Fig. 23.1
The EEG of a nonconvulsive seizure. This is an EEG (settings: 7 uV; low-frequency filter: 1 Hz; high frequency filter: 70 Hz; and notch filter turned on), from an 87-year-old man with atrial fibrillation, on oral anticoagulant agent (apixaban) who was admitted with a traumatic right-sided subdural hematoma. He had 8 electrographic seizures in the first 7 h of EEG recording. Seizures began with 0.5 Hz delta slowing with periodic spikes in the right parieto-temporal region (a), which then increased gradually to 3 Hz and spread to the entire right hemisphere (b beginning 31 s after the end of a). This was associated with non-specific bilateral arm movements, none of which was recognized as a seizure at the bedside
Table 23.1
Criteria for the diagnosis of nonconvulsive status epilepticus
EDs > 2.5 Hz, or |
EDs ≤ 2.5 Hz or rhythmic delta/theta activity (>0.5 Hz) AND one of the following: |
EEG and clinical improvement after IV ASDsa, or |
Subtle clinical ictal phenomena during the EEG patterns mentioned above, or |
Typical spatiotemporal evolutionb |
In addition to diagnosing NCSz, prolonged monitoring is also necessary for monitoring the efficacy of treatment, especially when using anesthetics, and for guiding treatment for NCSzs and NCSE. The significance of NCSz and NCSE in terms of the effects on cognition, neurologic function, development of epilepsy, and other morbidity and mortality will be discussed in separate chapters. Suffice to say there is extensive and growing evidence (though no prospective randomized trials of treatment versus no treatment) that NCSE or a high burden of some types of NCSz is associated with worse neurologic outcomes, including in cognitive function and later epilepsy. This chapter reviews indications for continuous EEG in SE (particularly for NCSE) under different settings, duration of monitoring, diagnosis of NCSz, data review, and cost effectiveness of C-EEG in SE.
Yield of Continuous EEG
Several studies over the years have established clearly the necessity and, higher yield, of prolonged EEG monitoring in critically ill patients for diagnosing NCSz and NCSE [8, 9]. Routine EEGs, even when performed repeatedly, have a lower yield for detecting seizures, particularly NCSz [10]. In a retrospective study, the monthly rate of diagnosis of NCSz in ICU patients increased significantly after introduction of C-EEG [9]. In another study examining the utility of C-EEG in a neurological-neurosurgical ICU, all patients underwent a routine 30-min EEG prior to continuous video EEG monitoring, with a mean duration of 2.9 days (range 1–17 days). In the 105 patients, NCSz were detected in 26.7% with C-EEG, as compared to 11.4% with routine EEG [8].
Who Should Be Monitored with Continuous EEG?
The purpose of C-EEG in critically ill patients admitted to ICUs can be divided into two broad categories: (a) diagnosis of NCSz and NCSE and, (b) assessing the efficacy of treatment for seizures and SE.
Diagnosis of Nonconvulsive Seizures and Nonconvulsive Status Epilepticus
Nonconvulsive seizures (NCSz) are common among critically ill adult patients, but the percentage of patients with NCSz varies depending on the population studied. Across different studies performed in all ICUs, the prevalence of NCSz has been between 8 and 38% (Fig. 23.2) [11–35]. Although there is a high incidence of NCSz among patients with acute brain injury, even patients without neurologic injury but with serious systemic illness, especially sepsis, are at risk of NCSz. A single center retrospective study examined all comatose adult patients admitted to the ICU with no overt clinical signs of seizures; patients with prior clinical signs of seizures or SE were excluded [33]. Of 236 patients, 19 (8%) were found to have NCSE on a 30-min EEG. Subsequently, another large single center retrospective study utilized C-EEG for all patients of all ages with unexplained alternation in consciousness in hospital settings including both wards and ICU [16]. Of 570 patients, 110 (19%) had seizures, and 92% were exclusively nonconvulsive. About half of the NCSzs qualified as NCSE (59/105, 56%). There was a higher incidence of NCSzs among the patients admitted to the Neuro ICU (61%) compared to those in other settings. In this study, younger age, coma at onset, history of epilepsy, convulsive seizures prior to monitoring, and a burst suppression pattern on the EEG were independent predictors of NCSz [16]. Another study found that profound alteration of consciousness, oculomotor abnormalities, and previous risk factors for epilepsy were significantly associated with NCSE [36]. A recent prospective study from a single center found a similar rate of NCSz or NCSE—21% among all patients (n = 170) with altered mental status in a Neuro ICU [37]. They also found that subtle oral twitching and eye deviation was associated with 50% of cases of NCSz and NCSE (though this is likely to be due to selection bias related to who undergoes C-EEG). Among these patients, those with a previous history of epilepsy, brain tumors, and meningitis or encephalitis were at the highest risk of NCSz.
Fig. 23.2
Incidence of nonconvulsive seizures (NCS) in different populations of critically ill children and adults. The confidence intervals were not reported by the studies, but were calculated based on the number of subjects in the study and the proportion of patients in whom nonconvulsive seizures were detected. Data are derived from: a Abend et al. [12], b Abend et al. [13], c Arndt et al. [14], d Carrera et al. [15], e Claassen et al. [16], f Claassen et al. [17] g Crepeau et al. [18], h Gilmore et al. [19], i Mani et al. [20], j O’Connor et al. [21], k O’Neill et al. [22], l Oddo et al. [23], m Payne et al. [24], n Ronne-Engstrom and Winkler [25], o Schreiber et al. [26], p Topjian et al. [27], q Vespa et al. [28], r Vespa et al. [29], s Westover et al. [30]. Adapted from Osman et al. [11], with permission
Convulsive Status Epilepticus. Often, patients remain obtunded after ‘successful’ treatment of convulsive status epilepticus (CSE), and it is difficult to differentiate a post ictal state from ongoing seizure activity. Studies have shown that even after successful treatment of CSE, approximately 43–48% of patients continue to have NCSz, and 14% have NCSE, on subsequent 24 h of C-EEG recording [6, 38]. Therefore, C-EEG is indicated in patients with convulsive seizures who do not show a clear improvement in alertness within 10 min of adequate treatment or who do not return to their functional baselines within 60 min of treatment with anti-seizure drugs (ASDs) [3, 39].
Acute Brain Injury. NCSz are a common occurrence in patients with acute neurologic injury, so C-EEG is indicated in all such patients with alteration of consciousness. Earlier studies from patients admitted to the Neuro ICU showed a high incidence of NCSz—up to 34%, with 76% of these being NCSE [40]. The variability in incidence of NCSz in these patients is partly attributable to methodologic variations in the nature of the studies, the underlying etiology, duration of monitoring, and possibly the use of ASDs and sedative medications.
Traumatic Brain Injury. All patients with traumatic brain injury (TBI) are at high risk of developing seizures, particularly those with depressed skull fractures, any intracranial bleeding, cortical contusions, or penetrating brain injuries. The estimated incidence of NCSz in TBI patients has varied from 3 to 53%. In 94 consecutive patients with moderate to severe TBI (defined as GCS of <12), seizures occurred in 22, and 52% of the seizures were exclusively nonconvulsive. The incidence of NCSE in the overall cohort was 6.3% [41]. The mean duration of monitoring was 7.5 (±4) days. The seizures occurred despite treatment with prophylactic ASDs. Thus, C-EEG is recommended for patients with moderate to severe TBI with alteration in consciousness, and in any TBI patients with penetrating injury or intracranial hemorrhage of any type.
Subarachnoid Hemorrhage. Patients with nontraumatic, aneurysmal subarachnoid hemorrhage (SAH) constitute another high-risk group for NCSz, with an incidence of NCSz of approximately 3–31% [42, 43]. Most of these studies were performed in poor-grade (more severe) SAH patients. Older age, female sex, higher Hunt and Hess grade of SAH, thick cisternal hemorrhage, need for ventriculostomy, cerebral edema on initial CT scan, and structural lesions were associated with a higher risk of NCSE [42, 43].
Intracerebral Hemorrhage. Early clinical seizures are common in patients with nontraumatic intracerebral hemorrhage (ICH); the incidence ranges from 4 to 14% [44–46]. The incidence of NCSz is even higher when C-EEG is carried out on these patients. In 102 consecutive patients with ICH who underwent C-EEG monitoring, seizures occurred in 31%, over half of them (18/32) exclusively nonconvulsive, and 7% (7/102) qualified as NCSE [17]. These studies may have overestimated the occurrence of seizures as only the patients who underwent C-EEG monitoring were studied, rather than the total ICH population. Nevertheless, Vespa and colleagues reported the incidence of NCSz to be similarly high at 28% in patients with nontraumatic ICH, even when consecutive patients were studied without a selection bias [29]. Larger hematoma volume has been consistently demonstrated as highly associated with increased risk of seizures [17, 46]. One study also indicated that cortical involvement is predictive of increased occurrence of seizures [45]. The study by Vespa and colleagues confirmed the same findings, but another study found an association in univariate analysis only (possibly underpowered for multivariate analysis) [17, 29]. Periodic discharges occur with a higher frequency in patients with hemorrhages in close proximity (<1 mm) to the cortical surface [17], which in turn may increase the risk of seizures. Whether NCSz contribute to increasing cerebral edema and midline shift in patients with ICH remains unclear. Nonetheless, NCSz have consistently been an independent predictor of poor outcome [17, 42]. As with other critically ill populations, 56% of patients with NCSz had the first seizure detected within the first hour of C-EEG monitoring, and 94% within 48 h [17]. Based on these findings, C-EEG monitoring is recommended in all patients with ICH and impaired consciousness, particularly those with larger or expanding hematomas and those with hemorrhages in close proximity to cortex, including the insula (<1 mm).
Acute Ischemic Stroke. The prevalence of early seizures among patients with acute ischemic stroke (within 7 days) is considerably less compared to that with hemorrhagic stroke and has been estimated to be around 2.5% [47, 48]. (This population of solely ischemic stroke has not been studied extensively with C-EEG.) In a prospective study of 232 patients, of whom 177 had acute ischemic stroke, 4.3% had SE (10% being NCSE), all within the first 24 h of stroke onset [48]. The incidence was similar among ischemic and hemorrhagic stroke. Another cohort of 100 consecutive patients (91 with ischemic stroke and 9 with hemorrhagic stroke) who were admitted to a stroke unit was studied with C-EEG for a mean duration of 17.5 h [49]. Two patients (2%) had NCSz, both in the acute ischemic group. Therefore, the evidence for utilization and yield of C-EEG in acute ischemic stroke patients is poor, and these patients probably remain a low risk group. Accordingly, C-EEG for acute ischemic stroke should probably be limited to those with large infarcts involving cortex and with altered consciousness, or those with fluctuations or deficits out of proportion to the infarct size and location.
Infections of the Central Nervous System. Infection of the central nervous system (CNS) is an established risk factor for seizures, particularly with viral infections (nearly 50% in patients with herpes encephalitis) as opposed to bacterial infection (15%) [50, 51]. A retrospective study monitored patients with suspected CNS infections for a mean duration of 2.5 days, excluding all patients who had neurosurgical procedures [15]. Of the 42 patients, 14 (33%) had seizures, of which 64% were nonconvulsive. In this cohort, 8 of 42 patients (19%) had NCSE. This may be an overestimate because the study was performed in selected patients and may not have represented the entire population with CNS infections. In this study, 67% of the population had clinical seizures prior to EEG, but there was no association between seizures prior to EEG and electrographic seizures during C-EEG. Further studies are required to assess the seizure frequency more accurately by studying a larger, unselected group of patients.
Brain Tumors. Neoplasms of the brain are an established risk factor for seizures and epilepsy. In a recent prospective study of critically ill patients in the Neuro ICU, those with a history of brain tumors had one of the highest associations, of all causes, with NCSz or NCSE [37]. Very infrequently, NCSE can be the presenting manifestation of metastatic and primary brain tumors [52, 53]. In a cohort of 259 patients with brain tumors who underwent C-EEG monitoring, 2% had NCSE, of which half were purely nonconvulsive [54]. Treatment resulted in resolution of NCSE in 22 of 24 patients (92%), indicating not only the high incidence of NCSE in this population, but also that it is easily treated, making it important to monitor these patients with C-EEG.
Patients With Systemic Illness and Altered Mental Status Without Neurologic Injury. Seizures are common in critically ill patients who do not have a primary brain insult. Retrospective study of all patients in the medical ICU (majority with sepsis) undergoing C-EEG and without a known neurologic injury found that 10% had electrographic seizures, 67% of which were purely nonconvulsive; it was unclear how many qualified as NCSE [23]. Also, the incidence of periodic discharges was 67% in this population. In a prospective study of 100 patients with sepsis and altered mental status, the incidence of NCSz was similar (11%), with all seizures considered definite or possible NCSE; 1-year follow up showed that none of the 35% of patients who survived developed subsequent unprovoked seizures [19]. There have been conflicting results concerning outcomes and any association with mortality for these patients with NCSz, which remain areas for research [19, 23].
Metabolic derangement, sepsis, and delirium are common in the postoperative period and make patients more susceptible to seizures. Kurtz and colleagues studied 154 patients undergoing C-EEG in a surgical SICU, mostly (65%) after abdominal surgery [55]. All patients with primary neurologic problems or neurosurgical intervention were excluded. With a minimum of 12 h of monitoring, NCSz were observed in 16% of patients, including 5% with NCSE; 29% had periodic discharges. Multivariate analysis identified coma and clinical seizures prior to C-EEG as predictors of NCSz. Kamel and colleagues studied all patients admitted to the MICU and SICU (excluding those with primary brain insults) and found NCSz in 11%, but the incidence of NCSE was not reported [56].
Cardiac Arrest Requiring Therapeutic Hypothermia and Pharmacologic Paralytics. Seizures are seen in 10–30% of patients with cardiac arrest who are treated with therapeutic hypothermia (TH) [57–59]. Seizures occur during the cooling phase, not just during the rewarming phase of TH [59]. During the cooling phase, patients are often paralyzed pharmacologically to prevent shivering and to maintain temperature goals—potentially masking the motor manifestations of seizures. C-EEG is thus required to detect NCSz and facilitate appropriate treatment. Cardiac arrest survivors also have frequent myoclonic seizures and myoclonic SE, and C-EEG is also required to differentiate epileptic from non-epileptic or subcortical myoclonus, although both are treated similarly. The neurointensive care guidelines from European Society of Intensive Care Medicine (ESICM) recommend C-EEG during the TH phase and for 24 h after rewarming for detection of NCSz [39]. Some institutions, including ours, have a protocol to monitor the patients continuously for 72 h beginning as soon as possible, but always within the first 12 h after arrest.
Patients With Routine EEG Showing Malignant Periodic Patterns. Several different periodic patterns can be found on the EEGs of critically ill patients. Studies in adults and children have shown that generalized periodic discharges (GPDs) are strongly associated with NCSz and NCSE [60, 61]. Similarly, lateralized periodic discharges (LPDs) (often labeled periodic lateralized epileptiform discharges, or PLEDs earlier) and lateralized rhythmic delta activity (LRDA) have a high association with acute seizures [62–64]. In a retrospective study, the incidence of seizures in patients with LRDA was 63%, similar to that associated with LPDs (57%) [63]. Patients who have these patterns on routine EEG should be monitored for 24–48 h due to the high risk (>50%) of developing seizures (mostly nonconvulsive). More details on periodic EEG patterns in critically ill patients are described in Chap. 5, “Periodic EEG Patterns”.
Patients With Abnormal Movements Or Other Subtle Signs of Possible Seizure Activity. Critically ill patients often have paroxysmal movements, such as tremors, myoclonus, clonus, paroxysmal abnormal movements of the mouth or eyes, posturing, and autonomic events that can mimic epileptic seizures. It is often difficult to determine whether these events are epileptic or not without continuous video EEG. One retrospective study examined all EEGs performed for observed movements in the ICU and found that 73% of the episodes were non-epileptic [65]. Therefore, video EEG is important in these patients, not only to make a diagnosis of seizures, but also to exclude seizures and prevent unnecessary use of ASDs.
Assessing the Efficacy of Treatment for Seizures and Status Epilepticus
The recent American Clinical Neurophysiology Society (ACNS) guideline for C-EEG in critically ill patients recommends that patients with previous CSE and those being treated for NCSE should be monitored for at least 24 h after cessation of all electrographic seizures. It is also recommended that critically ill patients being weaned off ASDs should be monitored for at least 24 h after stopping the medication [39]. For medications with long half-lives, longer monitoring is reasonable in selected cases.
Refractory status epilepticus (RSE) is defined as ongoing clinical or electrographic seizures despite adequate initial treatment with at least a benzodiazepine and one other appropriately chosen ASD [3]. Prolonged RSE is almost always nonconvulsive (even if it began as CSE), and all patients with RSE being treated with anesthetic doses of IV ASDs should be monitored continuously to assess treatment efficacy.
Duration of Monitoring
The recent ACNS consensus statement recommends monitoring for at least 24 h for critically ill patients with suspected NCSz [2]. In a retrospective study of 570 consecutive patients with unexplained alteration of consciousness, 88% had seizures detected within first 24 h of recording [16]. In the non-comatose patients, 95% of seizures were detected in 24 h of recording, and 98% after 48 h. In the comatose patients, however, only 80% of seizures were detected after 24 h, and 87% after 48 h. In comatose patients, more prolonged monitoring appears appropriate.
In places where resources are limited, early EEG findings may help determine the likely yield of prolonged monitoring. Several retrospective studies indicate that generalized slowing alone, with lack of epileptiform discharges on the initial EEG, predicts a low risk of seizures [30, 66, 67]. In one such study, no patients without epileptiform discharges in the first 4 h of recording developed seizures during subsequent monitoring—with a median monitoring duration of 24 h (range, 18–70 h) [66]. The same study, extended to 625 patients, found that if no seizures occurred after 16 h of C-EEG, the probability of recording a subsequent seizure decreased to <5% in patients with epileptiform discharges; in those patients without epileptiform discharges, this 5% threshold was reached after only 2 h [30] (Figs. 23.3 and 23.4).
Fig. 23.3
Ictal-interictal continuum. Demonstrates various EEG patterns, primarily based on frequency, depicted along the ictal-interictal continuum. The frequency of discharges (shown on the x–axis) has traditionally been the benchmark guiding the aggressiveness of treatment. This frequency-based division between interictal, continuum and ictal is arbitrary, conceptual, and does not take evolution of patterns into account. The evolution of EEG patterns can be subtle, especially when observing long epochs in critically ill patients, and it is often difficult to reach a consensus. Still, the presence of even subtly evolving patterns increases the possibility of them being ictal. If a clinical correlate is present with any of these patterns, it must be considered ictal by definition, regardless of the frequency. *At least 1 Hz with clear (unequivocal) evolution in frequency, morphology, or location is considered to be ictal. GCSE generalized convulsive status epilepticus, NCSE nonconvulsive status epilepticus, NCS nonconvulsive seizure, SIRPIDs stimulus-induced rhythmic periodic or ictal discharges. Adapted from Sivaraju and Gilmore [69], with permission
Fig. 23.4
Quantitative EEG and raw EEG on the ictal-interictal continuum. Patient is a 75-year-old woman with a history of cerebellar stroke and multiple vascular risk factors who presented with a change in mental status and was found to have sepsis. She continued to show the same impaired consciousness despite appropriate treatment of her infection and was metabolically and hemodynamically stable. The figure shows a gradual transition over an hour from interictal to ictal without a clear cut off, demonstrating the ictal-interictal continuum. The top portion of the figure shows 1 h of quantitative EEG, with time on the x-axis and frequency (0–20 Hz) on the y-axis (from Persyst 12™, Persyst Inc., San Diego, California). The top 2 rows are the rhythmicity spectrogram, showing gradually increasing rhythmicity, shown by the darkening tracing as time progresses to the right. The bottom 2 rows show the color density spectral array (CDSA), showing gradual evolution in the pattern over this hour (increasing delta and theta slowing as time progresses). The bottom panel shows the associated raw EEG for 10 s from 3 different time points. (EEG settings: sensitivity: 7 uV; low-frequency filter: 1 Hz; high frequency filter: 70 Hz; and notch filter turned off). a The initial epoch shows a background of quasi-rhythmic theta and delta slowing with only a hint of periodic discharges, interpreted as interictal. b There is gradual increase in frequency and prominence of generalized periodic discharges, on and off from 1.5 to 2.5 per second; this is considered to lie on ictal-interictal continuum. c The EEG shows the frequency of the GPDs at 2.5 Hz, evolving to 3–3.5 Hz rhythmic activity in the second half of this figure, and thus considered ictal, i.e., an ongoing seizure
Diagnosis of Nonconvulsive Seizure
Diagnosis of NCSz is often difficult or impossible without EEG as clinical manifestations may be very subtle. Even with C-EEG, the diagnosis can be difficult in critically ill patients. An electrographic seizure is defined as continuous rhythmic or periodic epileptiform discharges at >2.5 Hz, or epileptiform discharges that evolve in frequency, morphology, or location, and last at least 10 s (see Table 23.1). In critically ill patients, however, the background, interictal activity, and seizures are quite different as compared to those in otherwise healthy individuals. Thus, it can be difficult to distinguish ictal patterns (i.e., electrographic seizures) from interictal ones. Based on expert opinion, criteria have been proposed to define NCSz (see Table 23.1) [5]. Rhythmic discharges or periodic patterns >2.5 Hz are generally considered ictal, and those <1 Hz nonictal. Still, periodic patterns between 1 and 2.5 Hz (inclusive) may represent ictal or interictal phenomena based on different modifiers and are often considered to lie on an ictal-interictal continuum (IIC) (Figs. 24.3 and 24.4) [68, 69], but it is often challenging to make a diagnosis of NCSz based on EEG alone. In clinical practice, a trial of a rapid acting IV ASDs, e.g., benzodiazepines (BZDs) or less sedating IV ASDs such as phenytoin, valproate, lacosamide, or levetiracetam may be used to try to distinguish ictal from interictal activity. Because BZDs can abolish many kinds of abnormal electrical activity, the electrographic improvement must be accompanied by clinical improvement to make a definite diagnosis of NCSE. In one study, Hopp and colleagues found that in a cohort of 62 patients with suspected NCSE, a trial of IV BZD lead to clinical improvement in 35% [70]. All responders survived and had good functional outcomes. In clinical practice, however, the sedating effects of BZDs can confound the trial results, especially if large doses are used. Another recent study by O’Rourke and colleagues utilized BZDs or non-sedating ASDs in conjunction with clinical examination in order to identify ictal patterns [71]. They evaluated the response to BZDs and non-sedating ASDs in patients with sharp waves of triphasic morphology. Of 64 patients studied, 34.4% had a positive response (EEG and clinical improvement, though not necessarily immediately). Patients receiving non-sedating ASDs had a higher rate of definite or probable response compared to those receiving BZDs (26.7% vs. 18.9%), but the response was delayed in 20% of those given non-sedating ASDs. Our institution proposes a protocol using small, escalating doses of rapidly acting BZDs (e.g., midazolam) or non-sedating ASDs, and evaluating for change on the EEG and on clinical exam, i.e., resolution of the EEG abnormality and unequivocal improvement in clinical exam, or improvement of the EEG with return of previously absent normal EEG patterns, such as sleep spindles or a posterior-dominant rhythm (Table 23.2) [72].
Table 23.2
Anti-seizure drug trial for the diagnosis of nonconvulsive status epilepticus (From Hirsch and Gaspard [72], with permission)
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
