Electrographic Seizures in the Critically Ill

Introduction

Electroencephalography (EEG)¹ has become an integral component of neurological care since it was first used by Hans Berger to record surface human brain activity in 1924. In the 1990s, the advent of digitalized EEG revolutionized inpatient continuous EEG (cEEG) monitoring. The scale and scope of cEEG has expanded exponentially over the last several decades². The expansion of cEEG is driven in part by the recognition that non-convulsive electrographic seizures are relatively common in hospitalized patients³.

In critically ill patients, seizure incidence has been reported to be between 3.3 – 34%⁴ with more recent large studies suggesting an incidence closer to 12.0%⁵. Most of these seizures (75%) are nonconvulsive^3,6,7. It requires a high index of suspicion from the clinician to warrant EEG for days. Alternatively, using prophylactic anti-seizure medications (ASMs) without EEG tends to fall short in preventing and stopping seizures in those patients at risk and can cause unnecessary side-effects and even poor outcomes in those at low risk of seizure^8–10

American Clinical Neurophysiology Society (ACNS) Standardized Critical Care EEG terminology- 2021¹¹ introduced standardized terminology to define electrographic seizures, Ictal-interictal continuum (IIC) and terms such as brief potentially ictal rhythmic discharges (BIRDs). These terms and seizure risk stratification tools such as 2HELPS2B¹² offer more clarity in identifying potentially ictal patterns and those that may warrant longer continuous EEG (cEEG) monitoring and aggressive treatment. The presence of electrographic seizures and associated patterns has been tied to local metabolic failure^7,13,14and poor clinical outcomes^15,16.

A personalized approach to treatment of seizures and IIC patterns is essential and cEEG, clinical exam, along with ancillary testing with neuroimaging, multi-modal monitoring or serum biomarkers maybe necessary, but more study is required. We still do not have strong evidence guiding the selection and aggressiveness of treatment for a given EEG pattern.

Abbreviations	New Terms	Definitions – ACNS Standardized Critical Care EEG Terminology 2021
Esz	Electrographic seizures	a. Epileptiform discharges averaging >2.5 Hz for ≥10 seconds (>25 discharges in 10 seconds), OR b. Any pattern with definite evolution as defined above and lasting 10 seconds
ESE	Electrographic Status Epilepticus	an ESz for ≥ 10 continuous minutes or for a total duration of ≥20% of any 60-minute period of recording
ECSz	Electroclinical Seizures	any EEG pattern with either: a. Definite clinical correlate time-locked to the pattern (of any duration), OR b. EEG AND clinical improvement with a parenteral (typically IV) antiseizure medication
ECSE	Electroclinical Status Epilepticus	any seizure for ≥ 10 continuous minutes or for a total duration of ≥ 20% of any 60-minute period of recording. An ongoing seizure with bilateral tonic-clonic(BTC) motor activity only needs to be present for ≥ 5 continuous minutes to qualify as ECSE. This is also referred to as “convulsive SE,” a subset of “SE with prominent motor activity.”
BIRDs	Brief Potentially Ictal Rhythmic Discharges	focal (including Lateralized(L), bilateral independent(BI), Unilateral Independent(UI) orMultifocal(Mf)) or generalized rhythmic activity >4 Hz (at least six waves at a regular rate) lasting ≥ 0.5 to <10 seconds, not consistent with a known normal pattern or benign variant, not part of burst-suppression or burst-attenuation, without definite clinical correlate, and that has at least one of the following features: a. Evolution (“evolving BIRDs,” a form of definite BIRDs) b.Similar morphology and location as interictal epileptiform discharges or seizures in the same patient (definite BIRDs) c. Sharply contoured but without (a) or (b) (possible BIRDs)
IIC	Ictal-Interictal Continuum	synonymous with “possible Esz” or “possible electrographic SE”, is a purely electrographic term and not a diagnosis and can have the following patterns: a. Any PD or SW pattern that averages >1.0 and ≤2.5 Hz over 10 seconds (>10 and ≤25 discharges in 10 seconds); or b. Any PD or SW pattern that averages ≥ 0.5 Hz and ≤1.0 Hz over 10 seconds (≥5 and ≤10 discharges in 10 seconds) and has a plus modifier or fluctuation; or c. Any lateralized RDA averaging >1 Hz for at least 10 seconds (at least 10 waves in 10 seconds) with a plus modifier or fluctuation. This includes any LRDA, BIRDA, UIRDA, andMfRDA, but not GRDA. AND d. Does not qualify as ESz or ESE

Patients at risk/seizure prognostication

Around 12 – 25% of critically ill patients placed on continuous EEG monitoring will have at least one electrographic seizure detected^3,14,17–26. The underlying conditions that can be associated with seizures are varied and include systemic medical illnesses such as sepsis, liver and kidney failure as well as primary neurological injuries like stroke, intra-cranial hemorrhage, brain tumor, or traumatic brain injury^{3,14,17,19,21,22,24,27–33}. Clinical risk factors that have been associated with an increased chance of seizure include a history of epilepsy, age < 18 years, coma, and suspected acute symptomatic seizure³⁴.

In comparison to clinical factors, EEG factors are more accurate predictors³⁵. For example, coma is associated with an increased seizure risk but once EEG features are accounted for the added predictive power of coma is lessened. Similarly, the type of brain injury can influence the risk of seizure³⁶, but this variance is less powerful when EEG factors are taken into account³⁵.

Relevant EEG findings for seizure prediction include sporadic epileptiform discharges (SEDs), lateralized periodic discharges (LPDs), generalized periodic discharges (GPDs), Bilateral Independent Discharges (BIPDs), Brief Potentially Ictal Rhythmic Discharges (BIRDS) and lateralized rhythmic delta activity (LRDAs).

LPDs are seen most commonly in acute structural brain injury^37–40 such as CNS infection, stroke, brain tumors and carry an increased risk of seizures^37–41. LPDs with higher frequency (> 2Hz) and/or with rhythmic or fast activity - so called “plus” features pose a higher risk^42,43 of seizures. Lateralized rhythmic delta activity (LRDA) is found in similar patient populations as LPDs and presents a similar risk for seizures^13,38. GPDs occur more frequently in toxic-metabolic encephalopathy and anoxic brain injury^44,45. And similar to LPD, higher frequency (>1.5 Hz)⁵ and associated plus features portend increased risk of seizures, but their presence in of themselves does not have the same risk as the lateralized findings (LPD, BIPDS, LRDA). Generalized rhythmic delta activity (GRDAs), even with higher frequency and plus modifiers, do not seem to increase risk for seizures⁵. BIRDs⁴⁶ are brief seizure like patterns lasting <10 seconds and highly associated with seizures and may be conceptualized as an incompletely formed seizure¹¹.

2HELPS2B^12,47 is a scoring system that combines several of these EEG markers and a single clinical factor (history of clinical seizure) into a risk stratification scheme to guide seizure monitoring and decision to use prophylactic anti-seizure medications¹⁰. The variables in 2HELPS2B are 1) frequency >2 Hz 2) sporadic Epileptiform discharges 3) LPD/BIPD/LRDA 4) Plus features 5) prior Seizure 6) BIRDs(2). BIRDs are assigned 2 points while the rest of the variables get 1 point each and the total score ranges from 0–7. A score of 0 confers a seizure risk < 5 % and a score of 7 a risk of > 95 % with the scores in the range predicting a risk in between. 2HELPS2B score has been validated in an independent cohort study¹² and its proposed clinical algorithm can be used to assess seizure risk based on 1 hour of EEG. For a score >=2, cEEG of 24 hours is recommended, for score = 1, 12 hours of cEEG and for score = 0, no cEEG is recommended¹². Further areas of study include developing updating real-time seizure risk assessment aided by automated or semi-automated EEG interpretation.

Ictal-Interictal Continuum and Electrographic Seizures

There is growing evidence indicating that electrographic seizures cause neuronal damage leading to secondary brain injury^48,49. Magnetic resonance imaging (MRI) of patients with status epilepticus have been found to have restricted diffusion on diffusion weighted images (DWI) in the hippocampi, insula and cortex, indicating neuronal injury^50,51 Neuron specific enolase (NSE), which is marker of neuronal injury is elevated in patients with status epilepticus, indicating secondary brain injury^52,53. In the landmark study by Vespa et al., patients with traumatic brain injury (TBI) who had electrographic seizures were found to have ipsilateral hippocampal atrophy six months post discharge¹⁴. In another single center retrospective study, patients with super refractory status epilepticus (SRSE) were found to have global cerebral atrophy which was related to the duration of SRSE⁵⁴. The data is conflicting on whether the presence of seizures leads to poor patient outcomes^27,29,55,56. However, these studies did not account for seizure burden and there seems to be a dose dependent relationship between seizure burden and neurocognitive outcomes. In a cohort of patients with subarachnoid hemorrhage, every one hour of seizure burden was associated with 10 times higher odds of disability and death and worse cognitive outcome at 6 months as measured by T-MOCA¹⁸. Similarly in pediatric population, seizure burden of > 20% per hour (12 mins per hour), was found to be associated with worse functional outcomes at discharge¹⁶. These findings are reflected in the newly proposed criteria for electrographic status epilepticus (ESE) which is defined as total seizure burden >20% per hour¹¹, indicating a threshold beyond which recurrent seizures have been shown to have detrimental outcomes.

With the recent recognition of various rhythmic and periodic patterns on cEEG, the term ictal-interictal continuum (IIC) has been devised to remove the binary connotation of any given pattern to be either ictal or inter-ictal⁴⁴. Rather, these patterns are now thought to occur on a spectrum. Although the definition and spectrum of IIC continues to evolve over time, a consensus definition was recently proposed by ACNS in the 2021 nomenclature¹¹ to facilitate the further understanding of these patterns. Whether these patterns cause neuronal injury similar to electrographic seizures continues to be an area of active investigation. Ancillary imaging modalities such as fluorodeoxyglucose positron emission tomography (FDG-PET), single photon emission computed tomography (SPECT), CT perfusion, and MR perfusion, are being used to assess the extent of neuronal loss caused by IIC patterns⁵⁷. In a single center study, patients with lateralized periodic discharges were found to have concurrent hypermetabolism on using FDG-PET scan, that resolved following treatment¹³[figure 1]. Moreover, these changes of hypermetabolism directly correlated with the frequency of the LPDs, indicative of dose dependent relationship⁵⁸. Similarly, perfusion studies utilizing MRI and CT studies have demonstrated increased cerebral perfusion with periodic discharges similar to seizures, denoting their ictal nature^59–61.

Figure 1:

Metabolic correlates of EEG patterns

Advanced multi-modality invasive monitoring techniques have been instrumental in understanding the physiological underpinning of these patterns. Intracranial depth electrode monitoring has revealed that scalp IIC patterns often correlate with intracranial electrographic seizures⁷. In patients with traumatic brain injury (TBI), periodic patterns > 2 Hz produced metabolic crisis as increased lactate/pyruvate ratio (LPR) and decreased glucose levels in cerebral microdialysate akin to electrographic seizures⁴⁹. Similarly, in subarachnoid hemorrhage patients with multi-modality monitoring, higher frequency PDs (>2 Hz) were temporally related to reduced partial pressure of brain tissue oxygen (PbtO₂). There was a compensatory increase in cerebral blood flow for PDs up to 2 Hz however CBF did not increase for PD> 2 Hz, demonstrating an inadequately compensated state¹⁵. These findings substantiate the evidence that higher frequency PDs cause metabolic disturbances and neuronal injury similar to electrographic seizures.

Multiple prior studies have demonstrated that IIC patterns are associated with worse functional outcomes at discharge^45,62. In patients with subarachnoid hemorrhage (SAH) and acute ischemic stroke, burden of IIC patterns was found to be associated with worse functional outcomes despite adjusting for disease severity and other clinical variables^63,64. Recent work employing automated machine learning system to quantify the epileptiform activity (EA) burden demonstrated that peak EA burden in the 12-hour window within the first 72 hours was associated with worse functional outcome at discharge⁶⁵. Additionally, increasing peak EA burden from 0 to 100% increases the probability of poor neurologic outcome by 35%⁶⁵.

Ancillary Testing:

Decision making surrounding the management of IIC and electrographic seizures can be complex.

Ancillary testing can help guide treatment decisions by potentially revealing evidence of neuronal injury in a patient with abnormal EEG findings. Serum NSE is the best described serum test in this regard; NSE levels will rise in any process causing neuronal injury, thus, absolute levels (>20) and trends can be used to detect ongoing neuronal injury associated with an IIC pattern⁵². However, other pathological processes of acute brain injury can affect NSE levels, independent of epileptiform activity, limiting its use in most acute neurological disease state. NSE is also present in erythrocytes, platelets, and neuroendocrine tissues, further reducing its specificity⁵³.

Neuroimaging with MRI, FDG-PET or SPECT is perhaps the most used ancillary test in the clinical setting. Diffusion weighted imaging (DWI) changes are deemed to represent increased metabolic demand and theoretically, neuronal swelling⁵⁷. Accordingly, the presence or absence of DWI changes during IIC patterns or frequent repetitive seizures can guide decision to treat, although data on this practice is limited. MRI perfusion provides information about regional blood flow changes, which if correlates with epileptiform area on EEG, can serve as a metabolic signature of an abnormal EEG pattern. Another functional neuroimaging modality, FDG-PET measures cerebral metabolism by evaluating glucose uptake and can be particularly useful to define the nature of periodic discharges by providing good temporal and spatial resolution⁶⁶. SPECT is conceptually similar to FDG-PET but provides lower spatial resolution⁶⁷.

Invasive multimodal monitoring consists of various intracranial parenchymal monitors that can provide real-time information on secondary brain injury, detecting detrimental physiologic changes by tracking brain oxygen delivery-consumption, cerebral blood flow and metabolism. Cerebral microdialysis measures extracellular glucose, lactate, pyruvate, and glutamate⁶⁸. Lactate-to-pyruvate ratios (LPR) are useful parameters to trend and values greater than 40 indicate metabolic distress and enhanced anaerobic metabolism⁶⁹. Other measurements like brain tissue oxygen tension (PbtO2) and jugular venous oximetry (SjvO2) provide information on cerebral utilization of oxygen⁶⁹. Data obtained from these monitors are combined and interpreted along with other measurements to untangle the relationship between the abnormal EEG pattern and its metabolic and clinical impact.

It is important to keep in mind that ancillary tests provide only a metabolic and physiologic footprint, from which neuronal injury is inferred. Further, suspicion for neuronal injury related to EEG abnormality should be tempered with additional reflections as to whether the epileptiform activity is the underlying cause (as opposed to a sequalae) or the sole driving force of inferred neuronal injury.

Clinical cases

Case 1.

A 66 -year -old man presented to a local emergency room with acute onset of aphasia, and difficulty typing that started 1 hour prior, then had a witnessed generalized tonic clonic seizure. Head CT was reported to be normal. Given initial suspicion for stroke at the small community hospital, he received tPA, lorazepam and levetiracetam and then transferred to the tertiary center where he presented with persistent confusion. CT angiogram showed an enhancing lesion in the left inferior parietal lobule. cEEG initially showed diffuse slowing, but within 24 hours he developed left sided seizures progressing to focal status epilepticus (video 1). Lacosamide was added to levetiracetam, and patient was taken to the OR for resection of the tumor 3 days after presentation resulting in resolution of the status epilepticus. On post-operative day (POD) #2 he developed repetitive electroclinical seizures with intermittent eye deviation with persistent confusion, but was conversant, able to participate in care, and eating on his own. Phenytoin was added but he developed transaminitis. Clobazam was added subsequently and then Zonisamide finally with seizure cessation noted on POD # 5. MRI showed cytotoxic edema along resection bed. Pathology was consistent with a grade IV glioblastoma. He continued to improve in the hospital and was discharged on POD #7. At follow up, 2 months after discharge he had remained seizure free, was doing well clinically, even tolerating his 4 anti-seizure medications with a wean to fewer medications being planned.

Learning points:

• Surgery may be part of the treatment plan in focal status epilepticus related to a lesion. Sometimes, seizures may recur/persist in the immediate post-operative period.

• Not all refractory status epilepticus require escalation to anesthetic drips.

• Selecting medications based on interactions, potential side effects and varied mechanisms of actions can be beneficial.

Case 2

A 30-year-old man with history of intravenous (IV) drug use, endocarditis presented after witnessed cardiac arrest and having received CPR for 15 minutes. Post resuscitation altered mental status and clinical myoclonus was noted. EEG showed a burst suppression pattern initially. Burst were associated with clinical myoclonus [video 2], consistent with myoclonic status epilepticus. After some initial response to medications, he settled into a pattern of generalized periodic polyspike discharges ranging from 1–4 Hz on the ictal-interictal continuum [EEG figure 2a&b]. He was treated with multiple anti-seizure medications and anesthetics including with ketamine, midazolam, propofol and pentobarbital for a prolong period resulting in complications such as ileus. Eventually decision was made to terminally extubate, wean anti-seizure medications with using medications such as amantadine/methylphenidate to promote wakefulness. Post extubation, he was noted to follow one-step commands and interact some. A witnessed seizure on 7/19 and some attempted verbalization resulted in aggressive treatment being re-initiated. He was placed back on multiple medications, and re-intubated. Continuous EEG again showed subclinical status epilepticus and stimulation induced myoclonic seizures that subsided with increasing anesthetics. MRIs performed on Day 3 and on Day 8 and head CT on Day 23 were all normal. With continued treatment, EEG background steadily improved to mild slowing with continued stimulation induced myoclonus. Patient was eventually discharged to a long-term care facility. At discharge he was awake and alert, verbalizing answers with intermittent myoclonic jerking still noted.

Figure 2a:

EEG Background evolution on day 1 with more continuous activity noted which then progressed to a pattern of generalized periodic polyspike discharges ranging from 1–4 Hz on the ictal-interictal continuum

Figure 2b:

EEG then settled into pattern of generalized periodic polyspike discharges ranging from 1–4 Hz on the ictal-interictal continuum

Learning points:

• Not all post anoxic myoclonus with burst suppression results in a bad outcome, results can be heterogeneous⁷⁰

• While burst suppression with identical bursts are associated with poor outcomes, background improvement to continuous EEG in the first 12 hours is strongly associated with favorable neurological outcome^70,71.

Case 3.

A 67-year-old man with multiple cardiovascular risk factors was admitted for aorto-bifemoral bypass with superior mesenteric artery bypass. He had intraoperative bowel resection for ischemic bowel complicated by abdominal compartment syndrome, periapical abscesses of multiple teeth s/p dental extractions, severe AKI (on CVVT) and then cardiac arrest leading to intubation. A stroke code was called during admission for diffuse weakness and change in ability to follow commands with sedation wean. Patient had also been on multiple antibiotics including cefepime. Neurologic exam showed a comatose patient with eyes open, impaired leftward gaze, & no movement to pain or command. CT/CTA results were reassuring against a large infarct. CEEG was started to evaluate for possible seizures and showed multifocal sharps, generalized periodic discharges, reaching 2 Hz [Figure 3]. Cefepime was stopped and over the next 3 days, improvement was noted on EEG with decreasing frequency and improving morphology of periodic discharges that then resolved [figure 4]. Clinically also patient had improved mental status without the use of anti-seizure medications. Patient was eventually discharged to a rehab center.

Figure 3:

EEG shows an IIC pattern with generalized periodic discharges reaching up to 2 Hz

Figure 4:

Three days off Cefepime, patient’s EEG background improved with IIC pattern no longer seen

Learning points:

• The risk associated with IIC patterns must be interpreted within the clinical context.

• In this case, patient improved without the use of any anti-seizure medications, by just withdrawing the offending medication (cefepime).

Case 4

A 67-year-old woman with stage IV Melanoma and recent surgical resection of left parietal lobe metastasis presented with several days of right sided arm and leg twitching, which progressed to a generalized tonic-clonic seizure. Two weeks prior to her presentation, patient had received one dose of an immune checkpoint inhibitor therapy with subsequent development of immune mediated colitis requiring hospitalization for several days.

Upon presentation, patient was unresponsive with persistent right sided arm and leg twitching. She was treated with IV benzodiazepine and was emergently intubated for airway protection. Continuous EEG monitoring was commenced shortly after and showed continuous bilateral asymmetric LPDs lying on the IIC [figure 6]. Patient was treated with sequential addition of three ASMs (lacosamide, phenytoin, levetiracetam) and up titration of midazolam and propofol infusions till suppression of all epileptiform abnormality. Initial work up was notable for mildly elevated protein but otherwise unremarkable CSF. MRI Brain showed restricted diffusion and FLAIR hyperintensity in the left occipital and parietal lobes [figure 5a–c]. Empiric IV steroid therapy was initiated in conjunction with ASMs given suspicion for immune checkpoint mediated encephalitis.

Figure 6:

EEG showing continuous bilateral asymmetric LPDs+F, left posterior quadrant maximal.

Figure 5:

Top row: MRI showing DWI changes (a,b) in the left occipital and temporoparietal regions as well as related T2 flair hyperintensities(c). Bottom row- (d-f)asymmetrically increased FDG uptake in the left occipital region, and increased uptake in the bilateral thalami and mesial temporal lobes.

IV midazolam and propofol infusions were weaned and discontinued towards the end of the first week of admission. Upon completion of anesthetic wean, patient remained stuporous and sporadic epileptiform discharges began to reappear in left posterior quadrant, which over the course of several more days evolved into IIC pattern (LPDs with superimposed fast activity). Substitutions were made to ASMs but patient’s EEG continued to worsen with emergence of BIRDs and seizures [figure 7]. Ketamine infusion was commenced with quick resolution of BIRDs/seizures but persistence of IIC pattern. FDG-PET imaging along with repeat MRI Brain was obtained which demonstrated hypermetabolic foci in the left occipital lobe, bilateral thalami and mesial temporal lobe [figure 5d–f] and persistent restricted diffusion with increased FLAIR/T2 signal in the left occipital-parietal lobe respectively. Given imaging findings, decision was made to up titrate ketamine to achieve burst suppression.

Figure 7:

continuous periodic left posterior quadrant (P3/01) LPD+F at 1.5–2Hz with occasional focal seizures and BIRDs

After 24 hours of burst suppression, ketamine was discontinued, and patient was maintained on 3 ASM. Patient’s mental status and EEG background began to slowly improve. However, LPDs began to reappear several days later, occasionally meeting criteria for IIC [figure 8]. Given stable neurological exam, decision was made not to escalate treatment and continue to observe patient.

Figure 8:

EEG showing nearly continuous left posterior quadrant LPDs at 1–1.5 Hz s

Patient eventually underwent tracheostomy and was discharged to acute rehab 8 weeks following her presentation. At time of discharge, patient displayed some evidence of cognitive recovery, she was able to follow commands and was able to mouth words. EEG performed at this time showed mild to moderate slowing with sporadic sharp waves posteriorly.

Learning points:

• Ancillary tests such as MRI and PET CT can reveal evidence of brain injury or metabolic impact associated with an EEG pattern and provide guidance for clinical decision making.

• The duration of anesthetic use should be tailored to patient needs and early weaning be tried when appropriate.

Case 5

A 23-year-old, otherwise healthy woman presented with new onset bilateral tonic-clonic convulsions. A week prior, she was noted to have worsening headache, malaise, dizziness, and flulike symptoms. Around the same time, she received multiple vaccines including for COVID. Seizure semiology was described as forced head and eye deviation to the right followed by whole body convulsions. On EEG, multiple electro-clinical seizures, arising independently from right and left temporal regions were recorded [Videos 3 & 4]. Extensive evaluation including magnetic resonance imaging (MRI) of brain, computed tomography angiography (CTA) of head and neck, cerebrospinal fluid analysis including autoimmune and paraneoplastic panels were unrevealing. With a working diagnosis of new onset refractory status epilepticus (NORSE) and possible seronegative autoimmune encephalitis, her treatment included empiric steroids, IVIg, plasma exchange, multiple anti-seizure medications, anesthetics and ketogenic diet. She was eventually successfully weaned off anesthetics and status epilepticus resolved following a combination of above mention treatments, but sporadic seizures continued in the setting of illnesses, fevers in long term care/rehab on 6 different anti-seizure medications at therapeutic doses (LEV, LCM, TOP, PHB, PMP, PHT). Post discharge, she continues to have significant cognitive and motor deficits but continues to make progress in rehabilitation.

Learning point:

• Aggressive management including an extensive work up, anti-seizure medications, dietary therapy, immunotherapy and a prolonged hospital course with supportive measures are often part of the management in these difficult cases of NORSE and FIRES (febrile infection-related epilepsy syndrome).

Approach to Treatment:

Approach to management of the critically ill patient with epileptiform abnormality is twofold. First, a decision must be made as to whether treatment of the abnormal EEG pattern is warranted. Second, a treatment strategy is necessary, taking into consideration various medication options and assessment of treatment response. The clinician must also decide on treatment intensity and duration, balancing potential side effects from treatment against the risk of cumulative neuronal damage, understanding that both can lead to worsened patient outcomes.

In general, clinical seizures and EEG patterns meeting criteria for electrographic seizures or NCSE warrant treatment. Patterns on the IIC require special consideration; the continuum encompass a wide range of patterns with some more on the “ictal” end than others. While it is reasonable to approach treatment of patterns on the “ictal” end identical to seizures, data from existing literature is divided on whether IIC patterns carry the same significance and impact as seizures and status epilepticus. Additionally, IIC patterns lying in the intermediate zone can also have significant clinical impact, depending on the clinical scenario and underlying disease state. Ancillary testing (described in the previous section) can provide some guidance when faced with IIC pattern of unclear significance, but these tests are not always feasible to perform and can yield indeterminate or confounding results. In the absence of reliable ancillary testing, the treating clinician can opt to initiate empiric treatment trials or continue observation of patient with cEEG monitoring, only starting treatment upon evidence of clear seizures or clinical worsening.

Empiric treatment trial with low-dose benzodiazepines have historically been considered positive when both an electrographic and clinical improvement was observed⁴⁴. However, benzodiazepine trials are frequently equivocal, with apparent electrographic improvement and no corresponding clinical improvement. This phenomenon is largely due to the sedative effect of benzodiazepines worsening the typically encephalopathic critically ill patient. An alternate approach is to start a treatment trial with a non-sedating ASM that is rapidly titratable and has few drug interactions⁷². ASMs can then be added or substituted sequentially depending on clinical or electrographic response. Some experts advocate for starting with this method over a benzodiazepine trial to have better clarity on clinical improvement^73,74. In some cases where there is high suspicion for a malignant IIC pattern, it may be reasonable to choose a more aggressive method, with early initiation of intravenous anesthetics in addition to ASMs. This approach carries more risk but can provide a greater chance of rapid and effective treatment to prevent secondary neuronal injury.

Upon initiation of treatment, empiric or otherwise, it is essential to establish treatment endpoints. The highly desired outcome is electrographic resolution of abnormal patterns with corresponding clinical improvement, achieved with the least amount of medications. However, in some critically ill patients, it can be challenging to achieve this ideal scenario as attempts to completely suppress all abnormal EEG patterns can lead to oversedation or induced coma and paradoxically impair patient recovery⁷⁵. Furthermore, some EEG patterns in the critically ill patient may resolve spontaneously after addressing the underlying medical disorder or exacerbating agents (e.g medications, fever, uremia).

In a similar vein, it is unclear if patients with frequent electrographic seizures should be treated as aggressively as NCSE. Although some clinical studies have shown that the total amount of time spent in ictal activity during an ICU course can amount to a “seizure burden” associated with worse functional outcomes^16,76, a recent study among acutely encephalopathic critically ill children found that electrographic status epilepticus but not electrographic seizures was associated with unfavorable neurobehavioral outcomes⁷⁷. To date, there are no evidence-based studies on treatment of IIC patterns and treatment-related outcomes. Ultimately, treatment decisions should be individualized, incorporating all factors discussed in this section [figure 9].

Figure 9:

Suggested management algorithm.

* ASM (anti-seizure medication) ± BZD - (benzodiazepine) or escalating doses/drugs and sedatives as appropriate.

**: High risk patterns such as BIRDS, high frequency (>2.5Hz) periodic discharges. See section on Ictal Interictal Continuum.

Summary:

A clinician’s goal in treating a critically ill patient is to orchestrate a meaningful and fast recovery with minimal long-term sequelae, thereby improving their quality of life at best or preventing deterioration at worst. The path to achieving this goal becomes increasingly complex and foggy when seizures and IIC patterns are present. The first step is identifying patients at risk for seizures and determining the timing and duration of EEG monitoring. 2HELPS2B or other risk stratification schemes may help, but the ultimate goal should be overcoming the logistical and technical barriers that prevents near uniform EEG monitoring during critical illness, much in the same way that EKG monitoring is currently used.

Status epilepticus can cause secondary neurological injury but the effects of discrete seizure burden and IIC patterns is still under investigation. The ambiguous nature of these EEG patterns necessitates ancillary testing where possible, such as NSE, neuroimaging with MRI, FDG-PET, or SPECT to better understand their malignant features. Treating status epilepticus, seizures, and IIC patterns is a delicate balance where the benefits and drawbacks of escalating IV anesthetic agents and ASM must be continually weighed. In addition to treating the EEG pattern the clinician should focus on reversible causes such as metabolic disarray, offending drugs, infection or inflammation and surgically remediable causes like sub-dural hemorrhage. The optimal order and combination of ASMs and IV anesthetics is still lacking given the complex dynamic interplay of the forces in neurocritical care, but one can advocate for rational polypharmacy –i.e. using anti-seizure medications with different mechanisms of action and compatible pharmacokinetics. Weaning of ASMs in the acute and recovery stage of critical illness is also in need of further clarification.

Answering these management questions will require an expanded use of multimodal ancillary testing, multi-center collaboration with data harmonization and innovate trial designs for a data-drive approach to neurocritical care EEG with the goal of arresting neurological deterioration and improving neurological recovery.

Key Points.

• There is growing evidence indicating that electrographic seizures cause neuronal damage leading to secondary brain injury so early identification of patients at risk is necessary.

• Electrographic seizures and patterns on the ictal-interictal continuum are now better defined and their associated risks, supported by ancillary tests are under investigation.

• Choice of medications and duration of treatment need to be tailored to the individual patient.

Synopsis.

Identifying and treating critically ill patients with seizures can be challenging. In this article we review the available data on patient populations at risk, seizure prognostication with tools such as 2HELPS2B, Electrographic seizures and the various IIC patterns with their latest definitions and associated risks, ancillary testing such as imaging studies, serum biomarkers and invasive multimodal monitoring. We also illustrate 5 different patient scenarios, their treatment and outcomes and propose recommendations for targeted treatment of electrographic seizures in critically ill patients. Armed with these tools and our clinical acumen, a personalized approach to the management of these patients is now within reach.

Clinical care points:

• The incidence of seizures in critically ill patients is approx. 12%, with a majority being non-convulsive seizures.

• Compared to clinical factors, EEG factors are more accurate predictors of patients at risk.

• Initial EEG along with seizure prognostication tools such as 2HELPS2B can guide duration of recording as well as treatment choices.