Pitfalls and Overreading in Pediatric EEG
Pitfalls and Overreading in Pediatric EEG
John Giang, MD
Purpose: teach the interpretation errors that turn a normal child into an "epileptic" one, the errors that let real abnormality through, and the habits that prevent both.
Why it matters
- Over-read EEGs are a leading cause of epilepsy misdiagnosis: in Benbadis' series of patients eventually proven to have psychogenic non-epileptic seizures, the EEGs originally called epileptiform showed fluctuating sharply contoured background, wicket spikes, hyperventilation slowing, and hypnagogic hypersynchrony, with over-emphasis on "phase reversals" a recurring reason (Benbadis & Tatum)
- In children the risk is doubled because the normal record is itself full of sharp, rhythmic, paroxysmal activity that is age- and state-dependent; the emphasis of Mizrahi's classic review is on normal features misconstrued as epileptiform rather than on missed abnormality (Mizrahi 1996).
- A false-positive report is rarely reversed: it leads to years of medication, driving and activity restrictions, and anchoring of every later clinician.
Key points
1. Overcalling age-appropriate sharp and rhythmic activity
The specific variants (posterior slow waves of youth, mu, lambda, hypnagogic hypersynchrony, vertex waves, POSTS, 14 and 6 Hz positive bursts, wicket, RMTD) have been described in earlier sections. The pitfall section is about the circumstances in which they get misread:
|
Situation |
Why it is overcalled |
Protective check |
|---|---|---|
|
Drowsiness and sleep onset |
Paroxysmal hypnagogic hypersynchrony (peak 4–9 yr) often has intermixed sharp or spike-like components, which are only an admixture of normal frequencies, not epileptiform; sustained monorhythmic 4 Hz runs can suggest an ictal pattern |
Pattern occurs only in drowsiness or at sleep onset in an otherwise normal record (Mizrahi 1996) |
|
Vertex transients |
In 2–4-year-olds they can be very high voltage, frontocentral rather than confined to Cz, repeat at 0.5–1 Hz, and may at first appear on one side only |
Eventually bilateral, synchronous and symmetric; tied to stage II features |
|
Spindles |
Asynchronous spindles in infants under 18 months and high-voltage, slower (10–12 Hz), frontal-dominant spindles in up to 5% of children aged 3–12 look like paroxysmal sharp fast activity |
Uniform duration, eventual contralateral appearance, other sleep features |
|
Arousal |
Children may show only a high-voltage, sharply contoured or polyphasic slow transient, then frontal 2.5–3.5 Hz rhythmic activity that spreads posteriorly and can last minutes, mimicking an evolving discharge |
Relate to the stimulus and to the sleep stage before it |
|
Hyperventilation |
The build-up is more abrupt, higher voltage, and more persistent in children (maximal 8–12 yr) and can be paroxysmal; only spike-and-wave or clear focal/lateralizing change is unequivocal abnormality |
Judge amplitude, abruptness and persistence against the wide normal range |
|
Photic stimulation |
Photomyogenic response (flash-locked eyelid and scalp muscle potentials, normal) is read as a photoparoxysmal response (spike-and-wave not locked to individual flashes) |
Check flash-locking and whether it outlasts the train |
|
Posterior slow waves of youth |
Fusion with alpha produces a polyphasic sharp-and-slow look, sometimes asymmetric and repetitive |
Block with eye opening and vanish with drowsiness |
|
Any sharply contoured wave |
"It phase reverses" is treated as proof; a phase reversal only localizes, it does not mean epileptiform |
Apply the morphologic criteria (asymmetric slope, after-going slow wave, disruption of background, field) before localizing |
2. Confusing artifact with seizure
- The individual artifacts were covered in the previous section. The interpretive error here is reading a rhythmic run without asking the three ictal questions: does it evolve in frequency, morphology and field; does it have a physiologic distribution; is there a clinical correlate or postictal change on the video.
- Monomorphic, monorhythmic activity that never changes frequency is the signature of artifact (and of benign rhythmic variants such as RMTD), whereas an electrographic seizure evolves.
- Paroxysmal generalized or focal slowing in a neurologically abnormal child is not epileptiform just because it is rhythmic; postictal slowing without spikes is a transient finding that resolves on serial EEGs, and interictal spikes may be suppressed immediately after a seizure.
- Periodic complexes after focal injury or in hypoxic-ischemic or metabolic encephalopathy are not evidence of epilepsy on their own.
3. Missing background abnormality
- The reader hunts for spikes and skips the first step: comparing the record with the "ideal" EEG for that age (dominant rhythm frequency, regional frequency composition, sleep characteristics, response to activation, expected variants).
- Turning the machine off while the child falls asleep loses the most productive segment: spikes are most abundant in light non-REM (stages II–III) and in nocturnal slow-wave sleep, attenuated in REM, and children pass through light sleep quickly and unpredictably; prolonged deep sleep adds little.
- A normal EEG does not refute epilepsy: spike occurrence is time-modulated across the wake-sleep cycle, so a daytime awake record may contain no spikes in a child with known epilepsy. Approximately 9% to 10% of prolonged video EEGs in patients with confirmed epilepsy show no evidence of epileptiform discharges (interictal epileptiform discharges, or IEDs), meaning a prolonged EEG can still be normal in about 9% to 10% of people who have epilepsy.
- Medication effects are misread as encephalopathy or as epileptiform: phenobarbital can produce high-voltage, anterior-dominant, waxing and waning fast bursts, and phenytoin or carbamazepine slow the dominant rhythm with rising dose; conversely, not every slowing in a deteriorating child can be blamed on the drug.
- Treatment "decoupling": acute antiseizure therapy may stop the clinical seizure while the electrographic seizure persists, in neonates and in older children, so cessation of movements is not proof of cessation of seizure.
4. Over-interpreting a real finding
- Identification is seldom the problem; significance is. Spike foci in children are temporal 40%, occipital 29%, central 23%, frontal 8%; about 90% of temporal and 75% of frontal foci are associated with clinical seizures, but fewer than 54% of occipital and central foci are, and there is no simple relation between spike abundance and seizure frequency (Mizrahi 1996).
- Half or more of childhood spike foci disappear within 1–2 years and fewer than 10% persist indefinitely, so a focus should not be reported as a permanent lesion marker; most are not structural (Mizrahi 1996).
- Generalized abortive spike-and-wave (bursts ≤1.5 s despite sleep, HV, or photic, peak 2–10 yr): 67% of children with it have seizures and 37% do not; among those with epilepsy 95% have generalized seizures but a few have only focal seizures, so it is a marker that must be correlated with the history, not a diagnosis (Mizrahi 1996).
- Interictal spike abundance is a poor guide to seizure control in focal epilepsy; in absence epilepsy, spike-and-wave disappears first in REM, then wakefulness, then slow-wave sleep, and loss in wakefulness tracks absence control.
Common pitfalls
- Reading the record for spikes before establishing the age-expected background, state, and sleep architecture.
- Treating a phase reversal, sharpness, or rhythmicity as sufficient for "epileptiform"; each is a property of many normal pediatric waveforms.
- Calling drowsiness or arousal patterns ictal because they are generalized, high voltage, and spread anteriorly to posteriorly.
- Reporting a vigorous hyperventilation response or a photomyogenic response as abnormal.
- Calling an EEG "normal" when it was never recorded through drowsiness and light sleep, or inferring "no epilepsy" from a single spike-free awake record.
- Attributing drug-induced beta or alpha slowing to encephalopathy, or attributing true deterioration to the drug.
- Reporting a finding without the clinical question: the significance of a spike focus or abortive spike-and-wave depends on the referral history.
- Writing "possible", "suspicious for", or "sharply contoured, may be epileptiform" in the impression; an equivocal report can be read by the clinician as positive. Say normal, or describe the variant by name.