Version August 2025
INTRODUCTION —
The diagnosis of epilepsy is often challenging, and misdiagnosis is not rare [1]. A detailed and reliable account of the event (ictus) by an eyewitness is the most important part of the diagnostic evaluation but may not be available [2]. Even when present, the clinical history alone may be insufficient to clinch a diagnosis of epilepsy. This is especially true following a first-ever seizure, as the lifetime incidence of having a seizure approaches 10 percent, but less than half of these patients will actually have epilepsy [3,4]. For this reason, the International League Against Epilepsy (ILAE) incorporates imaging and electrodiagnostic information together with clinical presentation in the functional definition of epilepsy [5].
This topic discusses the use of scalp EEG in the diagnosis of seizures and epilepsy. The use of other diagnostic tests in the evaluation of patients with seizures and epilepsy is presented separately:
●(See "Evaluation and management of the first seizure in adults".)
●(See "Video and ambulatory EEG monitoring in the diagnosis of seizures and epilepsy".)
CLINICAL UTILITY —
To make the best clinical use of EEG in the evaluation of patients with possible epilepsy, the clinician must understand the strengths and weaknesses of EEG, specifically as they relate to the diagnosis of seizures and epilepsy.
●Advantages – EEG is an important diagnostic test in evaluating a patient with possible seizures and/or epilepsy. It can confirm epileptic etiology of spells of interest and can provide support for the diagnosis of epilepsy. It may also assist in classifying the underlying epileptic syndrome [6,7].
●Limitations – There are several reasons why EEG alone cannot be used to make or refute a specific diagnosis of epilepsy:
•Intermittent EEG changes, including interictal epileptiform discharges (IEDs), can be infrequent and may not appear during the relatively brief period of routine EEG recording. (See 'Normal EEG' below.)
•The EEG can be abnormal in some persons with no other evidence of disease. (See 'Abnormal EEG' below.)
•Not all cases of brain disease are associated with an EEG abnormality, particularly if the pathology is small, chronic, or located deep in the brain.
•Benign or normal EEG patterns can be misinterpreted as epileptiform. (See 'Benign and normal EEG patterns' below.)
•Most EEG patterns can be caused by a wide variety of different neurologic or systemic diseases.
•Many diseases can cause more than one type of EEG pattern.
ROUTINE EEG TECHNIQUE
Electrode placement and settings — During a routine electroencephalography (EEG), electrical activity is recorded from standard sites on the scalp according to the international 10 to 20 system of electrode placement, as illustrated in the figure (figure 1). The numbers “10” and “20” refer to the percentage of total distance (front to back or right to left) of the skull between adjacent electrodes, with odd numbers being on the left and even on the right [8]. Electrodes are also placed near the eyes (typically left lower canthus and right upper canthus) along with an electrocardiogram (ECG) and, if desired, a pulse oximeter.
Standard settings for display of the EEG include a time base of 30 mm/sec and sensitivity of 7 microvolts/mm. Filters are also typically applied to filter out low (high-pass) or high (low-pass) frequencies in which artifactual activity may be more likely found. These are commonly set to 1 and 70, respectively. Depending on the local power supply, a notch filter may also be applied to selectively filter out ambient electrical noise at 60 or 50 Hz.
Channels and montages — The EEG recording depends upon differential amplification: the output is always expressed as the difference between two inputs, in a tracing that is called a channel. Channels are then clustered into chains representing different head regions. Typically, 21 or more channels are displayed in a montage, which forms the basis of EEG interpretation.
Montages help electroencephalographers view electric potential fields in different ways, similar to the different views used in radiography (eg, axial, sagittal, coronal). Montages are defined by the manner of combining channel inputs and displaying them on the reading screen. With modern digital EEG technology, the electroencephalographer has virtually infinite ability to adjust the montages and other technical parameters in each recording to optimize interpretation and analysis.
There are two main types of montages, bipolar and referential.
●Bipolar montage – In the bipolar montage, each channel consists of a comparison of two adjacent electrodes. Bipolar montages use phase reversals to localize electric field maxima, with the more common negative phase reversals recorded on scalp EEG having neighboring channels pointing toward one another. There are several types of bipolar montages, including:
•Anterior-posterior longitudinal bipolar – The channels are arranged anteriorly to posteriorly within chains. Chains include temporal, parasagittal, and midline. This is the most versatile and commonly used montage for interpreting routine EEG recordings. It is also called a “double banana” montage by some readers. Some institutions may alternate left and right by chain, while others arrange hemispheres one over another.
•Transverse bipolar – Channels are arranged from left to right and organized into chains that cover the head anteriorly to posteriorly. This is also called a "coronal" montage.
•Others, such as hatband, mandibular notch montage, and institution-specific bipolar montages – These montages are in much less frequent use but may be chosen to display particular patterns.
●Referential montage – In the referential montage, each channel consists of the difference between a specific electrode and a chosen reference. Referential montages use amplitude to determine the location of electric field maxima. There are several types of references that can be used:
•Common reference – Each electrode is compared with a common electrode, such as Cz, ipsilateral or contralateral ear or mastoid, or a noncephalic reference (eg, the rarely used "balanced neck-chest" reference).
•Common average reference – Each electrode is compared with a weighted average of the signal from all other head electrodes. Generally, electrode positions most susceptible to artifact (Fp1, Fp2, O1, O2) are excluded from the average.
•Laplacian and weighted average references – Each electrode is compared with a reference, consisting of an average of the closest electrodes (Laplacian) or an average in which different electrodes are given different weights (weighted average). These montages are nonintuitive and require more extensive knowledge of EEG technology to avoid misinterpretation, so they are not in common usage.
No montage is perfect at detecting all types of abnormalities, and every montage is susceptible to artifact or representing the electrical field in a confusing manner. For this reason, electroencephalographers are advised to periodically switch montages during interpretation of every recording and to review any potential abnormality with more than one montage. However, a discussion of the strengths and weaknesses of particular montages is beyond the scope of this topic.
Routine activating techniques — A standard routine EEG usually includes hyperventilation and photic stimulation.
Hyperventilation — Hyperventilation increases the rate of generalized discharges in childhood absence epilepsy and other generalized epilepsies [9]. It is less productive in focal epilepsies, increasing the yield of focal interictal epileptiform discharges (IEDs) by less than 10 percent [10,11]. Two studies have suggested that the yield of hyperventilation in generalized epilepsy may also be low, approximately 12 percent [12,13]. However, hyperventilation is very effort dependent, and yield may vary as a result. One study in 80 patients undergoing long-term EEG monitoring found that hyperventilation had an activating effect on EEG recording, but only in those patients whose antiseizure medications were being tapered [14].
Hyperventilation with good effort also typically produces generalized slowing. This is hypothesized to be due to the vasoconstrictive reflex of cerebral blood vessel in response to hypocarbia induced by hyperventilation. Persistent slowing lasting greater than one minute after cessation of hyperventilation has many potential causes, classically hypoglycemia or unintended continued hyperventilation. In addition to generalized slowing, hyperventilation may also augment focal slowing.
Photic stimulation — Photic stimulation induces IEDs in some individuals with idiopathic generalized epilepsy and infrequently in patients with focal seizures arising from the occipital lobe [9,10]. A photoparoxysmal response can also be a familial trait [15] and, in this setting in particular, is a less specific finding for epilepsy than spontaneous IEDs. A photoparoxysmal response that is generalized and sustained (outlasting the period of photic stimulation) and occurs at a different frequency than the photic stimulation is more likely to be associated with epilepsy than when these features are absent [11].
To maximize the yield of photic stimulation, each laboratory should have a protocol that includes testing at several frequencies between 1 and 20 Hz, and patients should be tested with eyes both opened and closed at each frequency, if possible. The technologist should document the patient's clinical response to photic stimulation at each frequency. The procedures for photic stimulation vary widely, although there have been attempts to establish standardized methodology [16].
Sleep and sleep deprivation — If an EEG performed in a patient with possible epilepsy is normal and did not include sleep, a follow-up recording should attempt to capture sleep to increase the sensitivity of the study. Sleep is a neurophysiologic activator of epilepsy; 20 to 40 percent of epilepsy patients with an initial normal recording will have IEDs on a subsequent recording that includes sleep [17-19].
Sleep is sometimes captured on a routine EEG, but sleep deprivation increases the likelihood of recording sleep. Sleep can also be induced with a non-benzodiazepine or non-barbiturate sedative medication or acquired by an overnight study, with the latter more commonly performed.
In general, we prefer overnight EEG studies because sleep deprivation can be quite disruptive and carries some risk of seizure exacerbation. However, the choice of test should be individualized to the patient’s circumstance and test availability:
●Sleep-deprived EEG – An EEG performed after full or partial sleep deprivation likely increases the diagnostic yield of the study, though with a modest effect size. A 2024 meta-analysis of 12 studies comparing EEG in sleep-deprived and non-sleep-deprived groups found that sleep-deprived EEG increased the sensitivity for a diagnosis of epilepsy, but heterogeneity was high, and the finding was not statistically significant (odds ratio [OR] 1.51, 95% CI 0.74-3.67) [20]. With exclusion of outliers, sleep-deprived EEG had a favorable effect for the diagnosis of epilepsy, and confidence intervals were narrowed, although the effect just missed statistical significance (OR 1.56, 95% CI 0.99-2.48). Clinicians can order full or partial sleep deprivation, but it is unclear how much this affects the yield [21].
Sleep deprivation appears to increase IEDs to an extent not fully explained by the greater chance of recording sleep [21,22]. One study found that the additional yield of a sleep-deprived EEG was similar whether or not sleep was recorded [18]
However, sleep deprivation can be quite disruptive and carries some risk of seizure exacerbation.
●24-hour ambulatory EEG monitoring – 24-hour studies increase the yield of the EEG because they include sleep and are of a longer duration. When sleep-deprived EEG was compared with 24-hour ambulatory EEG monitoring in 46 patients with "presumed epilepsy," IED detection was similar (24 versus 33 percent) [23]. However, clinical seizures were also captured in 15 percent of the ambulatory EEGs and none of the sleep-deprived EEGs. (See "Video and ambulatory EEG monitoring in the diagnosis of seizures and epilepsy".)
●Drug-induced sleep – Sleep can be induced by administering a non-benzodiazepine or non-barbiturate sedative, but this is infrequently used in clinical practice since the introduction of ambulatory EEG. Drug-induced sleep EEG studies have lower rates of IEDs and seizures compared with sleep-deprived EEG [24,25]. The time it takes to perform an EEG with chloral hydrate sedation is significantly longer than for an EEG without sedation.
EEG BACKGROUND
Amplitude and frequency — The electrical activity in each electroencephalography (EEG) channel can be described in terms of amplitude and frequency. The amplitude of typical EEG recordings ranges from 5 to 200 microvolts, but most awake background EEG recordings are in the range of 20 to 50 microvolts. Frequency of EEG activity is expressed according to the following terminology:
●Delta – 0 to 4 Hz
●Theta – 4 to 8 Hz
●Alpha – 8 to 13 Hz
●Beta – 13 to 30 Hz
●Gamma – Greater than 30 Hz
Awake background — All EEG reports should describe the awake background. EEG procedure reports should comment on the continuity, symmetry, and reactivity of the background rhythms.
●Continuity – Continuity exists along a continuum from [26]:
•Continuous activity
•Nearly continuous (1 to 9 percent attenuated or suppression)
•Discontinuous (10 to 49 percent attenuated or suppressed)
•Burst attenuation or burst suppression (50 to 99 percent attenuated or suppressed)
•Diffusely attenuated or suppressed (100 percent)
Attenuation is defined by periods of lower voltage that are ≥10 microvolts but <50 percent of the higher voltage background. Suppression is defined by periods of lower voltage that are <10 microvolts [26].
●Asymmetry – Asymmetry may be due to hemispheric or regionalized slowing (frequency) or attenuation of faster frequencies (amplitude). Slowing suggests dysfunction of the subcortical circuity, whereas attenuation suggests cortical dysfunction and/or an overlying fluid collection (eg, subdural hemorrhage). In patients with epilepsy or a prior seizure, a regional attenuation of faster frequencies can be the result of a postictal cortical dysfunction, and the attenuation is often concordant with the side of seizure onset.
●Reactivity – EEG reactivity is assessed by changes in the EEG background frequencies or power in response to a stimulus. Modulation of the posterior-dominant rhythm (PDR) in response to eye closure is one example of reactivity of the EEG.
Posterior-dominant (alpha) rhythm — In the normal awake adult with eyes closed, there is an 8.5 to 12 Hz alpha rhythm, maximal in the posterior part of the head. This is also referred to as the posterior-dominant rhythm (PDR). The amplitude of the alpha falls off anteriorly where there is lower-voltage beta activity.
In some patients (eg, children, patients with mild cerebral dysfunction), the frequency of the "alpha" rhythm may be in the theta range, which introduces semantic confusion. Some electroencephalographers therefore prefer to use the term "posterior-dominant rhythm (PDR)" instead of "alpha rhythm."
●Variation in awake PDR – The PDR becomes lower voltage (attenuates) or disappears when the eyes open, and it becomes higher voltage (augments) when the eyes close. In this way, the alpha rhythm may be considered the "idling" rhythm of relaxed visual cortex.
In contrast to a resting pattern, lambda waves may be observed over the occipital head regions when the patient is scanning or processing complex visual information. Lambda waves typically appear as electropositive sharp transients maximal at O1 or O2, but they are more symmetric in their up- and downstrokes than epileptiform discharges and are known to be physiologic waveforms.
●Mu rhythm – A similar "idling" rhythm known as the mu rhythm is seen over Rolandic (pericentral) cortex. Mu is a rhythmic, 7 to 13 Hz, sharply contoured pattern that may have a comb-like appearance and may be observed over the central head regions when the contralateral limbs are at rest. This rhythm attenuates with contralateral movement or even with thinking about movement.
●Effect of drowsiness and sleep – With drowsiness (stage N1 sleep), the PDR gradually disappears, fronto-central beta activity may become more prominent, and diffuse theta activity emerges. Stage N1 sleep also sees emergence of vertex waves and positive occipital sharp transients of sleep (POSTS). These persist into stage N2 sleep, in which there is also emergence of sleep spindles and K complexes. The next stage is N3 or slow-wave sleep, characterized by high-amplitude delta activity. Rapid eye movement (REM) sleep is characterized by return of lower voltage activity and alpha frequency waveforms with rapid eye movements and loss of muscle tone.
The stages and architecture of normal sleep are discussed in more detail elsewhere. (See "Stages and architecture of normal sleep", section on 'Sleep staging'.)
ABNORMAL EEG FINDINGS —
Different abnormal electroencephalography (EEG) findings are variably associated with epilepsy. In the work-up of epilepsy, it is useful to classify EEG abnormalities as epileptiform or nonepileptiform.
Epileptiform activity — Examples of epileptiform activity include:
●Sporadic interictal epileptiform discharges (IEDs). (See 'Sporadic IEDs' below.)
●Periodic and rhythmic patterns that are highly epileptiform (ie, highly associated with seizures) include lateralized periodic discharges (LPDs) (waveform 1); generalized periodic discharges (GPDs) (waveform 2); and lateralized rhythmic delta activity (LRDA). (See 'Periodic discharges' below.)
Generalized rhythmic delta activity (GRDA) does not connote an increased risk of seizures and is therefore not considered epileptiform [27].
Sporadic IEDs
Definition — Sporadic interictal epileptiform discharges (IEDs) consist of sharp waves and spikes [28]:
●Sharp waves are epileptiform discharges defined as having a duration of 70 to 200 msec
●Spikes are similar to sharp waves but are briefer than 70 msec
There is no diagnostic or prognostic distinction between sharp waves and spikes [6]. A slow wave typically follows a sharp wave or spike and is usually higher in amplitude than the preceding sharp wave or spike.
IEDs may be focal or generalized, noting that bilateral synchronous (but not necessarily diffuse) potentials are classified as generalized.
To qualify as an IED, discharges should meet at least four of the following six criteria proposed by the International Federation of Clinical Neurophysiology (IFCN) [10,29]:
●Di- or triphasic wave with pointed peak.
●A different wave duration than the ongoing background activity.
●Asymmetry of the waveform.
●Followed by a slow after-wave.
●The background activity is disrupted by the presence of the IED.
●A voltage map with distribution of the negative and positive potentials that suggests a source in the brain corresponding to a radial, oblique, or tangential orientation of the source. In essence, a physiologically plausible potential field.
The above IFCN criteria have good reliability, with the receiver operating characteristic (ROC) curve having an area under the curve (AUC) value of 0.97 [29].
Additionally, candidate IEDs must not be one of the known benign variants or normal discharges such as wicket spikes, small sharp spikes (SSS), or vertex waves (table 1 and waveform 3A-G).
Sensitivity — An IED is found in 20 to 55 percent of persons with epilepsy on a first "routine" EEG [21,30-33]. Several factors can influence the sensitivity of finding an IED and, in turn, can influence the diagnosis of epilepsy.
●The number of EEG studies – With repeated recordings, the likelihood of finding IEDs increases from 20 to 50 percent to as high as 80 to 90 percent when four or more EEGs are obtained [30,31,33-36].
●EEG duration – A routine EEG is typically 30 to 60 minutes long. Longer EEG monitoring increases the yield of the study [37-41]. At some centers, ambulatory studies of 24 to 72 hours’ duration may be used when longer-term recording is desired than may be obtained with "prolonged" recording in the outpatient lab. Inpatient video-EEG monitoring for up to several days is most commonly used to capture events to determine whether they are epileptic; it is also used in situations where there is suspicion for nonconvulsive status epilepticus. (See "Video and ambulatory EEG monitoring in the diagnosis of seizures and epilepsy" and "Nonconvulsive status epilepticus: Classification, clinical features, and diagnosis", section on 'Routine and continuous EEG'.)
When monitoring for nonconvulsive seizures, the 2HELPS2B score can help determine length of required monitoring [42]. (See "Nonconvulsive status epilepticus: Classification, clinical features, and diagnosis", section on 'Assessing risk of seizure'.)
●Seizure frequency – More frequent seizures are associated with a higher prevalence of IEDs on an EEG tracing [31,43]. In this regard, a finding of IEDs may also be predictive of seizure recurrence. In a meta-analysis of 11 studies of EEG after a first unprovoked seizure in adults, the probability of seizure recurrence in patients with epileptiform EEG abnormalities was 50 percent compared with 27 percent in those with normal EEGs [5,44]. (See "Evaluation and management of the first seizure in adults".)
●Timing of EEG after seizure – Clinical seizures are temporally associated with IED frequency [30,31,45]. In a study of more than 600 outpatients with epilepsy or first seizure, the chance of finding epileptiform discharges on EEG decreased as time elapsed since the last seizure: 62 percent if ≤24 hours, 51 percent if 25 to 48 hours, 40 percent if 49 to 72 hours, and 31 percent if >72 hours [46]. In another case series, early performance of an EEG (within 24 hours of a seizure) appeared to have a similar yield of epileptiform abnormalities as did a later-performed sleep-deprived study [47]. (See 'Sleep and sleep deprivation' above.)
●Antiseizure medication therapy – There is limited information regarding the suppressant effect of antiseizure medications on IED detection [10,48-51]. Treatment with valproate, levetiracetam, and probably ethosuximide reduces the rate of generalized IEDs. Diazepam and phenobarbital can suppress IEDs acutely, but chronic therapy may have little impact. Another study found that antiseizure medication withdrawal was associated with fewer spikes on EEG [52].
●Epilepsy syndrome – The EEG is more likely to be abnormal in certain epilepsy syndromes. IEDs are almost invariably present in children with untreated infantile spasms, Landau-Kleffner syndrome, and self-limited epilepsy with centrotemporal spikes (SeLECTS), formerly known as benign Rolandic epilepsy. While medial temporal lobe epilepsy is usually associated with an abnormal interictal EEG, patients with frontal lobe epilepsy may have a normal interictal EEG [10,53]. However, the presence of IEDs often influences both the diagnosis of epilepsy itself as well as the specific epilepsy syndrome; as a result, this may be a source of substantial bias.
●Specialized techniques – Hyperventilation and photic stimulation are routinely done to increase the likelihood of capturing IEDs. If greater sensitivity is desired after an initial negative test, repeat testing may also utilize sleep deprivation or special electrode placement. In the EMU, more significant sleep deprivation and antiseizure medication withdrawal can also take place in a monitored setting. (See 'Specialized techniques' below.)
●Age – Factors that are variably reported to be associated with the prevalence of IEDs in persons with epilepsy include a younger age at the time of EEG, a longer duration of epilepsy, and an earlier age at epilepsy onset [31,43].
Specificity — IEDs are rare in patients without a history of seizures. In a 2024 systematic review and meta-analysis of 53 studies with over 73,000 individuals without a history of seizures, the overall prevalence of IEDs was 1.74 percent (95% CI 1.13-2.67) [54]. The prevalence of IEDs was somewhat higher in children 1 to 17 years of age (2.45 percent, 95% CI 1.41-4.21) and adults >65 years of age (5.96 percent, 95% CI 1.39-22.13) and lower in adults 18 to 64 years of age (0.93 percent, 95% CI 0.48-1.80).
A finding of IEDs is most helpful if the clinical history strongly suggests epileptic seizure [2,6,46]. However, certain caveats apply:
●The patient’s age and pattern of IEDs impact the specificity of these findings. Spikes and sharp waves are common in normal neonates during quiet (non-REM) sleep but disappear over the first six to eight weeks of life. By contrast, IEDs in adults are almost always associated with epilepsy [22,23,31,55,56]. In children, IEDs are less specific for epilepsy. In particular, central-midtemporal discharges, generalized spike-wave discharges, and photoparoxysmal responses may be asymptomatic manifestations of genetic traits [57-60]. In one series of EEG studies, only 40 percent of children with central-midtemporal spikes had epileptic seizures [23,61].
The table (table 2) lists IEDs that may be seen on EEG, their associated clinical significance, and their likelihood of association with epilepsy [10].
●Some conditions are associated with IEDs on EEG, but do not imply epilepsy. These include occipital spikes seen in blind people (especially those who are congenitally blind) [62].
●Withdrawal from short-acting barbiturates and benzodiazepines, certain metabolic derangements (eg, hypocalcemia, uremia, dialysis disequilibrium), as well as high drug levels of lithium, baclofen, neuroleptics (especially clozapine), bupropion, and tricyclic antidepressants have been associated with IEDs even in the absence of accompanying seizures [50,63,64]. These conditions are also associated with a lower seizure threshold.
●Inexperienced interpreters of EEG, unaware of benign variants and normal waveforms commonly seen on EEG, can misread EEGs and limit the specificity of the findings. (See 'Pitfalls in interpretation' below.)
Periodic discharges
●Lateralized periodic discharges (LPDs) – LPDs, previously known as periodic lateralized epileptiform discharges (PLEDs), are defined by lateralized, persistent spikes, sharp waves, or sharply contoured or blunt slow waves that occur at nearly regular intervals (varying by <50 percent from one cycle to the next in most cycle pairs), for at least six cycles (waveform 1) [26]. If the frequency of these LPDs is greater than 2.5 Hz over 10 seconds or if there is evolution (in frequency, morphology, or location), these are defined as an ictal pattern (electrographic seizure). When these are present at a frequency of 1.0 to 2.5 Hz for longer than 10 seconds, they fall along the ictal-interictal continuum (IIC) [26].
LPDs are most often seen in the setting of acute, relatively large, destructive lesions, such as cerebral infarction or hemorrhage, encephalitis, abscess, or rapidly growing cerebral malignancy [65-70]. In children, LPDs are also associated with chronic diffuse encephalopathies [71].
LPDs are highly associated with seizures, especially nonconvulsive seizures in critically ill patients [27]. Clinical seizures have also been reported in the majority of patients with LPDs (50 to 100 percent, depending on the population) [67,70,72-74]. Focal motor seizures are the most common clinical seizure type associated with LPDs [72,73].
LPDs usually resolve over several days to weeks with recovery from the acute illness. However, their presence increases the risk of developing remote symptomatic epilepsy, with that risk being higher if there is any superimposed rhythmic (LPD+R) or fast (LPD+F) activity [72,73,75,76].In a study of 118 patients with LPDs identified during a critical care admission, 47 percent had late seizures (after hospital discharge) [75].
●Bilateral independent periodic discharges (BIPDs) – BIPDs, previously known as BIPLEDs, are most often observed in association with acute central nervous system (CNS) infections (especially herpes simplex encephalitis), anoxic encephalopathy, and severe chronic epilepsy [67,77]. This pattern is also highly associated with seizures. Compared with LPDs, BIPDs are associated with more severe cerebral injury, worse neurologic status, and higher mortality, probably related to the severity of the underlying illness.
●Generalized periodic discharges (GPDs) – GPDs (waveform 2) are often observed in critically ill patients [75,78,79]. They are associated with seizures, even more so when seen in combination with superimposed rhythmic delta activity (GPD+R) or fast (GPD+F) activity [27]. (See "Nonconvulsive status epilepticus: Classification, clinical features, and diagnosis", section on 'Uncertain EEG patterns in critical illness'.)
Nonepileptiform activity — Examples of non-epileptiform abnormalities include:
●Slowing, which may be diffuse, regional, or localized (see 'Focal slowing' below and 'Generalized slowing' below)
●Amplitude attenuation, which may again be diffuse or focal
●Other deviations from normal patterns
While these findings are not considered epileptiform, they may be seen in patients with epilepsy, especially around the area of an epileptogenic focus (even if intermittent) and as a postictal phenomenon after seizures.
Such findings are nonspecific and are relatively common, especially in older individuals, patients with migraine, and those on centrally acting medications. These should not be interpreted as supporting a diagnosis of epilepsy – or even necessarily of a primary neurologic disorder.
Focal slowing — Focal delta slowing is a common postictal and interictal finding in patients with focal seizures. However, it is also frequently seen in other neurologic disorders, especially focal structural lesions, regardless of whether there are associated seizures [80-82]. In one case series, two-thirds of 100 patients with continuous, focal, polymorphic (nonrhythmic) delta activity had a structural lesion, but seizures occurred in only approximately 20 percent [80].
In patients with no clinical history of neurologic injury and a normal neuroimaging study, rhythmic as opposed to nonrhythmic focal slowing is more likely to suggest epilepsy:
●Lateralized rhythmic delta activity (LRDA) – LRDA (waveform 4) is a form of focal rhythmic slowing. The two most common causes of LRDA are acute brain injury and epilepsy. In one study, approximately 5 percent of critically ill patients had LRDA [83]. In this study, two-thirds of patients with LRDA had either clinical or electrographic seizures during their acute illness, an identical frequency to patients with LPDs in the same study. This suggests that the clinical significance of LRDA is similar to that of LPDs and much different than focal nonrhythmic slowing, in which much lower rates of seizures are observed. LRDA and LPDs commonly coexist, and patients with both patterns are at increased risk of acute seizures [27,83]. Superimposed spike/sharp activity may also be seen with LRDA, constituting LRDA+S and connoting an even greater risk for seizures [27]. (See 'Periodic discharges' above.)
●Temporal intermittent rhythmic delta activity (TIRDA) – TIRDA (waveform 5) is a particular form of focal slowing (and a subtype of LRDA) that is specific for temporal lobe localization in patients with refractory epilepsy. TIRDA is observed in as many as 25 to 40 percent of patients evaluated for temporal lobe resection [84,85]. TIRDA is often associated with temporal IEDs and has a high positive predictive value for temporal lobe localization in patients with refractory epilepsy. (See "Focal epilepsy: Causes and clinical features", section on 'Mesial temporal lobe epilepsy'.)
●Occipital intermittent rhythmic delta activity (OIRDA) – OIRDA is more common in young children and is rarely seen in patients older than 15 years of age [86]. It is a frequent interictal finding in generalized epilepsy syndromes, occurring in 15 to 38 percent of all patients with childhood absence epilepsy, and implies a good prognosis [86-89].
Generalized slowing — Generalized slowing of the background rhythms, including loss of the PDR or reactivity generally, may be seen in generalized epilepsies but is less specific for epilepsy than is focal slowing. Generalized slowing is not specific to primary CNS pathology and may be seen in toxic-metabolic encephalopathy, systemic infections, or with certain medications.
●Generalized rhythmic delta activity (GRDA) – GRDA is bilateral, synchronous, and symmetric slow activity in the delta frequency range (<4 Hz).
GRDA does not confer an increased risk of seizures and is therefore not considered a marker of epilepsy [27]. A large, multicenter study of 1513 critically ill patients with periodic or rhythmic activity found that GRDA was not associated with an increased risk of seizures, even at higher frequencies (>2 Hz) [27].
GRDA is most commonly intermittent in adults and usually has an anterior predominance. It has therefore also been called frontal intermittent rhythmic delta activity (FIRDA). Once thought to be associated primarily with deep midline cerebral lesions and raised intracranial pressure, frontally predominant GRDA is now recognized to be a nonspecific marker of encephalopathy regardless of etiology, as well as neurodegenerative disease and generalized epilepsy. It can occasionally be seen in normal individuals as well [90].
Ictal and possibly ictal EEG patterns
●Electrographic seizures – Electrographic seizures are any EEG pattern that either evolves in frequency, morphology, or location over 10 seconds, or consists of any periodic (PD) or spike-wave (SW) discharges at ≥2.5 Hz over 10 seconds [26]. Not all electroclinical seizures qualify as an electrographic seizure by EEG criteria. Any definite clinical change that correlates with a time-locked EEG change is an electroclinical seizure. Classic examples of this include myoclonic jerks time-locked with generalized spikes (myoclonic seizures) or spasms correlating with a broad slow wave and decrement (epileptic spasms).
●Brief potentially ictal rhythmic discharges (BIRDs) – BIRDs may be conveniently thought of as fragments of electrographic seizures. BIRDs are characterized by generalized or focal rhythmic activity at >4 Hz of duration 0.5 to 10 seconds with at least six cycles [26]. Possible BIRDs must be sharply contoured. Definite BIRDs must either have evolution or appear similar morphologically to a patient’s known ictal or interictal epileptiform activity.
●Ictal-interictal continuum (IIC) – The IIC describes electrographic patterns which lie along a spectrum between being interictal epileptiform activity (eg, sporadic sharp waves) versus definite electrographic seizures [26]. It is an area of active study to determine how much tissue injury these various patterns have as well as when and how aggressively they need to be treated.
The following electrographic patterns lie along the IIC:
•Any PD or SW pattern that averages >1 Hz to ≤2.5 Hz over 10 seconds
•Any PD or SW pattern that averages ≥0.5 to ≤1 Hz with a plus (+) modifier (ie, an additional feature which renders the pattern more ictal-appearing) or fluctuation (ie, pattern variability in frequency, location, or morphology over time that does not qualifying as a progressive change or “evolution”)
•LRDA averaging >1 Hz over 10 seconds with a plus modifier or fluctuation
GRDA is not considered to be an IIC pattern under any circumstance since it is not associated with risk for seizures [27]. Any definite evolution in frequency, morphology, or location qualifies this pattern as an electrographic seizure rather than IIC.
Epilepsy syndrome diagnosis — Specific interictal EEG findings that are associated with specific epilepsy syndromes are listed in the table (table 3).
Clinical correlation between the clinical seizure type and EEG findings is important. Agreement between these is generally sufficient to distinguish generalized from focal epilepsies [91]. In addition, when focal IEDs are strongly lateralized (more than 90 to 95 percent), they predict the side of seizure onset [56]. However, focal IEDs can manifest as secondary bilateral synchronous discharges due to rapid bisynchrony, while generalized epilepsy can have fragmentary expression of IEDs that appear more focal [9,92]. As a result, it is important to consider the clinical as well as the EEG manifestations when categorizing patients as having focal or generalized epilepsy.
The classification of seizures and epilepsy can be important for prognosis and treatment. In adult patients, the most important distinction is between primary generalized and focal epilepsy.
The evaluation and diagnosis of epilepsy are reviewed in detail separately. (See "Evaluation and management of the first seizure in adults" and "Seizures and epilepsy in children: Clinical and laboratory diagnosis".)
SPECIALIZED TECHNIQUES —
Specific methods can be employed to improve the detection of interictal epileptiform discharges (IEDs) and the sensitivity of the test.
Special electrode placement — Some highly epileptogenic areas, such as the mesial and inferior temporal lobes, are not well explored by the standard scalp electrodes. Scalp coverage with the standard 10 to 20 system detects only approximately 65 percent of epileptiform discharges from the temporal lobes [93]. The classic temporal chain electrodes (F7/F8 and T7/T8) lie close to the sylvian fissure and record activity from the infra- and suprasylvian regions (image 1).
Specialized electrode placement can improve IED detection (image 1). However, some of these electrodes can be uncomfortable for patients and are associated with increased artifact, which increases the potential for misinterpretation. (See 'Pitfalls in interpretation' below.)
Options include:
●Sphenoidal electrodes are wires inserted through the skin via a needle cannula inferior to the zygomatic arch, perpendicular to the sagittal plane, and parallel to the coronal plane, in an attempt to record activity from the anterior tip of the temporal lobe (image 1). These are no longer in common use.
●Nasopharyngeal electrodes (Np1 and Np2) are inserted through the nostrils into the nasopharynx to record from the anterior mesial surface of the temporal lobe (image 1). These are not in common use.
●Ear electrodes (A1 and A2) are inserted (noninvasively) into the ear canal to lie next to the tympanic membrane (image 1).
The following are additional surface electrodes:
●Superficial anterior temporal or Silverman's electrodes (T1 and T2) are placed on the skin 1 cm above and one-third the distance along the line from the external auditory meatus to the external canthus of the eye to record from the anterior-basal areas of the temporal lobe.
●Mandibular notch electrodes (Mn1 and Mn2) are placed 2.5 cm anterior to the external auditory meatus, inferior to the zygomatic arch, at the insertion site for sphenoidal electrodes (image 1).
●Inferior temporal chain electrodes (F9/T9/P9 and F10/T10/P10) are an extra chain of three electrodes placed a standard electrode distance inferior to the standard temporal chain and record from the temporal lobes (image 1).
Of these, sphenoidal electrodes have higher yield and are associated with considerably less artifact than nasopharyngeal electrodes. In one prospective study of 44 EEGs with simultaneous recording of sphenoidal, nasopharyngeal, and ear electrodes, a total of mesial temporal 875 spikes were recorded [94]; sphenoidal electrodes detected 99 percent of discharges, while nasopharyngeal electrodes detected 57 percent, and ear electrodes detected 54 percent. However, these electrodes are invasive, uncomfortable for patients, and not routinely used for diagnostic purposes.
The superficial anterior temporal and mandibular notch electrodes are slightly less sensitive than sphenoidal electrodes. However, they are noninvasive and are equivalent or superior in sensitivity and patient comfort to the nasopharyngeal, mandibular notch subdermal (also known as minisphenoidal), and ear electrodes [95-98].
Limited electrode montages — While a full-montage EEG is always preferred, a reduced electrode array can be useful when resources are limited or if there is going to be a delay in obtaining the full-montage EEG.
Full-montage EEG is considered the gold standard for identifying suspected subclinical electrographic seizures, but cost, time, and availability of trained technologists are limits to widespread utilization. In one report of urgent EEGs in an inpatient setting, a reduced-montage EEG (bitemporal chains only), using a headband device that could be applied without the need for a technologist, had a high sensitivity for detecting seizures, with detection rates very close to full-montage EEG [99]. Overall, for identifying whether a patient was having seizures during the EEG recording, the concordance between reduced-montage and full-montage EEG was 95 percent. In addition, limited montage EEG had a 96 percent positive predictive value for identifying clinically important abnormalities such as seizures, epileptiform discharges, or periodic discharges.
The same device now has artificial intelligence (AI)-based algorithms that can provide an estimate of seizure burden and are approved by the US Food and Drug Administration (FDA) for diagnosing nonconvulsive status epilepticus. In a retrospective report using expert blinded reviewers, the AI-based algorithms had an excellent predictive value for excluding nonconvulsive status epilepticus, with reasonable positive predictive values when a high seizure burden is reported [100].
The full-montage EEG should be used whenever it becomes available. (See "Nonconvulsive status epilepticus: Classification, clinical features, and diagnosis", section on 'EEG electrode array and placement'.)
Quantitative and automated EEG analysis — Several available quantitative measures can be used as adjunctive methods to identify seizures and other epileptiform abnormalities on EEG. These include seizure detection algorithms, automated detectors of epileptiform discharges, and quantitative parameters that aid in the visual detection of patterns or trends suggestive of seizures in long-term EEG recording.
While these methods, in their current form, should not replace interpretation of standard EEG patterns by a qualified reader, they have several roles. Automated seizure and epileptiform discharge detectors can serve as a screening method, especially in longer recordings, pointing the EEG reader to areas that deserve particular scrutiny and helping to ensure that subtle abnormalities are not missed [101]. These methods can also aid in quantifying spikes or seizures during longer recording periods, providing a perspective on longer-term trends. With further improvement and use of machine learning techniques, automated analysis will likely be of more clinical utility in the near future.
Intracranial electrodes — In selected cases, invasive intracranial EEG monitoring combined with video monitoring is used to localize the epileptogenic zone or to map eloquent cortex in patients who are candidates for epilepsy surgery. This is discussed elsewhere. (See "Epilepsy surgery: Presurgical evaluation", section on 'Invasive EEG monitoring'.)
Less commonly, invasive electrodes are used in critically ill patients with acute brain injury in conjunction with other invasive monitors ("multimodality monitoring") to enable advanced, individualized, physiology-driven decision-making [102].
PITFALLS IN INTERPRETATION —
Misinterpretation of EEG findings or over-reliance on the EEG frequently contributes to misdiagnosis [17,18,35,45,103].
Repeating or reinterpreting the EEG at a tertiary epilepsy center by a board-certified electroencephalographer can be helpful. One meta-analysis found that a more restrictive interpretation style that limits false-positive results improves diagnostic accuracy [104]. However, interobserver agreement is only moderate even among experienced, board-certified neurophysiologists [2,32,105].
Normal EEG — A normal electroencephalography (EEG) never rules out epilepsy [6,106]. Even with repeated EEGs, use of specialized techniques, or prolonged monitoring, a significant number of patients with epilepsy (10 to 20 percent) will not have interictal epileptiform discharges (IEDs) [19,106]. Even ictal recordings may not have an identifiable scalp correlate in many frontal lobe seizures as well as in focal aware seizures from any location, particularly with seizures from more limited cortical areas in deeper brain regions (eg, mesial temporal, insula, or cingulate cortex) [53].
Abnormal EEG — An "abnormal" EEG does not by itself define epilepsy; most abnormal findings are nonspecific. IEDs are the most specific finding for epilepsy, but these can occur in approximately 1 to 2 percent of healthy adults and approximately 2 to 7 percent of normal children [19,107]. (See 'Specificity' above.)
Benign and normal EEG patterns — There is wide variation in how EEGs are interpreted. When EEGs are read by clinicians without special training, many benign or normal patterns are often misinterpreted as epileptiform [17,18,103,108]. These include (table 1) [7,18,109,110]:
●Benign epileptiform transients of sleep (BETS), also termed small sharp spikes (SSS) (waveform 3A)
●Wicket spikes (waveform 3B)
●Rhythmic midtemporal theta of drowsiness (psychomotor variant) (waveform 3C)
●6 Hz "phantom" spike-and-wave complex (waveform 3D)
●Subclinical rhythmic EEG discharge in adults (SREDA) (waveform 3E)
●Positive occipital sharp transients of sleep (POSTS) (waveform 3F)
●Breach rhythm (waveform 3G)
●14 Hz and 6 Hz positive spikes (waveform 6)
●Hyperventilation-induced high-voltage paroxysmal slow waves
●Mu rhythm
●Lambda waves
●Repetitive vertex waves, especially in children
●Artifacts:
•Myogenic and chewing artifact
•Eye blink artifact
•Lateral eye movement and rectus spike artifact
•Glossokinetic artifact
•60 Hz (or 50 Hz) electrical artifact
•Electrode pop artifact
•ECG artifact
•Sweat artifact
SOCIETY GUIDELINE LINKS —
Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Seizures and epilepsy in adults" and "Society guideline links: Seizures and epilepsy in children".)
INFORMATION FOR PATIENTS —
UpToDate offers two types of patient education materials, "The Basics" and "Beyond the Basics." The Basics patient education pieces are written in plain language, at the 5th to 6th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-to-read materials.
Here are the patient education articles that are relevant to this topic. We encourage you to print or e-mail these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on "patient info" and the keyword(s) of interest.)
●(See "Patient education: EEG (The Basics)".)
●(See "Patient education: Epilepsy in adults (The Basics)".)
●(See "Patient education: Epilepsy in children (The Basics)".)
SUMMARY AND RECOMMENDATIONS
●Clinical utility – Electroencephalography (EEG) is an important diagnostic test in evaluating a patient with possible epilepsy, providing evidence that helps confirm or refute the diagnosis. During a routine EEG, electrical activity is recorded from different standard sites on the scalp according to the international (10 to 20) electrode placement system (figure 1).
EEG also assists in classifying the underlying epileptic syndrome and thereby guides management. (See 'Clinical utility' above.)
●Routine EEG – In the normal awake adult with eyes closed, a prominent 8.5 to 12 Hz posterior dominant (alpha) rhythm is observed in the posterior part of the head. This rhythm gradually disappears with drowsiness. The amplitude of the alpha falls off anteriorly where there is lower-voltage beta activity. (See 'Routine EEG technique' above and 'EEG background' above.)
●Epileptiform activity – Examples of epileptiform activity include sporadic interictal epileptiform discharges (IEDs; sharp waves and spikes) and periodic patterns including lateralized periodic discharges (LPDs) (waveform 1) and generalized periodic discharges (GPDs) (waveform 2). Focal rhythmic patterns including lateralized rhythmic delta activity (LRDA) are also highly associated with seizures. (See 'Epileptiform activity' above.)
•Interictal epileptiform discharges (IEDs) – IEDs include sharp waves and spikes – which may be differentiated by their duration of 20 to 70 msec for spikes versus 70 to 200 msec for sharp waves. The six International Federation of Clinical Neurophysiology (IFCN) criteria provide a useful framework for determining whether a sharp waveform qualifies as an IED. (See 'Sporadic IEDs' above.)
Overall, the specificity of IEDs for epilepsy is high: more than 90 percent in adults. However, inexperienced EEG interpreters can mistake artifact or benign EEG patterns for IEDs, lowering the specificity of the study. The specificity of this finding is also influenced by the pattern of IEDs, and by the patient's age, family history, and comorbid conditions. (See 'Specificity' above and 'Pitfalls in interpretation' above.)
•Lateralized periodic discharges (LPDs) – LPDs are usually seen in the setting of acute, relatively large cerebral injury, such as stroke, encephalitis, or rapidly growing cerebral malignancies (waveform 1).
Patients with LPDs have an increased risk of acute symptomatic seizures as well as new-onset remote symptomatic epilepsy after recovery. (See 'Periodic discharges' above.)
●EEG findings in patients with epilepsy – A single routine EEG has low sensitivity for detecting IEDs (20 to 55 percent) in patients with epilepsy. The sensitivity can be increased by repeating the study, recording for a longer time (such as overnight), including a recording of sleep (spontaneous or after sleep deprivation), performing the EEG within 24 hours of a seizure, and using special electrodes for temporal lobe epilepsy. (See 'Sensitivity' above and 'Specialized techniques' above.)
A normal EEG, however, can never rule out epilepsy, which is a clinical diagnosis; 10 to 20 percent of patients with definite epilepsy never have IEDs. (See 'Normal EEG' above.)
While some interictal EEG findings suggest a specific epilepsy syndrome (table 3), clinical correlation between the clinical seizure type and EEG findings is important. (See 'Epilepsy syndrome diagnosis' above.)
●Slowing – Generalized slowing on EEG is nonspecific and does not suggest epilepsy.
Focal slowing is also nonspecific but is a common postictal and interictal finding in patients with focal seizures. It may suggest epilepsy, particularly in patients with rhythmic slowing. As examples, LRDA is associated with seizures (waveform 4), including the specific subtype of temporal intermittent rhythmic delta activity (TIRDA), which is highly associated with temporal lobe epilepsy (waveform 5). (See 'Focal slowing' above and 'Generalized slowing' above.)
