05 Normal variants and how not to overcall

An EEG cannot cause a seizure, but it can cause an epilepsy diagnosis. A child who is labelled epileptic on the strength of an overread tracing acquires years of antiseizure medication, driving and activity restrictions later on, and a family that watches every stare. Reversing that label is far harder than avoiding it, and the record of how it happens is well documented. In patients referred to an epilepsy centre and found not to have epilepsy, the EEG findings that had been read as epileptiform were mostly nonspecific fluctuations of the background and sharply contoured normal activity, often over the temporal regions. Recognized normal variants were also overread, among them wicket waves, hypnagogic hypersynchrony and the slowing produced by hyperventilation (Benbadis 2003; Benbadis 2008).

The common thread is that none of these findings is a discharge at all. They are the ordinary texture of a normal record, read by someone who was looking for a spike and found a wave that was sharp enough. That is the error this module is built to prevent. A resident who can recite the benign variants but has no rule for deciding when a transient is epileptiform is still at risk, because the sharply contoured background that causes most overreads does not have a name in the variant tables.

Two habits that protect the patient

The first is to read the state and the age before reading any transient. Drowsiness and light sleep are where most variants live, and where the background itself becomes sharper and more irregular; a sharp wave that appears only as a child drifts off deserves more suspicion of the reader than of the brain. The second is to apply an explicit rule, the IFCN criteria taught later in this module, to every candidate transient, and to write down how many criteria it met. A transient that meets three criteria does not become epileptiform because the referral question was seizures. The clinical history belongs in the clinical correlation, as Module 2 describes; it must not leak into the description.

The variants have names because they recur, and each belongs to a state and an age. Learn them as a table, then learn what the table cannot do.

Ages are from the IFCN glossary (Kane 2017) and from St. Louis 2016. Wickets and small sharp spikes are rare in young children, but a pediatric reader meets them in adolescents, and in the parents whose records sometimes arrive for comparison.

What the variants have in common

Each is tied to a state. Hypnagogic hypersynchrony, the 14 and 6 Hz bursts, small sharp spikes and phantom spike-wave belong to drowsiness and light sleep, which is the same state in which the background becomes sharper and overreading is commonest. Posterior slow waves of youth belong to the awake record and behave like the alpha rhythm they ride on: they block with eye opening, as Module 4 describes. Each also has a morphology that fails the epileptiform criteria in a characteristic way. Wickets have no after-going slow wave. The 14 and 6 Hz bursts are positive, the wrong polarity for most epileptiform discharges. Phantom spike-wave is too low in amplitude and too brief. Hypnagogic hypersynchrony is rhythmic and diffuse, without a spike.

The table also has a limit worth stating plainly: the sharply contoured background that causes most overreads is not on it. A variant list is a set of exceptions; the rule that decides everything else is the subject of the next section.

The IFCN glossary gives six features of an epileptiform transient, and says that a transient is epileptiform if it meets at least 4 of the 6 (Kane 2017):

The list turns an impression into a checklist. The first three criteria describe the shape of the wave itself; the fourth and fifth describe what it does to the record around it; the sixth is the field, the same test of a cerebral generator introduced in Module 1 and used against artifact in Module 3. A sharply contoured piece of background usually scores on the first criterion and perhaps the second, and then stops: it has no after-going slow wave, it does not interrupt anything because it is the background, and it has no field beyond the electrode where it caught the eye.

Where to set the bar

The criteria were tested prospectively. In a validation study of 100 patients, requiring at least 5 of the 6 gave a specificity of 95.7% and a sensitivity of 81.5%, and the authors judged 5 the best threshold for clinical use (Kural 2020). The lesson is not that 4 is wrong; it is that the cost of a false positive is high enough, for the reasons in the first section, that a stricter bar is defensible. In practice, a transient meeting 4 criteria deserves a careful search for a fifth, and a transient meeting 3 should not be called epileptiform at all.

Using the rule at the bench

Score every candidate, and score it in more than one montage, because the field (criterion 6) can only be judged where a phase reversal or a referential maximum is visible. Write the score down. The sign-off for this module asks exactly this: ten sharp transients chosen by faculty, each scored against the six criteria, with agreement on at least eight.

Even a transient that passes the criteria is not a diagnosis. Epileptiform discharges occur in children who have never had a seizure and never will. In a prospective study of healthy children recorded with digital EEG, 6.5% had epileptiform discharges and 4.2% had rolandic (centrotemporal) discharges; earlier studies had reported 3.5% to 5% (Borusiak 2010). The centrotemporal spike is the one a pediatric reader must hold most lightly, because it is both the signature of the commonest self-limited focal epilepsy of childhood, described in Module 9, and a finding in roughly one healthy child in twenty-five.

Epilepsy is a clinical diagnosis

The EEG supports a diagnosis that the history makes; it does not make one. A child referred for headaches, or for a syncope, or for a sleep problem, whose record shows occasional centrotemporal spikes, does not have epilepsy on that basis. The correct report describes the discharges accurately, in the description, and then states in the clinical correlation that epileptiform discharges of this type occur in a small percentage of healthy children and that the diagnosis of epilepsy rests on the clinical events. A reader who instead writes "consistent with epilepsy" has converted a population base rate into an individual diagnosis.

The other direction

The same reasoning runs the other way, and Module 2 makes it a reporting rule: a normal routine EEG does not exclude epilepsy. Many focal and generalized discharges appear only in drowsiness and sleep, which is why a record without sleep is a weaker negative, as Module 8 discusses. The reader's job is to describe what the tracing shows and to state plainly what it can and cannot settle. The discharge count belongs to the EEG; the diagnosis belongs to the clinician who has the history, and often that is the same person wearing a different hat.

Knowing that a waveform is benign is half the job; the other half is writing it so that the next reader is not misled. Reports outlive their authors. A phrase like "sharp transients over the right temporal region" in a description, with no comment in the impression, will be found by a clinician years later and read as a discharge. The structure taught in Module 2 gives the rule: the description says what was seen, in standard terms, and the impression says whether it was normal.

Three acceptable ways to handle a variant

What is not acceptable is to hedge. "Sharp waves of uncertain significance" transfers the reader's uncertainty to the patient's chart and invites treatment. If the transient failed the criteria, say that the record contains no epileptiform discharges. If it truly met the criteria, describe it as epileptiform and let the clinical correlation carry the base rate.

A routine for the drowsy record

When a record is full of sharply contoured activity, slow down rather than speed up. Confirm the state from the polygraphic channels and the background. Look at the same activity in a referential montage to check for a field. Score any candidate transient against the six criteria and write the score in your notes, even if it does not go in the report. Then ask whether the finding, if real, would change anything the clinician does. That question, which Module 9 and Module 10 return to, is the one an EEG report exists to answer.