophysiology of AC is thought to be due to interruption of reticulothalamocortical pathways from alpha-generating cortical neurons. This may be the same in children, but the immature brain may produce variable responses to such deafferentation. Clinically, the AC EEG patterns in comatose children can have more variety in background frequencies (alpha, theta, spindle, and beta frequencies) possibly due to the immaturity of the young brain [4]. Alpha coma may be seen in about 30 % of comatose pediatric patients with HIE. When encountered, establishing the presence or absence of any reactivity is important. In the few published reports of AC, the mortality is high, approaching 40–50 %. However, these numbers are generally better compared with the adult population and may support that children may tolerate HIE better than adult patients. The prognosis is highly related with the underlying etiology of the AC.

Postanoxic myoclonus status epilepticus (MSE) is a commonly encountered symptom in adults after an anoxic event. This is typically seen within the first 24 h after hypoxic or anoxic brain injury and is associated with a very unfavorable prognosis. The incidence is estimated to be between 30 and 37 % of comatose adult survivors after cardiopulmonary resuscitation. There may be an additive detrimental effect of anoxic neocortical damage after cardiorespiratory arrest and prolonged myoclonic seizures [5]. Postanoxic MSE is occasionally seen in pediatric patients; however, there are no literature reports in terms of incidence and mortality related with postanoxic MSE and children. The EEG changes in patients with postanoxic myoclonus vary, but the majority shows bursts of generalized spikes and polyspikes activity. Other patterns include burst suppression, generalized low-voltage slow activity, periodic complexes, and alpha coma. In some instances, it can be challenging to determine if the myoclonus is of cortical origin when significant muscle artifact is dominating the EEG. In these instances, a brief trial of a neuromuscular blocker, such as rocuronium, can suppress the myogenic artifact allowing better interpretation of the EEG (Fig. 2).

Postanoxic MSE should be differentiated from chronic postanoxic nonepileptic myoclonus or Lance-Adams syndrome (LAS), which presents in survivors of hypoxia. LAS refers to intermittent myoclonic jerks that are induced by movement, startle, and tactile stimulation. These myoclonic jerks may lead to postural lapses, ataxia, and dysarthria. There is no consistent electrographic seizure discharge on EEG during this myoclonus, differentiating it from epileptic myoclonus. LAS can cause severe disabilities, and treatment during rehabilitation therapy is especially challenging [6]. This is more commonly observed in adults, and reports on the incidence in children are lacking. In the absence of a hypoxic event, myoclonic status epilepticus may not necessarily be associated with high mortality. MSE encountered in patients with idiopathic generalized epilepsy is typically reversible with standard treatment (benzodiazepines, valproate), and a full recovery is expected.

In adults with HIE, the American Academy of Neurology (AAN) practice parameter cites three EEG patterns prognostic of poor outcome. The first is myoclonic status epilepticus within the first 24 h after the event. The second is generalized suppression of the background to less than 20 μV. Finally, a burst-suppression pattern with generalized epileptiform activity or generalized periodic complexes on a flat background is strongly, but not invariably, associated with poor outcome [1]. A similar practice parameter is not available for children; however, as reviewed above, similar patterns are typically associated with a poorer prognosis in children. Further studies are needed to elevate the significance of the data to the level where a practice parameter could be created [7].

Other complementary tools to cEEG could be helpful for early seizure identification and background changes, including quantitative electroencephalography (qEEG) such as density spectral array (DSA). These techniques may provide information about subtle changes, such as vasospasm, that are easily overlooked with raw EEG data. Since qEEG tends to display data on a compressed time scale, it can provide trend data at the bedside to help determine response to therapy and improvements in background activity that may portend a better prognosis.

Evoked potentials (EPs) are another neurophysiologic modality that can be used to help determine prognosis after HIE. EPs are reproducible time-locked signals generated in the central nervous system and neural structures in response to a sensory stimulus. Typically, these electrical signals are serially repeated and averaged to produce waveforms that can be analyzed for amplitude and latency from the stimulus. Between the stimulus and recording electrode, absent or delayed waves suggest an anatomical or functional interruption in the conduction pathway. Brainstem auditory evoked potentials (BAEPs), visual evoked potentials (VEPs), and somatosensory evoked potentials (SEPs) have been used to predict short-term outcome after severe acute TBI and HIE in comatose children. Some EPs, such as BAEPs, have the advantage of not being as sensitive to therapeutic hypothermia and sedative medications as cEEG. Somatosensory evoked potentials (SEPs) have demonstrated a strong predictive value for outcome, especially in cases of nontraumatic pediatric coma. The absence of the N20 response following median nerve stimulation bilaterally has been associated with an unfavorable outcome (death, vegetative state, or severe disability resulting in dependence) in nearly all patients following severe HIE with a sensitivity ranging from 63 to 75 %. These studies are typically performed for prognostic purposes 3 days after the injury to improve predictive value. There is limited data on BAEPs and VEPs in pediatric patients with coma. However, some studies have evaluated both of these modalities, generally in multimodality evaluations with SEPs and/or EEG. Absent VEPs (flash stimulation) or BAEPs suggest a poor prognosis; however, there is variable sensitivity and specificity. This can be improved by combining various neurophysiologic techniques. The absence of multimodality evoked potentials would suggest a greater risk for poor outcome and a higher sensitivity and specificity than any single evaluation.