10 Seizures and status epilepticus on EEG

No single waveform identifies a seizure. Spikes occur between seizures, rhythmic activity occurs in sleep and in encephalopathy, and fast activity occurs with muscle. What separates an ictal pattern from all of these is that it changes over time in an orderly way: a seizure recruits and spreads, and the EEG shows that as evolution in frequency, morphology or location. The physiology is a cortical population that is progressively synchronized, so the discharge frequency shifts as the seizure builds, the waveforms change shape as more of the cortex joins, and the field moves as the seizure propagates.

The ACNS 2021 terminology turns this into a definition that can be applied at the reading station (Hirsch LJ 2021). Evolution is at least 2 unequivocal, sequential changes in frequency, morphology or location, with a frequency change of at least 0.5 Hz, and each stage lasting at least 3 cycles. Two changes, not one: a pattern that speeds up once and then holds has shown a change, not evolution. Sequential: the changes follow one another in time. Unequivocal: if you have to argue for it, it is not there. The 3-cycle rule stops a single odd wave from counting as a stage.

Muscle can hide the onset

A convulsive seizure floods the record with muscle artifact at exactly the moment you most want to see the brain. Look for the first change before the muscle starts: a focal attenuation, a run of rhythmic activity, a shift in the background. Then look at the post-ictal slowing after the muscle stops, which confirms that something cerebral happened and often lateralizes it. The seizure is read at its edges when its middle is obscured.

What this changes

Evolution is the criterion that separates an electrographic seizure from the rhythmic and periodic patterns of the ICU, which is why Module 11 builds on this module. It is also the reason a rhythmic artifact from patting or ventilation is not a seizure, as Module 3 explains: it does not evolve.

The 2021 revision of the ACNS critical care terminology gives explicit definitions that apply to children as well as adults (Hirsch LJ 2021). Learn the table; the questions at the end of this module and in Module 11 are answered from it.

Two routes to an electrographic seizure

The first route is frequency: epileptiform discharges averaging more than 2.5 Hz for at least 10 s. Note the word averaging and the word more: discharges at exactly 2.5 Hz do not qualify, and a brief run above 2.5 Hz inside a slower pattern does not either. The second route is evolution: any pattern with definite evolution, as defined in the previous section, lasting at least 10 s. A pattern that meets neither route is not an electrographic seizure, however alarming it looks, and the rhythmic and periodic patterns that fall short belong to the ictal-interictal continuum that Module 11 describes.

Electroclinical seizures

A pattern becomes an electroclinical seizure when it has a definite clinical correlate time-locked to it, or when both the EEG and the clinical state improve after a parenteral antiseizure medication. The second clause matters at the bedside: a pattern that does not meet the electrographic definition can still be treated as a seizure if a trial of medication improves both the tracing and the child. The video is therefore part of the read, and so is the medication record.

Applying the definitions

When a colleague asks whether a pattern was a seizure, answer with the definition: which route it met, for how long, and with what clinical correlate. That is what makes the read reproducible, and it is what the sign-off for this module tests.

The 2021 terminology defines status epilepticus in two ways, by continuous duration and by burden, and gives a separate rule for convulsive activity (Hirsch LJ 2021).

Electrographic status epilepticus is an electrographic seizure lasting at least 10 continuous minutes, or electrographic seizures occupying at least 20% of any 60-minute period. The burden rule captures the child who has many short seizures without any single long one: five 3-minute seizures in an hour add up to 15 minutes, which is 25% of the hour and meets the threshold, even though no seizure lasted 10 minutes. Reading a continuous record therefore means adding up seizure time, not only measuring the longest event, and the quantitative trends of Module 11 exist partly to make that arithmetic visible.

Electroclinical status epilepticus uses the same thresholds, at least 10 continuous minutes or 20% of any hour, with one change: only 5 continuous minutes if there is bilateral tonic-clonic activity. Convulsive status is declared sooner because its clinical danger is greater and because its treatment should not wait for a 10-minute mark.

Neonates are different

The neonatal definitions of Module 7 use their own thresholds, and mixing them up is a common error. A neonatal seizure needs only evolution and 10 s. Neonatal status epilepticus is seizures summing to at least 50% of any 1-hour epoch, not 20%. A burden that would be status in a 9-year-old is not status in a 2-day-old, and the reverse. Keep the two tables apart.

Why the thresholds exist

The burden rule is not arbitrary. Module 11 sets out the pediatric evidence that links seizure burden to outcome, and the thresholds in these definitions are the ones that evidence uses. Applying them consistently is what lets the report say "electrographic status epilepticus" and have the words mean the same thing to every reader.

Evolution is the rule, but knowing how seizures usually begin and look on the scalp makes the rule easier to apply.

Focal seizures often start as rhythmic theta or alpha over one region that slows and spreads as the seizure recruits neighbouring cortex, which is evolution in frequency and location happening in front of you. Some begin instead with focal attenuation, a flattening of the background over the region of onset, before rhythmic activity appears; the attenuation is easy to miss and is often the true onset. The post-ictal record then shows focal slowing over the same region, which helps lateralize a seizure whose onset was obscured.

Absence seizures are the exception to gradual onset. They start and stop abruptly with generalized 3 Hz spike-wave, the discharge of childhood absence epilepsy from Module 9. There is little to evolve: the pattern appears fully formed and ends as suddenly. The ictal and interictal discharge are the same pattern, and what makes a run ictal is its duration and its clinical correlate on video.

Tonic seizures and spasms show generalized fast activity or an electrodecrement, the abrupt drop in amplitude with low-voltage fast activity that Module 9 describes as the ictal signature of a spasm. These are the seizures most easily hidden by muscle, because the clinical event is a contraction, so the first change before the artifact and the recovery after it are the parts to study.

Reading with the pattern in mind

When a stereotyped event is recorded, ask which of these shapes it takes, then apply the definitions of the previous two sections: did it evolve, for how long, and what was the child doing? The sign-off for this module asks the resident to identify onset, evolution and offset in five recorded seizures, at least one generalized and one focal, which is the habit these patterns make possible.

A normal EEG during an event is strong evidence, but it is not proof that nothing happened in the brain. Some seizures do not reach the scalp.

Focal aware seizures may have no clear scalp correlate. A seizure confined to a small region of cortex, or to cortex that is deep or tangentially oriented, generates a field too small or too poorly aligned to be seen through the skull, the same geometry Module 1 describes for interictal discharges. Some deep or frontal seizures behave the same way: a mesial frontal or orbitofrontal onset may show only muscle and movement artifact until it spreads, by which time the onset is unreadable.

What this means for the read

A normal EEG during a stereotyped event does not always exclude a seizure, so the report should say what was recorded and what it does and does not show, and the clinician should correlate with the semiology. A rising epigastric sensation reported during a recording with no EEG change is not disproved by the tracing; it is a focal aware seizure until the history says otherwise. The opposite error is as real: a recorded event with no EEG change and a semiology that does not fit any seizure type is evidence for a non-epileptic event, and the report can say so. The tracing decides less than residents expect, and the video and the history decide more.

Practical rules