24.1 Status Epilepticus and Seizures (03.N)

Why the clock starts at five minutes

Status epilepticus (SE) is one of the few neurologic crises in which minutes of under-treatment convert a reversible electrographic storm into hypotension, aspiration, neuronal injury, and death. The neurocritical care examination tests whether you apply an operational definition, move through time-based stages without waiting for imaging, give first- and second-line drugs at trial doses, and know when continuous electroencephalography (cEEG) and anesthetic coma are required.

Operational definition

The International League Against Epilepsy (ILAE) 2015 framework treats SE as failure of the mechanisms that should stop a seizure, or as mechanisms that abnormally prolong one. Two times matter. t1 is when a seizure is already abnormally prolonged and treatment should start. t2 is when long-term consequences—neuronal injury and network reorganization—become likely.

For convulsive (tonic–clonic) SE, t1 is 5 minutes and t2 is 30 minutes. The bedside rule is therefore: 5 or more minutes of continuous convulsive activity, or 2 or more convulsive seizures without recovery of consciousness between them. Do not wait 30 minutes to treat. Thirty minutes is the injury horizon, not the treatment start time.

Focal SE with impaired awareness has a longer t1 (about 10 minutes) and t2 (about 60 minutes). Absence SE is longer still. Those distinctions matter for nonconvulsive syndromes. They do not justify delaying a benzodiazepine while a patient is convulsing in front of you.

Time-based stages

The American Epilepsy Society (AES) 2016 algorithm is the framework most examination items assume. The clock starts at seizure onset, not at hospital arrival. Benzodiazepines should be given as soon as SE is recognized. The 0–5 minute window is stabilization, not an excuse to withhold a benzodiazepine if the patient is already seizing on arrival. Many neurointensive care protocols therefore give the benzodiazepine during those first minutes and have a second antiseizure drug (ASD) infusing by 5–20 minutes if convulsions continue—compressing the AES labels because delayed benzodiazepine therapy is the dominant failure mode.

The RAMPART trial (2012) showed that intramuscular midazolam 10 mg was at least as effective as intravenous lorazepam 4 mg in prehospital convulsive SE, because IM injection avoids the delay of obtaining a line. If there is no IV, do not delay for access: give IM midazolam.

Underdosing lorazepam (for example 1–2 mg in a 70 kg adult) is a classic error. 0.1 mg/kg in a 70 kg adult is 7 mg, capped at 4 mg per dose, with a second 4 mg dose permitted—so many adults should receive 4 mg, then another 4 mg if still seizing, not a timid 2 mg. Respiratory depression after a correctly dosed benzodiazepine is more often a consequence of ongoing SE than of the drug; placebo-controlled data in the AES evidence review showed more respiratory problems with untreated convulsions than with benzodiazepines.

ESETT 2019: benzodiazepine-refractory convulsive SE

Once an adequate benzodiazepine has failed, the patient has established SE. The Established Status Epilepticus Treatment Trial (ESETT; Kapur and colleagues, New England Journal of Medicine 2019) randomized children and adults 2 years and older in the emergency department to one of three 10-minute infusions:

• Levetiracetam 60 mg/kg IV, maximum 4500 mg
• Fosphenytoin 20 mg phenytoin equivalents (PE)/kg IV, maximum 1500 mg PE
• Valproate 40 mg/kg IV, maximum 3000 mg

The primary outcome—clinically evident seizure cessation and improving consciousness at 60 minutes without additional ASD—occurred in 47% with levetiracetam, 45% with fosphenytoin, and 46% with valproate. The trial stopped for futility of showing superiority or inferiority. About half of patients still need further therapy. Numerically more hypotension and intubation occurred with fosphenytoin and more deaths with levetiracetam; those differences were not statistically significant. About 10% of enrollments were psychogenic nonepileptic events, a reminder to verify that the event is epileptic.

Choose among the three by contraindication and logistics, not by a belief that one agent is stronger:

• Prefer levetiracetam when liver disease, mitochondrial disease, pregnancy, or thrombocytopenia make valproate or fosphenytoin unattractive. Expect psychiatric or behavioral effects later, not hemodynamic collapse during the load.
• Prefer fosphenytoin when a sodium-channel agent is desired and the patient is not hypotensive or in high-grade atrioventricular block. Monitor blood pressure and ECG. Infuse no faster than 150 mg PE/min.
• Prefer valproate when idiopathic generalized epilepsy is the background (myoclonus, absence) or when a sodium-channel agent recently failed, unless the patient is pregnant, might be pregnant, has mitochondrial disease (especially POLG), or has liver failure.

Avoid valproate in pregnancy (neural-tube defects and valproate embryopathy), in suspected POLG-related mitochondrial disease (fulminant hepatic failure), and in significant hepatic synthetic failure or active pancreatitis. Carbapenems collapse serum valproate levels—do not add meropenem and expect valproate to hold.

Refractory and super-refractory SE

Refractory status epilepticus (RSE) is SE that continues despite an adequate benzodiazepine plus a correctly dosed second ASD. Roughly one-quarter to one-third of convulsive SE becomes refractory.

Super-refractory status epilepticus (super-RSE) is SE that continues or recurs 24 hours or more after the start of anesthetic therapy, including seizures that return while anesthesia is being weaned. Super-RSE should trigger a hunt for new-onset RSE (NORSE) etiologies: autoimmune encephalitis, occult infection, inborn errors of metabolism, and inflammatory syndromes such as FIRES in younger patients. Immunotherapy is considered when those diagnoses are plausible, not as a default third-line ASD.

Anesthetic coma, cEEG, and weaning

When third-line therapy is anesthesia, intubate, support blood pressure, and place cEEG. Clinical cessation of convulsions does not mean electrographic cessation. After convulsive SE, a substantial minority have nonconvulsive status epilepticus (NCSE) or frequent electrographic seizures. Treat electrographic SE, not only visible shaking.

Common ICU infusions:

• Midazolam: high-dose infusion; tachyphylaxis after many hours; an active metabolite accumulates in renal failure.
• Propofol: rapid on and off. Watch for propofol infusion syndrome after high dose for a prolonged time, especially in catecholamine-supported patients: unexplained metabolic acidosis, rhabdomyolysis, hyperkalemia, and cardiac failure. Cap prolonged high-dose use.
• Pentobarbital or thiopental: effective but long half-life, hypotension, ileus, immunosuppression, and delayed awakening that confounds prognosis.
• Ketamine: NMDA antagonism used increasingly in super-RSE; often raises blood pressure (helpful in anesthetic vasodilatation) and increases secretions.

The usual electrographic goal is seizure suppression. Burst-suppression is often targeted with barbiturates but is not proven to improve outcome compared with seizure control without deep suppression. After 24–48 hours of control, wean the anesthetic slowly on cEEG. If seizures recur, resume anesthesia, add or change a maintenance ASD, and re-investigate etiology. Do not extubate a patient who is still in NCSE simply because the motor activity stopped.

Nonconvulsive SE

Suspect NCSE in unexplained coma, fluctuating consciousness after a convulsion, subtle nystagmus or facial twitching, or failure to wake after sedation is lightened. Diagnosis is EEG-based; Salzburg criteria are the research standard. Treatment follows the same ladder. A carefully observed benzodiazepine trial on EEG can be diagnostic as well as therapeutic in selected patients. Do not declare anoxic injury or sedation hangover until NCSE is excluded with EEG.

Post-anoxic myoclonus

Early myoclonus after cardiac arrest is not a single entity and is not, by itself, a reason to withdraw life-sustaining treatment.

Malignant myoclonus or myoclonic status in a still-comatose patient, especially with a highly malignant EEG (suppression, burst-suppression, or identical bursts time-locked to jerks), is associated with poor outcome. Even then, Neurocritical Care Society 2023 neuroprognostication guidance, affirmed by the American Academy of Neurology, requires a multimodal approach. Myoclonus is one data point, not a death sentence, and assessment waits until after rewarming and clearance of sedatives (typically 72 hours or more).

Lance–Adams syndrome is chronic action and intention myoclonus that typically declares itself as consciousness returns, often after a primarily hypoxic (respiratory) arrest. Vertex-predominant spikes on a continuous background, described by Elmer and colleagues, mark a phenotype with a meaningful chance of recovery. These patients may need levetiracetam, valproate, or clonazepam for the myoclonus—not an automatic pentobarbital coma and not early withdrawal.

The examination trap is a vignette of a temperature-managed post-arrest patient with multifocal jerks at hour 12. The correct move is cEEG, continued temperature control in the 32–37.5 °C range, and delayed multimodal prognosis—not myoclonus equals no chance.

Examination traps

• Giving 2 mg of lorazepam and calling benzodiazepines a failure.
• Starting fosphenytoin before any benzodiazepine.
• Combining tiny doses of three ASDs instead of one full load.
• Using valproate in a woman who might be pregnant, or in unexplained liver failure.
• Stopping the workup after a CT: missed meningitis, autoimmune encephalitis, and ASD nonadherence still need a search.
• Treating all post-arrest myoclonus as super-RSE or as certain poor outcome.