Now we will learn about an electrical emergency whose treatability changes minute by minute. The first attending-level decision is not which anesthetic infusion to use. It is when to stop calling an event “a seizure,” declare status epilepticus, and activate a coordinated treatment sequence. Waiting too long permits a potentially self-limited seizure to become self-sustaining; treating every abnormal movement as status exposes patients with syncope, rigors, dystonia, posturing, toxic clonus, or functional seizures to unnecessary intubation and anesthetic coma. The bedside task is therefore simultaneous recognition, resuscitation, treatment, and diagnostic revision.

The historical thirty-minute definition is no longer an acceptable treatment threshold. The International League Against Epilepsy framework separates two operational times. Time point one, or t1, is when ongoing seizure activity is abnormally prolonged and treatment should begin. Time point two, or t2, is when continued activity may carry increasing risk of neuronal injury, network alteration, and other long-term consequences. For bilateral tonic–clonic status, t1 is five minutes and t2 is approximately thirty minutes. For focal status with impaired consciousness, the best estimates are ten minutes and more than sixty minutes. For absence status, t1 is approximately ten to fifteen minutes and t2 remains uncertain. These are phenotype-specific estimates, not biologic cliffs. Thirty minutes is not a safe observation window, and crossing thirty minutes does not guarantee irreversible injury. Etiology, age, temperature, oxygen delivery, arterial pressure, seizure burden, and the underlying brain all modify risk.

A patient with continuous bilateral tonic–clonic activity for five minutes has status epilepticus. So does a patient with recurrent seizures who does not recover meaningfully toward baseline between them. If the onset time is unknown and the patient is still convulsing when the treatment team arrives, the practical assumption should be that t1 has already been crossed; starting a new five-minute observation period is an avoidable error. The classification should then remain multidimensional. Semiology asks whether prominent motor manifestations are present. Etiology distinguishes acute symptomatic, remote symptomatic, progressive, and unknown causes. EEG describes the electrographic correlate, and age changes both likely causes and treatment risk. These axes matter because convulsive status can evolve into nonconvulsive status: tonic–clonic movements may become less vigorous as motor output fails while cortical seizure activity continues. The disappearance of dramatic movement is therefore a clinical observation, not proof of cerebral seizure termination.

The reason time changes treatment response is partly synaptic. Normal seizure termination depends on recruitment of inhibition, restoration of ionic gradients, neurotransmitter clearance, and interruption of synchronized excitation. During prolonged experimental status epilepticus, synaptic benzodiazepine-sensitive GABA-A receptors are progressively internalized, while NMDA and AMPA receptor–mediated excitation becomes more prominent. Intracellular calcium rises, mitochondrial buffering is stressed, extracellular potassium accumulates, astrocytic clearance becomes less effective, and excitatory transmission increasingly sustains itself. Much of the detailed temporal evidence comes from animal and cellular models, so it should not be translated into a precise human countdown. The clinically durable conclusion is narrower: benzodiazepines become less reliable as treatment is delayed, and prolonged status increasingly requires drugs acting through additional mechanisms. The modern synthesis of this receptor-trafficking biology is reviewed by Naylor.

The systemic physiology also evolves. Early generalized convulsions produce sympathetic activation, hypertension, tachycardia, increased cerebral blood flow, high cerebral glucose and oxygen consumption, lactate generation, and hyperthermia. Initially, increased perfusion may partially match metabolic demand. With continued seizures, hypoxemia, aspiration, apnea, hypercapnia, hypotension, acidosis, hyperthermia, rhabdomyolysis, and pulmonary edema can erode that compensation. Cerebral metabolic demand may then exceed substrate delivery, particularly in previously injured or ischemic tissue. This is why seizure termination, oxygenation, ventilation, temperature control, and hemodynamic support are parts of the same neuroprotective intervention. It is also why severe lactic acidosis immediately after generalized convulsions does not by itself prove shock or mandate bicarbonate; its trajectory after seizure control is more informative. Failure to clear should prompt reassessment for continued motor or electrographic seizures, hypoperfusion, sepsis, toxin exposure, or another source of lactate.

Recognition begins with a clock and a witness. Record the last-known baseline, actual or estimated seizure onset, intervening recovery, medications already administered, and the time visible convulsions end. Obtain video when feasible because semiology often disappears before the consultant arrives. Functional seizures remain an important source of iatrogenic harm: approximately ten percent of ESETT enrollments were ultimately judged to have psychogenic nonepileptic events. No single feature reliably excludes epilepsy during an emergency, and an uncertain patient with sustained bilateral convulsions generally warrants the first benzodiazepine. The diagnostic obligation becomes more important with each escalation. Repeated anesthetic dosing and intubation should not proceed automatically when the evolving semiology, physiology, video, and EEG increasingly argue against epileptic status.

The first actions occur in parallel. Protect the patient from injury, position and suction the airway, provide oxygen for hypoxemia, and assist ventilation if apnea or ineffective breathing develops. Apply ECG, blood-pressure, and oxygen-saturation monitoring; obtain a finger-stick glucose; establish intravenous or intraosseous access; draw essential studies; and prepare both the benzodiazepine and the durable antiseizure load. One person should own the timeline and announce when treatment milestones are reached. Another should obtain the focused history and investigate the cause. The CT scanner, pharmacy preparation, and airway assessment must not become sequential barriers to seizure treatment.

Intubation is indicated for persistent apnea, refractory hypoxemia, loss of airway protection, major aspiration, severe hemodynamic deterioration, or ongoing convulsions requiring anesthetic-dose therapy. It is not required merely because an adequately treated patient is briefly postictal, acidemic, or somnolent. Conversely, fear of intubation should not cause benzodiazepine underdosing. Controlled trial data reviewed in the American Epilepsy Society guideline show that respiratory complications are importantly attributable to untreated status itself; benzodiazepine-treated patients did not have more respiratory depression than placebo-treated patients in the foundational prehospital evidence. Prepare to support the airway, but give the antiseizure dose capable of working.

For an adult with intravenous access, lorazepam is given at 0.1 milligram per kilogram, capped at 4 milligrams per dose, and may be repeated once after approximately five minutes if convulsions continue. Without immediate intravenous access, intramuscular midazolam at 0.2 milligram per kilogram, capped at 10 milligrams—and practically 10 milligrams in an adult over 40 kilograms—is preferred. Intravenous diazepam at 0.15 to 0.2 milligram per kilogram, capped at 10 milligrams, is another effective option and can be repeated once. Dividing an intended dose into serial 1-milligram lorazepam aliquots is not cautious treatment; it delays attainment of an effective brain concentration while the seizure becomes less responsive.

The RAMPART trial demonstrated the operational importance of route. Intramuscular midazolam stopped status before emergency-department arrival without rescue treatment in 73.4% of participants, compared with 63.4% receiving intravenous lorazepam. Intravenous lorazepam acted faster after entering the bloodstream, but establishing access delayed its delivery; intramuscular midazolam therefore won the race from treatment-box opening to seizure cessation. The lesson is not that intramuscular therapy is intrinsically stronger. It is that the fastest reliable route is superior to a theoretically ideal route that has not yet been obtained.

While the benzodiazepine is being administered, the durable antiseizure medication should already be selected and prepared. The Neurocritical Care Society guideline recommends urgent-control therapy after the initial benzodiazepine for patients presenting with status, even when visible convulsions terminate, unless an immediately reversible cause such as hypoglycemia has been definitively corrected. Benzodiazepines redistribute, and apparent early success can be followed by recurrence. The time windows in published algorithms are outer organizational boundaries, not instructions to wait twenty minutes before ordering the next medication.

For status persisting after an adequate benzodiazepine, ESETT compared levetiracetam 60 milligrams per kilogram up to 4,500 milligrams, valproate 40 milligrams per kilogram up to 3,000 milligrams, and fosphenytoin 20 milligrams phenytoin equivalents per kilogram up to 1,500 milligrams phenytoin equivalents, each infused over ten minutes. The stringent composite of no clinically apparent seizures and improving responsiveness at sixty minutes without additional antiseizure treatment occurred in 47%, 46%, and 45%, respectively. No agent was demonstrably superior, and major safety outcomes did not differ significantly. Importantly, failure to meet the composite did not always mean that convulsions persisted; some patients had stopped seizing but had not yet improved in responsiveness. ESETT nevertheless established that a single conventional second-line drug produces complete early success in only about half of benzodiazepine-refractory cases. The 2024 ACEP clinical policy consequently gives a Level A recommendation that adults with seizures refractory to appropriately dosed benzodiazepines receive a second-line agent and recognizes fosphenytoin, levetiracetam, and valproate as similarly efficacious options.

The choice among them should be physiologic rather than habitual. Levetiracetam has few drug interactions and little direct hemodynamic or cardiac effect, making it attractive when history, organ function, and seizure type are initially uncertain. A token 1,000-milligram dose is not equivalent to the ESETT regimen in a large adult. Renal dysfunction generally changes subsequent maintenance dosing more than the emergency loading decision. Valproate is broad-spectrum and avoids sodium-channel-mediated cardiac effects, making it useful when generalized, myoclonic, or absence mechanisms are possible. It is less attractive with severe hepatic dysfunction, known mitochondrial or urea-cycle disease, substantial thrombocytopenia, hyperammonemic vulnerability, or pregnancy when an effective alternative is available. Fosphenytoin has extensive experience in focal and structurally provoked convulsive status but requires ECG and blood-pressure monitoring. It is less attractive with hypotension, bradyarrhythmia, major conduction disease, sodium-channel-blocker poisoning, or a generalized epilepsy phenotype dominated by absence or myoclonic seizures. The dose must be prescribed in phenytoin equivalents; confusing milligrams of fosphenytoin with milligrams phenytoin equivalents is a preventable medication error.

Phenobarbital, usually loaded at 15 to 20 milligrams per kilogram, remains an effective alternative, particularly when sedative-hypnotic withdrawal is driving the physiology, but hypotension and respiratory depression become increasingly likely after benzodiazepines. Lacosamide is frequently used because it is easy to administer and has relatively few interactions, but its evidence base in established convulsive status is not equivalent to the three ESETT agents, and PR prolongation remains relevant. In a patient with known epilepsy who missed an intravenous-compatible home medication, reloading that drug may be sensible, but familiarity should not delay an adequately dosed proven urgent-control agent.

The seizure and its cause must be treated simultaneously. Finger-stick glucose comes first because hypoglycemia is rapidly reversible. Initial testing usually includes sodium, potassium, bicarbonate, calcium, magnesium, renal and hepatic indices, blood count, blood gas when indicated, antiseizure drug concentrations when measurable, toxicologic testing guided by context, and a pregnancy test when relevant. Known epilepsy prompts questions about missed doses, vomiting, absorption, interacting medications, alcohol or benzodiazepine withdrawal, sleep deprivation, infection, and recent regimen changes. New-onset status expands the search to ischemic stroke, intracranial hemorrhage, cerebral venous thrombosis, tumor, trauma, meningitis, encephalitis, autoimmune disease, posterior reversible encephalopathy syndrome, eclampsia, hypoxia, metabolic disturbance, and toxins.

Targeted etiologic therapy may matter more than adding another conventional antiseizure medication. Hypoglycemia requires dextrose; thiamine should be given promptly in a patient at risk for deficiency but must not delay glucose. Symptomatic severe hyponatremia requires hypertonic saline. Eclampsia requires magnesium sulfate. Isoniazid-associated seizures require pyridoxine. Sodium-channel-blocker toxicity may require sodium bicarbonate, and phenytoin is a poor reflex choice in that setting. Alcohol or benzodiazepine withdrawal requires restoration of GABAergic tone. Suspected bacterial meningitis or encephalitis warrants timely empiric antimicrobials and acyclovir when appropriate rather than waiting for perfect imaging or lumbar-puncture conditions.

Once circulation and visible convulsions are controlled, neuroimaging and CSF evaluation should follow the clinical phenotype. Most adults with new-onset status, focal findings, trauma, immunocompromise, malignancy, anticoagulation, or failure to recover require urgent CT, with vascular imaging when stroke, hemorrhage, venous thrombosis, or another vascular process is plausible. MRI becomes important when CT is unrevealing, deficits persist, encephalitis is suspected, or peri-ictal abnormalities must be distinguished from a primary structural lesion. Lumbar puncture should be pursued when infection, inflammation, or autoimmune encephalitis remains possible, but diagnostic testing should not interrupt time-critical seizure treatment.

Visible cessation is necessary but not always sufficient. After generalized convulsions, postictal coma, medication effect, hypercapnia, or an underlying structural lesion can explain delayed awakening; continued nonconvulsive status can look identical. Urgent EEG is warranted when purposeful improvement does not begin as expected, subtle facial or limb twitching persists, the eyes remain tonically deviated, unexplained nystagmoid movements occur, seizures recur, or neuromuscular blockade has removed the motor examination. A paralytic stops movement, not cerebral seizure activity. The current ICU synthesis emphasizes that EEG is not required to recognize or begin treating convulsive status, but it is required to diagnose ongoing nonconvulsive status reliably. The recent Lancet Neurology review similarly places rapid supportive care, etiologic assessment, benzodiazepines, durable antiseizure therapy, EEG, and ICU escalation within one continuous pathway.

If clinical or electrographic seizures continue after an adequate benzodiazepine and one fully dosed appropriate second-line medication, the patient has refractory status epilepticus. This is a response-defined state; no additional thirty- or sixty-minute duration requirement must be satisfied. Waiting for a serum drug concentration, repeating small benzodiazepine aliquots indefinitely, or sending the actively seizing patient through another diagnostic study is not an acceptable holding strategy. The team should prepare for airway control, anesthetic-dose seizure suppression, continuous EEG, and management of the hemodynamic consequences while continuing to identify the cause.

A 2026 proposal termed “Stage 1 Plus” places an additional warning point between t1 and t2—approximately ten minutes for generalized convulsive status, twenty minutes for focal status with impaired consciousness, and thirty to forty minutes for nonconvulsive status with coma—and suggests that treatment-naïve patients first encountered this late may merit earlier combination therapy. This is a biologically coherent but explicitly unvalidated proposal, not an ILAE treatment guideline. It does not establish routine initial triple therapy or justify replacing benzodiazepines with ketamine. Its useful bedside message is narrower: a patient arriving after twenty minutes of untreated convulsions should not be moved leisurely through a fresh sequential algorithm. Give the benzodiazepine, prepare and administer the durable load in parallel, and anticipate refractoriness. The proposal’s limitations are stated by Magro.

The first-hour endpoint is therefore not simply “the shaking stopped.” It is documented clinical and, when necessary, electrographic seizure control; adequate oxygenation, ventilation, temperature, and perfusion; a durable antiseizure regimen; treatment of immediately reversible causes; and a diagnostic plan capable of identifying structural, infectious, inflammatory, metabolic, and toxic precipitants. The next installment begins at the point where those measures fail: refractory and super-refractory status epilepticus, including anesthetic selection, ketamine, EEG treatment targets, duration of therapeutic coma, and the distinction between seizure suppression and burst suppression.

References