While most of this chapter focuses on seizure management, two important overall management components that must occur simultaneously are medical stabilization and identification of precipitating etiologies requiring specific therapy.

The NCS’s guideline provides a timed treatment pathway [15]. Steps in the initial 2 min include non-invasive airway protection and with head positioning, and vital sign assessment. Steps in the initial 5 min include neurologic examination and placement of peripheral intravenous access for administration of ‘emergent’ ASDs and fluid resuscitation. Steps in the initial 10 min include intubation if airway or gas exchange is compromised or intracranial pressure is elevated. Intubation may be necessary because of seizure associated hypoventilation, medication associated hypoventilation, inability to protect the airway, or other causes of oxygenation or ventilation failure. Steps in the initial 15 min include vasopressor support if needed [15].

Multiple studies have characterized the various etiologies for SE [7, 25–27]. Acute symptomatic conditions are identified in 15–20% of children with SE [7, 26, 28]. Rapidly reversible causes of seizures should be diagnosed and treated rapidly, e.g. with hypoglycemia, hyponatremia, hypomagnesemia, and hypocalcemia. The American Academy of Neurology’s (AAN) practice parameter on the Diagnostic Assessment of the Child with SE reported that abnormal results among children who underwent testing included low ASD levels (32%), neuroimaging abnormalities (8%), electrolyte abnormalities (6%), inborn errors of metabolism (4%), ingestion (4%), central nervous system infections (3%), and positive blood cultures (3%) [29]. The NCS’s guideline provides suggestions regarding etiologic testing including bedside finger stick blood glucose (0–2 min) and serum glucose, complete blood count, basic metabolic panel, calcium, magnesium, and ASD levels (5 min). In some patients, diagnostic testing may include neuroimaging or lumbar puncture (0–60 min), additional laboratory testing (including liver function tests, coagulation studies, arterial blood gas, toxicology screen, and screening for inborn errors of metabolism), and C-EEG monitoring if the patient is not returning to baseline after clinical seizures cease (15–60 min) [15]. These recommendations are similar to those of the AAN practice parameter [29].

Among children with SE, neuroimaging abnormalities have been reported in 30% of children and appear to alter acute management in 24% [26]. If no etiology is identified by computerized tomography (CT), magnetic resonance imaging (MRI) may still identify lesions. Among 44 children who underwent both CT and MRI, 14 had normal head CT but an abnormal MRI [26].

EEG monitoring may be indicated urgently if psychogenic SE is suspected [to avoid escalation of ASDs, with potential adverse effects] or if there is a concern that EEG-only seizures persist after termination of clinically evident seizures [30, 31]. A multi-center study of children who underwent C-EEG monitoring while in the pediatric ICU reported that 33% of 98 children who presented with CSE had subsequent electrographic seizures identified. Among those with seizures, electrographic SE occurred in 47% of subjects, and EEG-only seizures in 34% [32].

Central nervous system (CNS) infections are a common cause of acute symptomatic SE [26, 33, 34]. The clinical presentation of encephalitis and other CNS infections is highly variable depending on the pathogen involved and specific host factors. Fever and other clinical signs of infection may be absent in young children, individuals who are immunocompromised, or individuals who have received recent antibiotics. Lumbar puncture should be performed in all children with SE without a definite non-infectious cause. Guidelines from the Infectious Disease Society of America for adults recommend obtaining screening head imaging prior to lumbar puncture in patients who are immunocompromised, have a known space-occupying lesion or shunt, papilledema, or a focal neurologic deficit [35]. A lumbar puncture should also be obtained if an autoimmune etiology is suspected, as neuro-inflammatory processes will often yield cerebrospinal fluid pleocytosis, elevated cerebrospinal fluid protein, and intrathecal immunoglobulin synthesis (oligoclonal band profile, IgG index, and IgG synthesis rate). Rarer infectious, metabolic, autoimmune and paraneoplastic, and genetic etiologies may be considered in specific situations, or when other etiologies are not identified [20, 36].

Administration of appropriate ASDs should occur as the patient is medically stabilized and diagnostic studies are performed. Table 28.3 provides a summary of recommended medications and doses.

The NCS’s guideline notes that benzodiazepines are the “emergent” medications of choice, with lorazepam for intravenous administration, diazepam for rectal administration, and midazolam for intramuscular, buccal, or intranasal administration. Repeat dosing may be provided in 5–10 min if seizures persist [15]. With regard to the initial benzodiazepine medications, the AES’s guideline concluded that intravenous lorazepam and diazepam are efficacious at stopping seizures lasting at least 5 min (level A evidence) and that rectal diazepam, intramuscular midazolam, intranasal midazolam, and buccal midazolam are probably effective at stopping seizures lasting at least 5 min (level B evidence). It indicates that there are three equivalent first-line options including intravenous lorazepam (0.1 mg/kg/dose; repeat once if needed), intravenous diazepam (0.15–0.2 mg/kg/dose; repeat once if needed), and intramuscular midazolam (10 mg for >40 kg; 5 mg for 13–40 kg; single dose) (level A evidence) [16]. A double-blind randomized trial of 273 children with CSE in the emergency department compared intravenous lorazepam (0.1 mg/kg) and diazepam (0.2 mg/kg). A half-dose of either medication could be administered if seizures persisted after 5 min. The primary outcome, SE cessation by 10 min without recurrence in 30 min, was not significantly different in the two groups (72% with diazepam; 73% with lorazepam). Subjects receiving lorazepam were more likely to be sedated (67% with lorazepam, 50% with diazepam) but there was no difference in the requirement for assisted ventilation (18% with lorazepam, 16% with diazepam). The study concluded that the data did not support preferential use of lorazepam over diazepam [37]. If intravenous access cannot be obtained, rectal, intramuscular, buccal, or intraosseous benzodiazepines can be administered. For buccal or nasal dosing of midazolam, the intravenous formulation of the drug is generally used in the United States.

Administration of benzodiazepines may result in respiratory depression and hypotension, so continued medical monitoring and stabilization is important. The AES’s guideline noted that respiratory depression was the most common adverse event associated with ASD treatment (level A evidence) and that there was no difference in respiratory depression among midazolam, lorazepam, and diazepam by any administration route (level B evidence) [16]. In the clinical trial described above, assisted ventilation was required in 16% of the diazepam group and in 18% of the lorazepam group [37]. If the seizure does not terminate 5–10 min following initial benzodiazepine administration, a second benzodiazepine dose should be administered. Care should also be taken to assess whether pre-hospital administration of a benzodiazepine occurred, as excess benzodiazepine dosing increases the risk of respiratory suppression [4].

SE will persist in about one-third to one-half of children receiving benzodiazepines [11, 18, 37], but there are few comparative data evaluating the ASD options available for this stage [38]. Common options include phenytoin (or fosphenytoin), levetiracetam, phenobarbital,  and valproate. (see Table 28.1). The AES’s guideline concluded that there was insufficient evidence to evaluate phenytoin or levetiracetam as second-line therapy (level U evidence) but that IV valproic acid has similar efficacy but better tolerability than IV phenobarbital (level B evidence) [16]. Optimal decisions may depend on patient characteristics, seizure characteristics, and also practical institutional factors, such as which drugs are most rapidly available as some need to be ordered and dispensed from a pharmacy.

Phenytoin is reported as the second-line medication by most respondents in surveys of pediatric emergency medicine physicians [39] and neurologists [40]. Phenytoin has demonstrated efficacy in pediatric SE management [41, 42], but its historical role as the most commonly used second-line medication is based on few data, and there are no studies showing it to be more effective than other options such as levetiracetam, phenobarbital, or valproate. A recent meta-analysis of drugs administered for benzodiazepine-refractory CSE found that phenytoin had lower efficacy (50%) than levetiracetam (69%), phenobarbital (74%), and valproate (76%) [38]. More information may be available in the near future as the NIH funded Established SE Treatment Trial (ESETT) will compare phenytoin, valproate, and levetiracetam for CSE in children and adults [43].

Both phenytoin and fosphenytoin are considered ASDs effective in focal seizures, but they may be ineffective or worsen SE in the generalized epilepsies. Phenytoin is prepared with propylene glycol and alcohol at a pH of 12, which may lead to cardiac arrhythmias, hypotension, and severe tissue injury if extravasation occurs (the “purple glove syndrome”). Fosphenytoin is a pro-drug of phenytoin, and it is dosed in ‘phenytoin equivalents’ (PE). Cardiac arrhythmias and hypotension may be less common with fosphenytoin because it is not prepared with propylene glycol, but they may still occur. There are numerous drug interactions due to strong hepatic induction and high protein binding, so free phenytoin levels may need to be checked [44]. Phenytoin causes little respiratory depression, particularly when compared to some other ASDs. The NCS’s guideline classifies phenytoin and fosphenytoin as appropriate “emergent,” “urgent”, or “refractory” SE treatments, with an IV loading dose of 20 mg/kg (or for fosphenytoin, 20 “PE”/kg) [15]. The AES’s guideline notes that there are insufficient data to compare phenytoin and fosphenytoin for efficacy (level U evidence) but that fosphenytoin is better tolerated than phenytoin (level B evidence). It recommends fosphenytoin (20 PE/mg) as an appropriate second therapy phase medication [16].

Valproic acid is a broad spectrum ASD often used for SE and refractory SE management. It has multiple mechanisms of action, including some independent of GABA receptors, making it potentially useful when benzodiazepines have been ineffective. In several meta-analyses, valproic acid was found to have the highest relative efficacy among typical second-line ASDs [38, 45–47]. Valproate may be administered rapidly IV and is considered appropriate “emergent,” “urgent,” or “refractory” SE treatment by the NCS’s guideline, with a typical IV loading dose of 20–40 mg/kg [15]. The AES’s guideline recommends valproic acid dosing of 40 mg/kg [16].

Several studies have evaluated valproate as an urgent medication for SE. One study compared it to phenobarbital in children; 60 children with CSE or acute prolonged seizures were assigned randomly to receive valproate (20 mg/kg) or phenobarbital (20 mg/kg). There was a nonsignificant trend (p = 0.19) toward greater seizure termination with valproate (in 27 of 30; 90%) than with phenobarbital (23 of 30; 77%). Clinically significant adverse effects (mostly lethargy) occurred significantly less often with valproate (24%) than with phenobarbital (74%). Seizure recurrence within 24 h was also lower with valproate than phenobarbital (15% vs. 52%) [48].

Several studies that included children have compared valproate to phenytoin. An open-label study of patients with SE (mostly adults, but some children) randomized patients to receive either valproate (n = 35, 30 mg/kg) or phenytoin (n = 33, 18 mg/kg) as a first-line medication. Valproate was more effective than phenytoin (66% vs. 42%) as first-line therapy, and after crossover to second-line therapy (79% vs. 25%). Adverse effects were not different between the two groups [49]. Another study of patients with SE refractory to benzodiazepines (mostly adults, but some children) compared valproate (n = 50) and phenytoin (n = 50), and found no difference in efficacy (84% vs. 88%, respectively). Among patients under 18 years of age, seizures terminated in 91% with valproate and in 75% with phenytoin [50]. In another study, patients older than 15 years with RSE received either valproate (mean loading dose, 1000 mg) or phenytoin (mean loading dose 743 mg). The two groups were the same in terms of seizure control, time to seizure control, hospitalization duration, and mortality [51].

Several studies have reported that valproic acid was effective in terminating RSE in 78–100% of children, without adverse effects [52–56]. A prospective study of 41 adults and children with refractory CSE included 5 patients younger than 5 years old treated with valproate, 30 mg/kg IV load, followed by an infusion at 6 mg/kg/h. Seizures terminated in 88% within one hour, and no adverse effects were observed [54]. Another prospective open-label study of 40 children with RSE randomized subjects to receive IV valproate (30–40 mg/kg IV bolus) or a continuous diazepam infusion (10–80 mcg/kg/min). There was no difference in efficacy for seizure control (80% with valproate and 85% with diazepam infusion), but seizure control was more rapid with valproate (5 min with valproate, and 17 min with diazepam). No patient in the valproate group had adverse effects including need for ventilation or hypotension; 60% of those who received diazepam required ventilation, and 50% developed hypotension [55]. Several retrospective studies have also demonstrated that valproate often terminates RSE without major adverse effects [52, 53, 56].

Adverse events are infrequent with IV administration of valproic acid, but hypotension, thrombocytopenia, pancytopenia, platelet dysfunction, hypersensitivity reactions, pancreatitis, and hyperammonemia may occur. Valproic acid is a potent hepatic enzyme inhibitor and may increase levels of other medications. There is a Federal Drug Administration black box warning for hepatotoxicity, which is most common in children under 2 years of age receiving ASD polypharmacy and in those suspected of having mitochondrial or metabolic disorders. In the outpatient setting, valproic acid is estimated to cause hepatotoxicity in 1 in 500 children younger than 2 years of age and in children with metabolic disease; it must be used with caution in young children with SE of unclear etiology. A practice parameter on SE in children noted that data from 9 class III studies showed that an inborn error of metabolism was diagnosed in 4.2% of children with SE [29].

Levetiracetam is a broad spectrum ASD. Previously considered only for RSE, it has been used earlier in the course of SE due to its ease of dosing and lack of drug interactions. Levetiracetam has no hepatic metabolism, which may be beneficial in complex patients with liver dysfunction, or metabolic disorders. It has no drug interactions. In comparison to other ASDs available for IV administration, levetiracetam has a very low risk for sedation, cardio-respiratory depression, or coagulopathy. Since levetiracetam clearance is dependent on renal function, maintenance dosage reduction is required in patients with renal impairment. A growing number of observational [57] and retrospective case series and reports [58–67] indicate that levetiracetam may be safe and effective for treating both SE and acute repetitive seizures in children at IV loading doses of 20–60 mg/kg and without major adverse effects. The NCS’s guideline considers levetiracetam to be an appropriate “emergent,” “urgent,” or “refractory” SE treatment option at 20–60 mg/kg [15]. The AES’s guideline states that levetiracetam is an appropriate second therapy phase medication at 60 mg/kg [16]. A meta-analysis of drugs administered for benzodiazepine-refractory CSE found levetiracetam efficacious in 69% of subjects [38].

Phenobarbital is often considered a 3rd or 4th -line drug for pediatric SE. The NCS’s guideline considers phenobarbital appropriate for treatment of emergent, urgent, or refractory phases of SE [15]. The typical IV loading dose is 20 mg/kg, with an additional 5–10 mg/kg if needed. A recent meta-analysis of drugs administered for benzodiazepine-refractory CSE found that phenobarbital was efficacious in 74% [38]. One study of 36 children with SE found that phenobarbital stopped seizures more quickly than a combination of diazepam and phenytoin; safety was similar with both [68]. Several reports have described the use of high-dose phenobarbital to control RSE and allow withdrawal of pharmacologic coma [69–71]. Phenobarbital may cause sedation, respiratory depression, and hypotension, so cardiovascular and respiratory monitoring is generally required. It is a hepatic enzyme inducer leading to drug interactions.

RSE is characterized by seizures that persist despite treatment with adequate doses of initial ASDs. Definitions vary in seizure duration (with no time criteria, 30 min, 1 h, or 2 h), and by response or lack of response to different numbers (2 or 3) of ASDs. The NCS’s guideline states that patients who continue to have either clinical or electrographic seizures after receiving adequate doses of an initial benzodiazepine followed by a second acceptable ASD have RSE [15]. Depending on the definitions and cohorts described, RSE occurs in about 10–40% of children with SE [12, 13, 41, 72].

In a subgroup of patients, RSE may last for weeks to months, despite treatment with multiple medications— which has been referred to as “malignant” RSE [73] or super-RSE [74, 75]. They are associated with infectious or inflammatory etiologies, younger age, previous good health, and high morbidity and mortality [73, 76, 77]. They have been referred to as de novo cryptogenic refractory multi-focal SE [77], new-onset RSE (NORSE) [76, 78, 79], and febrile infection-related epilepsy syndrome (FIRES) [80–82]; these terms may overlap in describing similar or identical entities [83].

The NCS’s guideline indicates that when a benzodiazepine and “urgent” control medications are ineffective, clinicians may give another “urgent” control medication or proceed to initiate pharmacologic coma medications [15]. Additional urgent control ASDs may be reasonable if they have not been tried yet or if the patient needs to be transferred or stabilized prior to administration of continuous infusions. If an initial urgent control medication fails, definitive seizure control should be initiated with continuous infusions. There should not be many trials of ‘urgent control’ medications before advancing to pharmacologic coma induction.

The management of pediatric RSE has been reviewed [20, 22, 84–87]. There is some variability among suggested approaches, as there are few data to guide evidence-based management [88], but all pathways administer either additional ASDs such as phenytoin, fosphenytoin, phenobarbital, valproate sodium, or levetiracetam, or they proceed to pharmacologic coma induction. The NCS’s guideline recommends rapid advancement to pharmacologic coma induction rather than sequential trials of many “urgent control” ASDs [15]. Substantial delays have been described before administration of pharmacologic coma induction, indicating that attention to timing is important [3].

The medications used most often for induction of coma are midazolam, pentobarbital, and propofol. Midazolam use usually involves an initial loading dose of 0.1–0.2 mg/kg followed by an infusion at 0.05–2 mg/kg/h, titrated as needed to achieve clinical or electrographic seizure suppression or EEG burst–suppression. Pentobarbital dosing usually involves an initial loading dose of 5–10 mg/kg followed by an infusion at 0.5–5 mg/kg/h titrated similarly. If seizures persist with midazolam or pentobarbital, an escalating dose with additional boluses and increase in the infusion rate are needed to increase levels and terminate seizures rapidly. Increasing the infusion rate without additional bolus dosing will lead to a very slow increase in serum levels—inconsistent with the goal of rapid seizure termination. Anesthetics such as isoflurane are also effective in inducing a burst–suppression pattern and in terminating seizures but often cause vasopressor-requiring hypotension; seizures often recur upon weaning, and few data are available to guide use [89, 90]. Propofol may terminate seizures and induce burst–suppression rapidly [91], but it is rarely used in children due to its Federal Drug Administration black box warning of the propofol infusion syndrome [92].

Patients treated with continuous infusions or inhaled anesthetics require intensive monitoring due to concerns about numerous potential complications. First, continuous mechanical ventilation is usually needed for airway protection and to maintain appropriate oxygenation and ventilation. Second, central venous access and arterial access is important due to frequent laboratory sampling and a high likelihood of developing hypotension requiring vasopressor or inotropic support. Third, temperature regulation must be managed because high-dose sedatives and anesthetics can blunt the shivering response and endogenous thermoregulation. Fourth, assessment is important for development of lactic acidosis, anemia, thrombocytopenia, and end organ dysfunction such as acute liver or renal injury. Fifth, patients must be monitored for infections, with multiple risks for secondary infections due to indwelling catheters (central catheters, endotracheal tubes, Foley catheters) and immune suppressing medications (e.g., pentobarbital).

It remains unclear whether the EEG treatment goal of pharmacologic coma induction should be termination of seizures or a burst–suppression pattern on the EEG; either goal is considered appropriate by the NCS guideline [15]. It is also unclear how long the patient should be maintained in pharmacologic coma. The NCS guideline recommends 24–48 h of electrographic seizure control prior to a slow withdrawal of continuous infusion ASDs [15], consistent with a survey of experts in SE management [93]. A meta-analysis of midazolam for RSE found that much higher doses of midazolam were needed, with longer times to seizure control, in studies using EEG monitoring, compared to studies in which care was guided by clinical seizure identification, indicating that ongoing seizures may not be identified and targeted for treatment without EEG monitoring [88].

Electrographic or electro-clinical seizures frequently recur during weaning of pharmacologic coma medications [94–97], indicating that pharmacologic coma should be considered to provide a window to evaluate further for precipitant etiologies, initiate specific management when possible, and optimize the ASD regimen to provide seizure control as coma-inducing medications are weaned. Case reports and series have described several add-on medications, and other techniques have been reported useful in reducing seizure recurrence as pharmacologic coma is weaned, but there are no large studies to guide management. As summarized in several review papers, [20, 84, 87, 98] case series and case reports have described benefit with topiramate [71, 99–104], lacosamide [105–107], phenobarbital [68, 69, 108, 109], ketamine [110–114], pyridoxine [115–120], neurosteroids [121], lidocaine [122–124], the ketogenic diet [71, 83, 125–131], therapeutic hypothermia [132–136], immunomodulation [137, 138], epilepsy surgery [111, 139–149], vagus nerve stimulation [150], and electroconvulsive therapy [151–153].

Seizures and SE are very common in critically ill children. Observational studies of interdisciplinary neurologic critical care teams at large pediatric institutions describe seizures and SE as the most commonly managed conditions, with EEG and EEG monitoring often performed [1, 2]. Studies of critically ill children undergoing clinically indicated C-EEG monitoring report that electrographic seizures occur in 10–50% of patients (Fig. 28.3); about one-third of critically ill children with electrographic seizures may be categorized as in electrographic SE (Fig. 28.4) [30, 31, 154–172]. The indications for C-EEG monitoring varied across these studies. Some included only patients with known acute structural neurologic disorders (e.g., hypoxic-ischemic brain injury, encephalitis, or traumatic brain injury) while others included patients with encephalopathy due to broader and more heterogeneous diagnoses (e.g., both primary neurologic and primary medical conditions). Inclusion criteria variability may explain the broad range of reported electrographic seizure incidence—lower in studies with broader inclusion criteria. Additionally, many studies were small, as reflected in the wide 95% confidence intervals in Fig. 28.3. When individual subjects from these studies are analyzed together, the overall electrographic seizure incidence is 34%.

The largest epidemiologic study of C-EEG monitoring in pediatric ICUs was a retrospective study in which 11 tertiary care pediatric institutions each enrolled 50 consecutive subjects, thereby yielding 550 subjects. Electrographic seizures occurred in 30% of subjects. Among those with electrographic seizures, electrographic SE occurred in 33%, and EEG-only seizures in 35% [31]. These data are consistent with other single-center studies [30, 156, 159, 161–163, 165–168, 171]. Additionally, EEG-only seizures occurred in children who had not received paralyzing medications, recently or ever, during their ICU stay [30, 167], indicating the occurrence of an electromechanical uncoupling or dissociation, and not simply the masking of clinically evident seizures by paralytic medications.

C-EEG monitoring is resource-intense. Seemingly small utilization and workflow changes have substantial impacts on equipment and personnel needs [173, 174]. Identifying children at higher risk for having electrographic seizures may be beneficial in optimally directing limited C-EEG monitoring resources. There are several risk factors for electrographic seizures: 1. Younger age (infants as compared to older children) [30, 31, 161, 164, 168]; 2. Convulsive seizures [31, 162, 166] or convulsive SE [161] prior to the initiation of EEG monitoring; 3. Acute structural brain injury [160–162, 164–166, 168, 171]; and 4. Interictal epileptiform discharges [31, 161, 165, 166] or periodic epileptiform discharges [156]. These risk factors may have limited clinical utility in selecting patients to undergo C-EEG monitoring because the absolute difference in the proportion of children with and without electrographic seizures, depending on the risk factor, is often only 10–20%. Seizure prediction models combining multiple risk factors might allow better targeting of EEG monitoring in a given center [175].

Observational studies of critically ill children undergoing C-EEG note that about 50% of patients with electrographic seizures are identified within 1 h of C-EEG monitoring initiation, and 90% within 48 h of C-EEG monitoring initiation (Fig. 28.5) [30, 156, 157, 159, 161, 162, 166, 167]. Most studies calculated C-EEG duration from the onset of C-EEG rather than from the time of the acute brain insult. Patients generally underwent 1–3 days of C-EEG, and seizures may have occurred later, after C-EEG was discontinued. The NCS’s Guideline for the Evaluation and Management of SE strongly recommends performing C-EEG monitoring for 48 h to identify electrographic SE in comatose children following an acute brain insult [15]. Similarly, the American Clinical Neurophysiology’s (ACNS) Consensus Statement on Continuous EEG Monitoring in Critically Ill Children and Adults recommends performing C-EEG monitoring for at least 24 h in children at risk for electrographic seizures [176]. A survey of neurologists about EEG monitoring utilization noted that most perform 24–48 h of C-EEG monitoring when screening for electrographic seizures [177].