The electrical end of status epilepticus is not the end of the illness. Seizures may stop while the patient remains comatose, delirious, ventilator-dependent, profoundly weak, or cognitively altered. The resulting uncertainty is uncomfortable: the EEG may be improving, yet the bedside examination is not; sedatives have been discontinued, yet meaningful interaction has not returned; an MRI shows cortical diffusion restriction, and the family wants to know whether it represents permanent injury. These are not peripheral questions. They determine whether clinicians continue diagnostic work, reduce treatment burden, initiate rehabilitation, or begin discussing limits of life-sustaining therapy.

Recovery after status epilepticus should be considered across several domains rather than reduced to a single awakening time. Electrographic recovery means that seizures remain controlled through anesthetic withdrawal. Neurologic recovery includes arousal, language, memory, executive function, motor function, and behavior. Medical recovery requires liberation from ventilation and resolution of infection, hypotension, metabolic disturbances, and treatment-related organ dysfunction. Functional recovery concerns independence and participation. Long-term recovery includes epilepsy control, medication burden, mood, sleep, relationships, and return to work or education. These domains recover at different rates, and improvement in one does not guarantee improvement in the others.

The first principle of status epilepticus recovery is that prolonged encephalopathy is common. In a retrospective cohort of 232 adults with a first nonanoxic episode, 85.3% developed at least moderate postictal encephalopathy, and encephalopathy persisted beyond 14 days in 24.5% of survivors. Its severity was independently associated with etiology, duration of sedation, age, and premorbid disability—not consistently with status duration, seizure semiology, peri-ictal MRI changes, or specific ictal EEG features. This does not prove that seizures are biologically harmless. It demonstrates that the examination after seizure termination reflects a mixture of the underlying disease, treatment, premorbid reserve, and critical illness rather than seizure duration alone (Bode et al., 2024).

Postictal should therefore be a provisional physiologic description, not a diagnostic conclusion. A patient who has not awakened may still have intermittent electrographic seizures, recurrent nonconvulsive status epilepticus, or an ictal–interictal continuum pattern contributing to impaired consciousness. The patient may instead be experiencing residual anesthetic effect, accumulation of multiple antiseizure medications, renal or hepatic dysfunction, sepsis-associated encephalopathy, hypercapnia, hyperammonemia, withdrawal, delirium, a new structural lesion, progression of encephalitis, or the consequences of the original brain injury. Several of these processes frequently coexist.

The practical response is to construct a recovery timeline. The team should document the last definite electrographic seizure, the end of each anesthetic infusion, the cumulative sedative and antiseizure medication exposure, neuromuscular blockade, organ function, metabolic abnormalities, infectious complications, and the evolution of the examination. Serial findings are more useful than an isolated Glasgow Coma Scale score. Reappearance of spontaneous eye opening, tracking, localization, reproducible command following, purposeful asymmetry, sleep–wake cycling, or reliable responses to family members may be more informative than a single formal examination conducted during fever, dialysis, medication administration, or overnight sleep.

When consciousness does not improve as expected, seizure control should be reconfirmed rather than presumed from the absence of motor activity. Continuous EEG is preferable when intermittent seizures are plausible, particularly during and after anesthetic withdrawal. A short negative EEG reduces but does not eliminate that possibility. Conversely, rhythmic or periodic activity should not automatically be relabeled as recurrent status epilepticus. Frequency, evolution, burden, background, clinical correlation, and response to a carefully chosen treatment trial all matter. Suppression of an EEG pattern without improvement in the background or clinical state may demonstrate pharmacologic EEG suppression rather than treatment of an injurious ictal process.

Medication exposure deserves particular attention. Adequate benzodiazepine treatment should never be withheld during active convulsive status epilepticus out of fear that it will obscure the later examination. Once seizures are controlled, however, the indication for every continuing sedative and antiseizure medication should be reassessed. Prolonged infusions, repeated loading doses, active metabolites, organ dysfunction, and polytherapy can collectively delay awakening. Bode and colleagues found an exploratory association between greater midazolam exposure and more severe postictal encephalopathy, but their observational design cannot determine whether midazolam caused the encephalopathy or merely marked more severe disease. The appropriate inference is to minimize unnecessary continuing exposure, not to undertreat active status epilepticus (Bode et al., 2024).

MRI is similarly valuable but easy to overinterpret. Peri-ictal abnormalities can include cortical or hippocampal diffusion restriction, FLAIR hyperintensity, thalamic involvement, and focal or regional hyperperfusion. In a prospective study of 206 patients, 45% had a peri-ictal MRI abnormality. Among those with follow-up imaging, 59% of abnormalities had resolved by approximately one week; among patients with persistent changes who underwent later imaging, most resolved by approximately three weeks. A minority developed lasting structural abnormalities. Thus, cortical diffusion restriction after status epilepticus is not synonymous with completed infarction, although extensive or persistent abnormalities may still reflect clinically important seizure-related injury (Bosque Varela et al., 2023).

The correct interpretation depends on distribution, vascular anatomy, perfusion, the timing of imaging relative to seizures, the underlying etiology, and evolution on follow-up. Peri-ictal MRI abnormalities may identify a brain region exposed to intense metabolic stress without establishing irreversibility. They may also carry information about later epilepsy. A 2026 observational cohort of 135 adults with de novo status epilepticus found a four-year cumulative seizure probability of 61% among patients with peri-ictal DWI or FLAIR abnormalities, compared with 36% among those without a peri-ictal abnormality and 13% among those with isolated ASL hyperperfusion. This is potentially useful for follow-up and medication planning, but it is not a validated rule for predicting cognitive recovery or treatment futility (Orav et al., 2026).

The recovery period is also a second phase of critical illness. In one ICU cohort, 96% of patients remained altered after termination of status epilepticus, with a median duration of two days. Eighty percent developed at least one complication, 55% developed delirium, and 21% developed a nosocomial infection. Mechanical ventilation, hypotension, infection, immobility, disrupted sleep, restraints, catheters, malnutrition, and medication toxicity can all prolong encephalopathy and degrade functional recovery. These associations do not prove that every complication is causal, because the sickest patients experience both more complications and worse outcomes. They do identify a clinically actionable period in which seizure freedom alone is insufficient care (Baumann et al., 2020).

Prognostication should begin with etiology, premorbid function, physiologic reserve, and the severity of the continuing brain disorder. A 2026 two-center observational study of 967 adults found that status epilepticus was terminated in 95%, 48.5% returned to premorbid neurologic function by discharge, in-hospital mortality was 7.9%, and 30-day mortality was 13.9%. Acute symptomatic status epilepticus and intracranial hemorrhage were associated with a lower likelihood of recovery, whereas known epilepsy was associated with a greater likelihood. Nonconvulsive status epilepticus with coma had the strongest and most consistent association with mortality and failure to recover. These are cohort-level associations. They can inform the prior probability communicated to a family but cannot determine the outcome of a sedated individual whose etiology remains uncertain (De Stefano et al., 2026).

Severity scores are useful for research stratification, triage, and communication, but they are weak instruments for individual decisions to withdraw treatment. A systematic review and meta-analysis found only moderate discrimination for commonly used scores such as STESS, EMSE, and END-IT. Their negative predictive values were relatively high, but positive predictive values for in-hospital death were only approximately 20%–35% at commonly studied thresholds. A high score therefore identifies a high-risk population but misclassifies too many survivors to function as a stand-alone futility rule (Yuan et al., 2023).

Treatment refractoriness also changes prognosis without eliminating the possibility of recovery. In a post hoc analysis of 996 episodes from the prospective SENSE registry, status epilepticus was successfully treated in 84.1%. The probability of seizure termination, survival, and favorable functional outcome fell after failure of the first two treatment steps, supporting the clinical distinction of refractory status epilepticus. Importantly, the probability of control did not continue collapsing toward zero with every subsequent treatment attempt. Approximately one third of refractory episodes remained treatable even after several unsuccessful steps. Treatment escalation must still be proportional to etiology, medical complications, treatment toxicity, and the patient’s values, but the number of failed medications is not itself a neurologic death sentence (Beuchat et al., 2024).

NORSE illustrates both the possibility and the incompleteness of recovery. In a single-center series of 25 adults, 96% developed super-refractory status epilepticus. Sixteen survived hospitalization, and 13 of those 16 improved functionally over time; median modified Rankin Scale score improved from 4 at discharge to 2 at the last available follow-up. The result is encouraging, but the cohort was small, retrospective, and affected by survivor selection and decisions to withhold or withdraw therapy. It demonstrates the existence of delayed meaningful recovery, not its probability in every patient (Stretti et al., 2024).

A multicenter study of 48 patients discharged after cryptogenic NORSE provides the necessary counterweight. Nine subsequently died, mostly within six months. At a median follow-up of 23 months, only two of the original 48 had completely recovered, while 18 had achieved a favorable modified Rankin Scale outcome. Among the 39 longer-term survivors, 79% had significant cognitive impairment and 54% had psychiatric morbidity; only seven of 30 previously employed patients had returned to work. The apparent contrast with the Zurich cohort reflects differences in population, outcome definition, denominator, and follow-up—not a contradiction. Functional independence may improve substantially while memory, executive function, mood, epilepsy, and vocational participation remain profoundly affected (Costello et al., 2024).

When the examination remains heavily confounded and no devastating underlying injury has been demonstrated, a time-limited trial of continued support can be reasonable. The duration cannot be standardized across etiologies. The trial should define what is being treated, which confounders are being removed, when EEG or imaging will be repeated, what constitutes meaningful improvement, and which burdens would make continued treatment inconsistent with the patient’s values. This is not an instruction to continue every treatment indefinitely. It is an attempt to make uncertainty explicit and prevent an arbitrary awakening deadline from masquerading as evidence.

Rehabilitation should begin before prognostic certainty. In the ICU, this means restoring day–night cues, reducing unnecessary sedation and restraints, treating pain, correcting sensory deprivation, mobilizing when safe, assessing swallowing and communication, and engaging physical, occupational, and speech therapy. For NORSE and FIRES, international consensus strongly recommends motor and cognitive rehabilitation, serial neuropsychological assessment, and screening for mood and sleep disorders. The same document explicitly acknowledges that direct trial evidence for a specific rehabilitation strategy is absent; these are expert-consensus recommendations supported by the high burden of observed disability rather than randomized comparisons (Wickström et al., 2022).

Medication rationalization is another component of recovery. A patient who required several agents during refractory status epilepticus should not automatically remain on every drug indefinitely, but neither should the regimen be rapidly dismantled without considering etiology, EEG, MRI, treatment refractoriness, prior epilepsy, and seizure recurrence. In a cohort of 257 survivors of a first nonhypoxic status episode, 21% developed an unprovoked seizure during follow-up, usually months later. Five-year risk varied markedly—from under 10% for some acute secondary or toxic etiologies to approximately 30% after acute primary CNS injury and substantially higher with progressive disease. This heterogeneity argues against a universal duration of antiseizure treatment (Lattanzi et al., 2024).

A pragmatic approach is to identify which agents are providing durable protection, which were added only during escalation, and which are impairing arousal, gait, behavior, appetite, or cognition. Simplification should generally occur sequentially, with particular caution around medications whose abrupt withdrawal can provoke seizures. The discharge plan should specify maintenance dosing, rescue treatment, what to do after a missed dose, warning signs of recurrent nonconvulsive seizures, and how follow-up EEG or imaging may influence later reduction. The 2026 MRI data may eventually refine this process, but they remain observational and should complement rather than replace clinical judgment.

Follow-up should not end with seizure freedom. A practical, although not trial-validated, pathway includes early epilepsy or neurocritical-care review within several weeks, reassessment of cognition and medication burden after the acute confusional period has cleared, and repeated evaluation over the following months. Formal neuropsychological testing is most informative once participation is reliable, but screening for memory, executive dysfunction, depression, anxiety, irritability, sleep disturbance, and caregiver strain can begin earlier. Driving restrictions and occupational precautions must follow local law and the individual’s seizure risk. Return to work or education should be treated as a rehabilitation outcome rather than assumed from an acceptable modified Rankin Scale score.

The central discipline after status epilepticus is to preserve two truths at once. Prolonged seizures, the underlying neurologic disease, and intensive treatment can produce death or enduring disability. Meaningful recovery can nevertheless unfold over days, weeks, or months, including after refractory and super-refractory episodes. Good care neither minimizes the injury nor converts uncertainty into nihilism. It repeatedly verifies seizure control, removes reversible barriers to awakening, treats the complications of critical illness, initiates rehabilitation early, and updates prognosis from serial evidence rather than from a predetermined clock.

References

Baumann, S. M., Semmlack, S., De Marchis, G. M., Hunziker, S., Rüegg, S., Marsch, S., & Sutter, R. (2020). Frequency and implications of complications in the ICU after status epilepticus: No calm after the storm. Critical Care Medicine, 48(12), 1779–1789. https://doi.org/10.1097/CCM.0000000000004642

Beuchat, I., Novy, J., Rosenow, F., Kellinghaus, C., Rüegg, S., Tilz, C., Trinka, E., Unterberger, I., Uzelac, Z., Strzelczyk, A., & Rossetti, A. O. (2024). Staged treatment response in status epilepticus: Lessons from the SENSE registry. Epilepsia, 65, 338–349. https://doi.org/10.1111/epi.17817

Bode, C. M., Kristensen, S. B., Olsen, H. T., Cornwall, C. D., Roberg, L., Monsson, O., Krøigård, T., Toft, P., & Beier, C. P. (2024). Postictal encephalopathy after status epilepticus: Outcome and risk factors. Neurocritical Care, 40(3), 1025–1035. https://doi.org/10.1007/s12028-023-01868-1

Bosque Varela, P., Machegger, L., Oellerer, A., Steinbacher, J., McCoy, M. R., Pfaff, J., Trinka, E., & Kuchukhidze, G. (2023). Imaging of status epilepticus: Making the invisible visible. A prospective study on 206 patients. Epilepsy & Behavior, 141, 109130. https://doi.org/10.1016/j.yebeh.2023.109130

Costello, D. J., Matthews, E., Aurangzeb, S., Doran, E., Stack, J., Wesselingh, R., Dugan, P., Choi, H., Depondt, C., Devinsky, O., Doherty, C., Kwan, P., Monif, M., O’Brien, T. J., Sen, A., & Gaspard, N. (2024). Clinical outcomes among initial survivors of cryptogenic new-onset refractory status epilepsy (NORSE). Epilepsia, 65(6), 1581–1588. https://doi.org/10.1111/epi.17950

De Stefano, P., Baumann, S. M., Fisch, U., Grzonka, P. S., Rochat, T., De Marchis, G. M., Dittrich, T. D., Hunziker, S., Rüegg, S. J., Kleinschmidt, A., Quintard, H., Seeck, M., & Sutter, R. (2026). Impact of etiology on mortality and recovery in patients with status epilepticus. Neurology, 106(5), e214624. https://doi.org/10.1212/WNL.0000000000214624

Lattanzi, S., Orlandi, N., Giovannini, G., Brigo, F., Trinka, E., & Meletti, S. (2024). The risk of unprovoked seizure occurrence after status epilepticus in adults. Epilepsia, 65(4), 1006–1016. https://doi.org/10.1111/epi.17912

Orav, K., Bosque-Varela, P., Lauth, W., Machegger, L., Jannone-Pedro, N., Leitinger, M., Pfaff, J., Trinka, E., & Kuchukhidze, G. (2026). Peri-ictal MRI abnormalities and risk of unprovoked seizures after de novo status epilepticus. Neurology, 106(7), e214774. https://doi.org/10.1212/WNL.0000000000214774

Rajajee, V., Muehlschlegel, S., Wartenberg, K. E., Alexander, S. A., Busl, K. M., Chou, S. H. Y., Creutzfeldt, C. J., Fontaine, G. V., Fried, H., Hocker, S. E., Hwang, D. Y., Kim, K. S., Madzar, D., Mahanes, D., Mainali, S., Meixensberger, J., Montellano, F., Sakowitz, O. W., Weimar, C., … Varelas, P. N. (2023). Guidelines for neuroprognostication in comatose adult survivors of cardiac arrest. Neurocritical Care, 38(3), 533–563. https://doi.org/10.1007/s12028-023-01688-3

Stretti, F., Bögli, S. Y., Casagrande, F., Eisele, A., Galovic, M., Keller, E., & Brandi, G. (2024). Long-term outcome in new onset refractory status epilepticus: A retrospective study. Critical Care, 28, Article 72. https://doi.org/10.1186/s13054-024-04858-7

Trinka, E., Cock, H., Hesdorffer, D., Rossetti, A. O., Scheffer, I. E., Shinnar, S., Shorvon, S., & Lowenstein, D. H. (2015). A definition and classification of status epilepticus—Report of the ILAE Task Force on Classification of Status Epilepticus. Epilepsia, 56(10), 1515–1523. https://doi.org/10.1111/epi.13121

Wickström, R., Taraschenko, O., Dilena, R., Payne, E. T., Specchio, N., Nabbout, R., Koh, S., Gaspard, N., Hirsch, L. J., & International NORSE Consensus Group. (2022). International consensus recommendations for management of new onset refractory status epilepticus including febrile infection-related epilepsy syndrome: Statements and supporting evidence. Epilepsia, 63(11), 2840–2864. https://doi.org/10.1111/epi.17397

Yuan, F., Damien, C., & Gaspard, N. (2023). Prognostic scores in status epilepticus: A systematic review and meta-analysis. Epilepsia, 64, 17–28. https://doi.org/10.1111/epi.17442