Page 1262 · ЕЕГ-моніторинг судомної активності після зупинки серця — continuous EEG surveillance for hypoxic-ischemic brain injury after cardiac arrest
Global cerebral hypoxia-ischemia after cardiac arrest injures neurons unevenly — cortical layers III and V, hippocampal CA1, cerebellar Purkinje cells, and thalamus are especially vulnerable. This selective injury creates an electrically unstable substrate: damaged, disinhibited networks that can generate seizures in the first hours to days after return of spontaneous circulation (ROSC). Critically, a substantial share of that seizure burden produces no visible clinical sign at all.
Global hypoxia-ischemia does not damage the brain uniformly. Watershed cortical zones, hippocampal CA1 pyramidal neurons, and thalamocortical relay circuits are selectively vulnerable to the combination of oxygen deprivation and reperfusion injury. The result is a patchwork of dying, dysfunctional, and disinhibited neurons.
This patchwork is electrically unstable for several converging reasons:
• Loss of inhibitory interneurons: GABAergic interneurons are disproportionately sensitive to hypoxic injury, tipping local circuits toward net excitation • Excitotoxicity: excessive glutamate release during ischemia and reperfusion overactivates NMDA receptors, priming hyperexcitable networks • Reperfusion injury: free radical generation and calcium dysregulation continue to injure tissue for hours after ROSC, extending the window of instability • Cortical-subcortical disconnection: thalamocortical injury can generate the periodic, rhythmic discharges characteristic of post-anoxic EEG
The combination explains why seizures cluster in a defined post-arrest window rather than occurring randomly — they reflect an evolving injury process, not a fixed epileptic focus.
Post-arrest seizures are not "epilepsy" in the traditional sense. They are an acute symptomatic phenomenon of evolving hypoxic-ischemic injury, which is part of why their management differs from chronic epilepsy — and why their prognostic meaning is still being actively studied.
The central reason EEG monitoring matters so much after cardiac arrest is that a large share of post-arrest seizure activity is electrographically silent at the bedside. In a comatose, often sedated or paralyzed patient, the usual clinical hallmarks of seizure — limb jerking, eye deviation, tongue biting — may be absent, blunted, or masked entirely.
Without continuous EEG, this activity is invisible: a clinician examining the patient at the bedside sees no convulsion and has no reason to suspect ongoing electrographic seizures or even nonconvulsive status epilepticus. Only the EEG trace reveals it.
This is fundamentally different from seizures in an awake, unsedated patient, where the clinical exam is often sufficient. In the post-arrest ICU population, the clinical exam is an unreliable seizure detector — which is the core rationale for EEG as a monitoring tool rather than a one-time diagnostic test.
Not every post-arrest patient carries equal seizure risk. Features that raise suspicion and often prompt earlier or more intensive EEG monitoring include:
• Longer duration of cardiac arrest and delayed ROSC (longer no-flow/low-flow time) • Non-shockable initial rhythm (asystole, PEA) versus shockable (VF/VT) • Deeper or more prolonged coma on neurological exam • Myoclonic jerks observed clinically, which may represent the visible tip of a much larger electrographic iceberg • Absence of early spontaneous eye opening or purposeful movement
These factors do not diagnose seizures on their own — they simply identify which patients most warrant the continuous EEG surveillance discussed in the next stage.
A routine EEG samples 20–30 minutes of brain activity — a narrow window in a 72-hour period of dynamic post-arrest physiology. Seizures and epileptiform patterns can be paroxysmal, evolving, or intermittent, appearing and disappearing across hours. Continuous EEG (cEEG) monitoring is the only method that reliably captures this activity across the full acute risk window, particularly once sedation and neuromuscular blockade have removed the clinical signal.
The fundamental limitation of a routine, one-time EEG is temporal sampling. If a seizure or periodic pattern begins outside the 20–30 minute recording window, it is simply not seen — the report may read as reassuringly normal while a clinically important event unfolds an hour later.
Continuous EEG addresses this by recording uninterrupted for the duration judged necessary — typically at least 24 hours in comatose post-arrest patients, extended to 48 hours if the initial period is inconclusive, the patient remains comatose, or epileptiform activity is seen. This extended window matters because:
• Seizure onset after arrest is not confined to the first hour — it can emerge, evolve, or recur across the first two to three days • Background patterns evolve over time (e.g., burst suppression may transition toward continuous or discontinuous activity as recovery — or injury — progresses) • Reactivity and background trends, which carry prognostic value, are best assessed by observing change over time, not a single snapshot
A single "normal" 20-minute EEG performed on hospital day one does not rule out a seizure that begins on hospital day two. Continuous monitoring converts EEG from a diagnostic snapshot into a surveillance tool matched to the actual time-course of post-arrest brain injury.
Post-arrest ICU care commonly involves sedation for ventilator tolerance and comfort, and — particularly during targeted temperature management — neuromuscular blocking agents to suppress shivering. Both interventions have a side effect relevant to seizure detection: they blunt or eliminate the motor manifestations that would otherwise alert bedside staff to a seizure.
A patient under neuromuscular blockade cannot generate visible convulsive movements no matter how active the underlying electrographic seizure is. Deep sedation similarly dampens or eliminates subtle motor signs (eye flutter, facial twitching) that an alert observer might catch in a lighter sedation state.
This creates a specific clinical trap: the very interventions that make post-arrest critical care tolerable and effective can simultaneously erase the bedside seizure signal — reinforcing that EEG, not the clinical exam, must carry the burden of seizure detection in this population.
Delivering continuous EEG in the ICU setting involves more than attaching electrodes:
• Standardized electrode montage (typically international 10–20 system) applied and maintained for the monitoring duration • Real-time or near-real-time review by trained EEG readers, often supported by quantitative EEG (qEEG) trend displays that compress hours of raw data into browsable trends (e.g., amplitude-integrated EEG, spectrogram) • Standardized terminology (such as the American Clinical Neurophysiology Society Critical Care EEG Terminology) so that "generalized periodic discharges," "burst suppression," and other patterns are described consistently across readers and institutions • Coordination with temperature management and sedation protocols, since both interact with the EEG background independent of any underlying seizure activity • A defined threshold and workflow for escalating findings to the treating team — continuous data is only useful if it changes bedside decisions in a timely way
Not all post-anoxic EEG abnormalities are equivalent. Myoclonic status epilepticus, generalized periodic discharges (GPDs), and burst suppression are recurring, describable patterns — but each carries different implications for treatment urgency and prognosis, and distinguishing them (and their reactive versus unreactive variants) requires EEG expertise rather than a cursory read.
Myoclonic status epilepticus refers to prolonged, repetitive, often generalized myoclonic jerking arising in the acute post-anoxic period, frequently within the first 24 hours. On EEG it may correlate with generalized epileptiform discharges time-locked to the jerks, though the EEG-motor correlation is not always tight.
Historically, MSE was regarded as an almost uniformly grave prognostic sign (classically associated with the eponymous "Lance-Adams" contrast — a distinct, more benign post-anoxic myoclonus that emerges later, in patients who are waking up). More recent series have shown outcomes are heterogeneous: some patients with early MSE, particularly when the EEG background remains continuous and reactive, can have better-than-expected recovery. This has shifted practice away from treating MSE alone as an automatic trigger for withdrawal of life-sustaining therapy, and toward reading it in the context of the full EEG background and other prognostic modalities.
Generalized periodic discharges are repetitive, roughly regularly spaced sharp waveforms across the scalp. They sit on a spectrum sometimes called the "ictal-interictal continuum" — at one end, clearly interictal (background) patterns; at the other, clearly ictal (seizure) activity; in between, patterns of uncertain epileptogenicity that may or may not represent ongoing seizure-equivalent activity contributing to injury.
Features that push a reading toward the more concerning, seizure-like end of the spectrum include: • Discharge frequency approaching or exceeding ~2.5 Hz • Evolution in frequency, morphology, or spatial distribution over time (a hallmark of true electrographic seizures) • Associated rhythmic or fast activity superimposed on the periodic discharges • Clinical correlate (even subtle) that improves with a trial of antiseizure medication
Because GPDs are common after diffuse anoxic injury and do not automatically equal "seizure," their interpretation is one of the more nuanced tasks in post-arrest EEG reading.
Burst suppression describes an EEG pattern alternating between bursts of higher-amplitude activity and periods of marked attenuation ("suppression"), reflecting severe, diffuse cortical dysfunction. Its prognostic weight depends heavily on qualifiers:
• Unreactive burst suppression (no change with auditory, tactile, or noxious stimulation) is one of the more concerning post-anoxic patterns • Burst suppression with identical bursts (highly stereotyped, repeating burst morphology) is regarded as particularly malignant in several classification schemes • Reactive burst suppression, or burst suppression evolving toward a more continuous background over time, carries a comparatively less pessimistic signal
Reactivity testing — observing whether the EEG background changes in response to standardized stimulation — is itself an independent prognostic marker across post-anoxic patterns, not just burst suppression. A reactive background, whatever its baseline appearance, is generally a more favorable sign than an unreactive one.
The same word — "burst suppression" — can describe patterns with meaningfully different prognostic weight depending on reactivity and burst morphology. This is precisely why post-anoxic EEG interpretation is a specialist skill, not a simple checklist.
Detecting seizure activity is only half the problem. Clinically evident, convulsive seizures are usually treated promptly using standard status epilepticus protocols with little controversy. Purely electrographic, subclinical seizure activity is a genuinely harder call — aggressive antiseizure treatment brings real sedation burden, and the net benefit of suppressing electrographic-only activity in this population is still being actively studied.
When a post-arrest patient has a witnessed convulsive seizure, or a clear clinical correlate accompanying epileptiform EEG activity, the treatment logic resembles standard status epilepticus management: rapid-acting benzodiazepine for acute termination, followed by a loading dose of a second-line antiseizure medication (commonly levetiracetam or valproate, sometimes fosphenytoin) to prevent recurrence, escalating to anesthetic-dose therapy for refractory cases.
The rationale here is uncontroversial: uncontrolled convulsive status epilepticus causes ongoing metabolic stress, risk of aspiration and hemodynamic compromise, and additional neuronal injury layered on top of the primary hypoxic-ischemic insult. Treating it is standard of care regardless of the underlying post-arrest context.
The more difficult clinical question arises when EEG shows seizure or ictal-interictal-continuum activity with no visible clinical correlate — either because the patient is deeply comatose, sedated, or under neuromuscular blockade. Here, aggressive treatment is not an automatic decision, for several reasons:
• Sedation burden: escalating antiseizure medication, particularly anesthetic-dose therapy (e.g., propofol or midazolam infusions targeting burst suppression), deepens sedation — which delays and confounds the neurological examination that is itself central to later prognostication • Uncertain benefit: whether suppressing electrographic-only activity meaningfully improves neurological outcome, versus simply treating a marker of the underlying injury without altering its course, remains an area of active clinical investigation rather than settled fact • Reversibility trade-off: heavier sedation can be weaned later, but time lost to a deeply sedated, unexaminable patient during the critical prognostication window is a real cost
The pragmatic approach adopted by many centers is graded: treat clear seizures and rhythmic/periodic patterns that show a treatment-responsive clinical or physiological correlate, while using shared decision-making and often a trial-and-reassess strategy (e.g., a benzodiazepine trial to see if a periodic pattern resolves along with any subtle clinical correlate) for more ambiguous ictal-interictal-continuum patterns.
Randomized trials such as TELSTAR (treating rhythmic and periodic EEG patterns in critically ill patients) have specifically tested whether more aggressive suppression of these patterns changes outcome — reflecting that this is a genuinely open clinical question, not one where guidelines can simply mandate maximal treatment.
A useful mental framework for the electrographic-only case weighs three axes:
1. Pattern characteristics — frequency, evolution, and superimposed rhythmic/fast activity that push toward the more clearly ictal end of the continuum favor treatment 2. Background context — a continuous, reactive background alongside the discharges is more reassuring and may support a more conservative approach than an already-suppressed, unreactive background 3. Trajectory and timing — activity that is worsening, or persists despite a reasonable treatment trial, argues for escalation; activity that is stable or improving may support continued observation
Throughout, the decision should be revisited as new EEG data accrues rather than fixed once at first detection — this is precisely why the "continue monitoring" default exists in this simulator's treatment-consideration logic when seizure activity has not been flagged.
EEG findings after cardiac arrest are genuinely informative for prognosis — but they are contributory, not determinative. Contemporary guidelines converge on a multimodal approach that combines EEG with clinical neurological examination, somatosensory evoked potentials, neuroimaging, and serum biomarkers, deliberately avoiding any single test as the sole basis for a prognosis or a decision to withdraw life-sustaining therapy.
Every prognostic test after cardiac arrest — including EEG — has false-positive potential, meaning it can look falsely severe in a patient who nonetheless goes on to a meaningful recovery. Sedative and metabolic confounders can transiently depress EEG background, clinical exam findings, or evoked responses independent of the severity of the underlying injury.
Because the practical stakes of a prognostic decision after cardiac arrest — potentially informing withdrawal of life-sustaining therapy — are so high, current guidance deliberately requires convergence across multiple independent modalities before a poor-prognosis conclusion is drawn, rather than relying on any one signal, however concerning it looks in isolation.
The core safeguard in modern post-arrest neuroprognostication is redundancy: a single malignant EEG pattern raises concern and warrants careful evaluation, but by design it does not, on its own, dictate a decision.
EEG findings are typically interpreted alongside:
• Clinical neurological examination: pupillary light reflex, corneal reflex, motor response to pain — absent bilateral pupillary and corneal reflexes at defined time points are established poor-prognosis indicators, particularly free of sedative confounding • Somatosensory evoked potentials (SSEP): bilateral absence of the cortical N20 response is one of the most specific (low false-positive rate) poor-prognosis findings available, though it does not assess the full breadth of cortical function • Neuroimaging: diffuse cerebral edema, loss of grey-white differentiation on CT, or extensive diffusion restriction on MRI reflect the anatomic extent of hypoxic-ischemic injury • Serum biomarkers: neuron-specific enolase (NSE), among others, trending at defined post-arrest time points adds a biochemical injury-severity signal
EEG contributes patterns like unreactive burst suppression, suppressed/isoelectric background, or status epilepticus arising from an unreactive background as part of this constellation — its value comes from how it corroborates or diverges from the other four domains, not from acting alone.
Guidelines generally recommend deferring definitive neuroprognostication until at least 72 hours after ROSC, and later still if targeted temperature management, prolonged sedation, or significant renal/hepatic impairment could be delaying drug clearance and confounding both the exam and the EEG background.
An EEG read as maximally concerning on hospital day one, while the patient is still hypothermic and heavily sedated, carries far less prognostic weight than the same pattern persisting on hospital day three or four after sedative washout. This is why continuous EEG monitoring (Stage 2) and careful pattern recognition over time (Stage 3) feed directly into responsible neuroprognostication — the trajectory across days is at least as informative as any single reading.