02 A systematic read and the report

An EEG is ordered to answer a question, and the first job of the reader is to know which questions a routine study can answer. It can show the state of the background, which reflects diffuse cortical function at the moment of recording. It can show interictal epileptiform discharges, which raise or lower the probability of an epilepsy and point to a syndrome. It can capture an event, if one happens during the recording, and settle whether it had an electrographic correlate. It can grade the severity of an encephalopathy and follow it over time.

What it cannot do is decide whether a child has epilepsy by itself. A normal routine EEG does not exclude epilepsy. A routine study samples a short window, usually with the child awake for much of it, and many discharges appear only in drowsiness or sleep, or only on some days. The converse is also true: an interictal discharge does not by itself make the diagnosis. Epileptiform discharges occur in a small proportion of children who have never had a seizure, and sharply contoured normal activity is often over-read as epileptiform. Module 5 gives the numbers and the criteria. Epilepsy is a clinical diagnosis, and the EEG supports or weakens it.

Framing the indication

The indication shapes the read. A child referred after a first unprovoked seizure needs a careful search for a syndrome-defining pattern, because that changes prognosis and treatment. A child referred for staring spells needs adequate hyperventilation, because the absence of spike-wave after a good effort argues against childhood absence epilepsy. A child referred with regression needs sleep, because some patterns are marked only in sleep. When the indication and the study do not match, say so in the report and recommend the study that would answer the question. A report that interprets an inadequate study as if it were adequate misleads everyone downstream.

The read begins before the first page. Four facts set the expectations for everything that follows, and each belongs in the technical description at the top of the report.

Age. A record is normal or abnormal only for age. For neonates, state the postmenstrual age, not the days of life, because a pattern normal at 30 weeks is abnormal at term; Module 6 makes this the whole lesson. For older children, the expected posterior rhythm, sleep features and responses to activation all change with age, as Module 4 sets out.

Indication. The clinical question decides what you look for and what an adequate study needs to contain, as the previous section argued.

Medications and sedation. Drugs change the record, and a reader who does not know what the child received will misread what they see. Sedatives add beta activity and can alter sleep, so a record full of fast activity in a sedated child is not a finding. Antiseizure medications can suppress discharges and slow the background, so a normal record on treatment is a weaker negative than one off treatment, and mild diffuse slowing in a treated child may be the drug. State every relevant medication and any sedation given for the study.

State. Say what state the child was in during the recording: awake, drowsy, asleep, or all three, and whether the child was crying, feeding or being held. The state determines which features are expected and which are abnormal. A pattern that is normal in drowsiness, such as hypnagogic hypersynchrony, is exactly the kind of activity that gets over-read when the state is not noticed first.

A reader who has written these four lines has already protected the report against the commonest errors: reading a neonate against the wrong age, calling drug effect an abnormality, and calling a state-dependent normal pattern epileptiform.

Read every record in the same order, and describe each element before moving to the next. The order below is the one the report will follow, so reading in it means the report writes itself.

Background. Start with the posterior dominant rhythm: its frequency and whether it is reactive, blocking with eye opening. Then judge symmetry between the hemispheres, continuity (relevant in neonates and in encephalopathy), amplitude, and the anterior-to-posterior gradient, meaning that faster, lower-voltage activity should dominate anteriorly and the slower dominant rhythm posteriorly. A background that is slow for age, asymmetric, discontinuous, or lacking its gradient is abnormal before any discharge has been considered, and diffuse background changes often matter more for the child than a single sharp wave.

Sleep. Say whether sleep was reached. If it was, describe the sleep features expected for age, such as spindles and vertex waves, and whether they are present and symmetric. Absent or asymmetric sleep features are findings in their own right. If sleep was not reached, say so, because the record is then a weaker negative for discharges that appear only in sleep.

Activation. Say whether hyperventilation and photic stimulation were done, and whether they were done adequately. Adequacy is not a formality. In an untreated child, 3 minutes of well-performed hyperventilation without 2.5 to 4 Hz spike-wave argues against childhood absence epilepsy (Hirsch E 2022). That negative finding is only as strong as the effort: a child who "attempted" hyperventilation for a few breaths has not tested the hypothesis, and the report should say the effort was inadequate and suggest repeating it. Marked slowing during hyperventilation is a normal response in children and should not be called abnormal.

Reading in this order is what stops the eye from jumping to the one abnormal-looking wave. By the time you reach the abnormalities, you already know the age, the state, the background and the sleep, and the wave is read in context.

Once the background, sleep and activation are described, turn to the abnormalities. Each one is described in four dimensions before anything is said about what it means.

Use the shared vocabulary

Standard terms exist so that a report written in one hospital means the same thing in another. The IFCN glossary defines the terms most commonly used by electroencephalographers and proposes a report format (Kane 2017); it is free in full text and worth reading once from beginning to end. For rhythmic and periodic patterns in critically ill patients, the ACNS critical care terminology supplies the main terms and modifiers that Module 11 teaches. ACNS Guideline 7 sets out how a report should be structured (Tatum 2016). SCORE is a structured, computer-based alternative that turns the report into coded fields (Beniczky 2017); whether or not your laboratory uses it, its structure is a good model of what a complete description contains.

Description is not interpretation. "Sharp waves at T3, occasional, in drowsiness and sleep, with a field to F7 and T5" is a description. It contains no diagnosis, and a reader who disagrees with your interpretation can still use it. Keep the words that carry clinical meaning, such as "epileptiform" and "consistent with", for the impression and the correlation, and use the description to record what was seen.

The report ends with two short sections that do different jobs, and the discipline of this module is to keep them separate.

The impression states whether the record is normal or abnormal, then lists the findings as a numbered list. Each item is a finding, not a diagnosis: "1. Occasional sharp waves, maximum at T3, in drowsiness and sleep. 2. Normal background and sleep architecture for age." A numbered list makes the report scannable and lets a clinician see at a glance how many findings there are and which is first.

The clinical correlation says what the findings mean for this child, and it is stated as support, not diagnosis. "The findings support a diagnosis of focal epilepsy" is a correlation. "The patient has focal epilepsy" is a diagnosis, and it is not the electroencephalographer's to make from the tracing alone. The correlation is also where the study's limits are stated: sleep not obtained, hyperventilation inadequate, a normal record on treatment. Those limits tell the referring clinician whether the question was actually answered.

Why the separation matters

When description and correlation are mixed, the reader cannot tell what was seen from what was concluded. "Sharp waves at T3" belongs in the description; "consistent with a focal epilepsy" belongs in the correlation. Keeping them apart means a colleague who disagrees with the conclusion can still use the observation, and it keeps the reader honest: a finding that is hard to describe in standard terms is often one that should not be interpreted at all.

The sign-off for this module is five full reports that a faculty reader accepts with minor edits. The commonest edits a faculty reader makes are moving interpretive language out of the description, adding the state to a described abnormality, and adding a sentence about adequacy of activation. Write the first draft with those three in mind and the edits will be few.