Sensitivity is an important measure of clinical utility of a test finding. For our purposes, sensitivity of IEDs can be defined as the prevalence of IEDs in people with a diagnosis of seizures or epilepsy determined by other means. Many studies have addressed these issues and allow some general conclusions. However, many factors affect sensitivity, and these must be considered when making clinical decisions.

In three studies of large groups of mostly adult patients evaluated at epilepsy centers, the initial routine EEG demonstrated IEDs in 29% to 55% of patients.^2,50,108 However, serial EEGs performed during varying periods of time ultimately demonstrated IEDs in 80% to 90%.^2,108 Of those patients who had IEDs recorded at some point, 90% had such findings by the fourth EEG.^2,108 If the first EEG demonstrated nonspecific abnormalities but no IEDs, subsequent EEGs were more likely to demonstrate IEDs.¹⁰⁸ Another investigation¹³⁵ continuously screened adult patients undergoing video-EEG monitoring for IEDs with standard IED-detection software. Analysis was confined to patients with recorded seizures, allowing a secure diagnosis that was not influenced by the presence of IEDs. No IEDs were detected in 19% of patients, although EEG was screened continuously for an average of 6.9 days.

Studies of patients with single seizures or patients in whom antiepileptic drugs are being discontinued provide some information about the prevalence of IEDs in patients with infrequent seizures (Table 2). Such studies are less likely to be affected by selection bias. From 12% to 50% of this population have IEDs recorded during the initial routine EEG. Nonepileptiform abnormalities are noted in 4% to 45%, and 43% to 74% have normal EEGs. Only one study¹³¹ examined the yield of repeated EEGs in patients with single seizures. Twelve percent of a mostly adult group had IEDs recorded with the first EEG. Electroencephalography was repeated in almost all of those with an initially nonepileptiform study, and an additional 14% had IEDs, resulting in a total yield of 26% after two EEGs. The wide variability in the prevalence of IEDs in these groups with infrequent seizures can be related to a variety of factors, including the study of different populations at different centers and the use of different criteria for the detection of IEDs. The prevalence of normal EEGs was less variable, with five of seven studies finding that about 50% of patients had normal initial EEGs.

Interictal epileptiform discharges are recorded more frequently in children than in adults.² Furthermore, IEDs are more frequent when epilepsy begins earlier in life.² IEDS are more prevalent and more persistent in some epilepsy syndromes such as hypsarrythmia, Landau-Kleffner syndrome, untreated childhood-onset absence, and benign rolandic epilepsy. However, the interictal findings usually form a critical component of diagnosis in these cases, and so such statements may constitute circular reasoning. In adults with partial epilepsy, IEDs are more common when seizures originate in the temporal lobes than when seizures originate elsewhere.^2,135

Some antiepileptic drugs appear to affect the likelihood of recording IEDs in certain situations.^6,112 Benzodiazepines and barbiturates consistently decrease the prevalence of IEDs acutely. These effects appear to wane with long-term therapy. Withdrawal of barbiturates may be accompanied acutely by the appearance of generalized epileptiform activity or the appearance of noncharacteristic foci of seizure onset. Variable effects on IEDs have been reported with long-term administration of phenytoin and carbamazepine, and no clear trend has emerged. Valproate dramatically suppresses generalized IEDs. In a series of studies, valproate decreased the number of generalized spike-wave discharges in 76% of patients 10 weeks after the drug was started, and the number of generalized spike-wave discharges was still decreased in 57% of patients 1 year after therapy was begun. Photoparoxysmal response was eliminated in 25% of patients with this finding 10 weeks after treatment and in 75% at 1 year after treatment; it is interesting that suppression of photoparoxysmal response persisted even after valproic acid was discontinued and serum levels were no longer detectable. Activation of generalized spike-wave activity with hyperventilation was not affected by treatment at 10 weeks but was eliminated in about half of patients with this finding at 1 year after initiation of treatment.^16,133 Although treatment with benzodiazepines, barbiturates, or valproic acid clearly affects the number of IEDs recorded in certain situations, these drugs are not usually discontinued when no IEDs are recorded and epilepsy is suspected. Atypical IEDs are often seen when barbiturates or benzodiazepines are withdrawn, and the suppressive effect of valproic acid is probably too lengthy to allow safe discontinuation for diagnostic purposes in most cases.

Seizure frequency was associated with greater likelihood of recording IEDs in one study,² but the opposite was found in another.¹³⁵ The first series included many children, whereas the second studied only adults, which may account for the discrepancy. Contemporary studies agree that IEDs occur more frequently in the period immediately after epileptic seizures.⁵² Nonetheless a small minority of people with a clear diagnosis of epilepsy will lack IEDs in spite of several routine EEG recordings. There are several potential explanations. IEDs may be infrequent and therefore not detected, even with aggressive sampling. They may be present but not detectable with traditional scalp electrode arrays. IEDs are frequently recorded with intracranial electrodes when results of simultaneous scalp recordings are normal.^1,125 More than 10 cm² of brain surface must be involved with an IED before the IED can be detected at the scalp, and fields of IEDs recorded with intracranial electrodes often occupy a smaller area.^1,125 Furthermore, many areas of the cortex, such as the basal frontal and medial temporal regions, are not directly accessible to scalp recording, and IEDs
originating in these regions may not be detected by scalp electrodes. Finally, some patients with epilepsy may have no IEDs even with intracranial recording, although experience suggests that this situation is rare, at least in intractable epilepsy.

The yield of IEDs can be increased by activation methods (sleep, sleep deprivation, hyperventilation, photic stimulation, and others; see Chapter 94), by longer EEG recordings, and by employing more electrodes than the routine number used with the 10-20 International System. Each of these may increase sensitivity of interictal EEG in supporting the diagnosis of epilepsy.

The increased yield of IEDs after sleep recordings has been accepted since the early reports of Gibbs and Gibbs.⁴⁵ Several reviews provide detailed information.^26,29,35,110 An EEG recorded during sleep will demonstrate IEDs in 40% or more of people with epilepsy in whom EEGs recorded during wakefulness reveal no IEDs. Chloral hydrate is most commonly used to hasten sleep onset. There is no evidence that spontaneous sleep is more effective than sedated sleep; however, sedative drugs, including chloral hydrate, may affect EEG background activity. The majority of studies have shown that 24 hours of sleep deprivation further increase the yield of IEDs by 20% or more after accounting for the increased yield expected with sleep and multiple EEGs.³⁵ Furthermore, activating effects of hyperventilation and photic stimulation are consistently potentiated by sleep deprivation. Many laboratories employ lesser amounts of sleep deprivation, even though these have not been proved to be as effective.

The occurrence of IEDs and changes in IED distribution and morphology differ depending on epilepsy syndrome and sleep stage. Hypsarrythmia is usually potentiated by sleep, and the EEG becomes increasingly abnormal with deeper stages of sleep; however, the EEG pattern often reverts to normal during rapid eye movement (REM) sleep. Rates of slow spike-wave activity, centrotemporal discharges, and 3-c/s spike-wave activity all increase during deeper stages of sleep. The activating properties of sleep are considered so profound in some of these syndromes that most epileptologists do not consider the EEG evaluation for infantile spasms, Lennox-Gastaut syndrome, Landau-Kleffner syndrome, or benign rolandic epilepsy complete unless a reasonable amount of sleep was recorded. In contrast, the polyspike-wave bursts seen in juvenile myoclonic epilepsy may decrease with sleep and are especially prominent on forced arousal from sleep. Although all studies agree that temporal IEDs are more frequent in non-REM sleep than in wakefulness or REM sleep, there is controversy as to whether frequency is higher in deep or light sleep.^81,109

Hyperventilation increases the frequency of generalized spike-wave activity in 50% to 80% of patients with absence seizures and often precipitates overt absences, especially in untreated patients.¹¹¹ However, altered clinical responsiveness can occur during hyperventilation in the absence of a spike-wave discharge and is a nonepileptic phenomenon observed even in normal children.³⁹ In contrast, hyperventilation increases the yield of focal IEDs in <10% of patients.²⁶

Photic stimulation induces IEDs in 10% of people with epilepsy.¹⁴² More than 80% of such patients have childhood absence epilepsy, juvenile absence epilepsy, juvenile myo-clonic epilepsy, or epilepsy with tonic–clonic seizures on awakening.¹⁴² When partial seizures are activated, clinical features usually suggest a posterior cerebral onset.

An incidental photoparoxysmal response in people without other IEDs or a clear history of seizures can be difficult to interpret. Photoparoxysmal response must be distinguished from the photomyoclonic response and potentiated visual-evoked responses, which are not associated with an increased risk for seizures.^{26,79,105,142} Initial studies reported that photoparoxysmal responses persisting beyond the period of photic stimulation were associated with epilepsy, whereas while those limited to the period of photic stimulation were not.¹⁰⁵ However subsequent work has not found this distinction to be useful.^65,120 The very large studies of highly selected airplane crew members report a prevalence of <0.3%.⁵⁴ Few adults with an incidental photoparoxysmal response eventually develop epilepsy.^54,114,120 Twin studies of the photoparoxysmal response indicate a concordance of virtually 100%, so the family history should be scrutinized if a photoparoxysmal response is found in someone without seizures. Consequently, most clinicians interpret an incidental photoparoxysmal response conservatively and do not recommend treatment on this basis alone.

IEDs may occasionally be provoked by tactile stimuli; these do not necessarily imply potential epileptogenicity.⁷⁶

Increasing the number of EEGs clearly increases the likelihood of recording IEDs. Salinsky et al.¹⁰⁸ found that approximately 60% of patients studied with up to seven EEGs demonstrated IEDs. About half of these had IEDs in the first EEG, and 90% of these had them after four EEGs. Walczak et al.¹³⁵ studied a similar population with continuous detection of spikes and seizures and obtained a somewhat higher yield of 80%. No study has compared the yield of continuous IED detection with that of serial routine EEGs.

Additional electrodes may provide useful information when IEDs are not recorded with the standard 10-20 International System array. This is especially relevant with temporal IEDs. Nasopharyngeal electrodes, true anterior temporal electrodes, and sphenoidal electrodes are all positioned closer to the anterior temporal lobe than are the F_7/8 electrodes. Several studies^{50,51,60,106,107,122} have compared how often temporal IEDs are recorded with these four electrodes. In patients in whom IEDs were recorded at some point during the study, IEDs were recorded in 43% to 58% with the standard electrode array, in 57% to 69% with nasopharyngeal electrodes, in 81% to 90% with true anterior temporal electrodes, and in 75% to 100% with sphenoidal electrodes. More artifacts were seen with nasopharyngeal recordings than with other electrode types. Consequently, nasopharyngeal electrodes have been largely abandoned in favor of anterior temporal electrodes. Sphenoidal electrodes are often used during evaluation for epilepsy surgery but are rarely used during routine EEG. Orbitofrontal electrodes help to record from the frontal pole and nasoethmoidal electrodes from the frontal operculum,¹⁰² and these may be helpful when seizures are thought to originate in the frontal lobe. It is not clear how often such electrodes record IEDs that would not be detected by the standard 10–20 International System electrode array.

Specificity is another important measure of the clinical utility of a test finding. Strictly speaking, the specificity of an IED is defined as the number of individuals not having a diagnosis of epilepsy that do not have IEDs. However, IEDs are rare in people without epilepsy, and so prevalence of IEDs in people without epilepsy is often discussed. All studies of EEGs in large numbers of normal individuals indicate that IEDs are seen in rare individuals without a clinical history of seizures. The likelihood of epilepsy developing in such people appears to be increased, although the likelihood is much less than in patients with paroxysmal clinical episodes who have IEDs. Prevalence of IEDs in nonepileptic individuals is influenced by age, general condition, and recording conditions, among other factors. Only one community-based study is available.¹⁹ Some degree of selection occurred in the remainder.

Table 3 summarizes the available information. Although the studies are few in number and have many limitations, several trends tentatively emerge. In healthy individuals, IEDs are more common in children (2.2% to 3.5%) than in adults (0.2% to 0.5%). Seizures appear to develop eventually in a higher percentage of healthy children with IEDs.¹²⁷ In the single study reporting results in hospitalized adults without neurologic or psychiatric disease, prevalence of IEDs was similar to that found in healthy persons.⁸ The higher rate in the frequently quoted study of Zivin and Ajmone-Marsan¹⁴⁴ may be attributable to the inclusion of patients who had cerebral neoplasms and underwent craniotomies. The similar rate reported among psychiatric inpatients¹⁵ may be related to the inclusion of patients who were withdrawing from barbiturates or who had anorexia. Interictal epileptiform discharges are more frequent in these conditions and are not necessarily associated with epilepsy.

The types of IEDs seen in normal individuals appear to differ from the types seen in large series of people with epilepsy. Centrotemporal IEDs, generalized IEDs, and the photoparoxysmal response account for the great majority of IEDs found in normal individuals, especially children.^8,19,33,54 On the other hand, focal or multifocal IEDs, especially temporal IEDs, predominate in series of people with epilepsy.^2,37,71 The types of IEDs observed in most nonepileptic subjects with IEDs appear to have a lower association with epilepsy than do other types of IEDs. Centrotemporal IEDs appear to be a heritable EEG pattern with incomplete and age-dependent penetrance.⁵⁶ Only approximately 40% of patients with centrotemporal IEDs had epileptic seizures in one large retrospective study.⁷¹ Generalized IEDs occur in approximately 10% of parents and 35% of other family members of probands with both tonic–clonic seizures and generalized epileptiform discharges.⁹¹ The photoparoxysmal response accounts for as many as 63% of IEDs found in individuals without epilepsy.⁵⁴ Most studies indicate that seizures develop infrequently in patients with this pattern (see earlier discussion).

Sharp waves or transients are frequently seen in normal neonates and appear to be a normal part of cerebral maturation as revealed by EEG. Frontal sharp transients are isolated sharp waves first present at 34 weeks of conceptional age, with maximum expression at approximately 36 weeks of conceptional age and then with gradual diminution in number and voltage following 44 weeks of conceptional age, and so they are rarely present following 6 weeks postterm. Frontal sharp transients are bilaterally synchronous and symmetric from the time of their first appearance. Focal sharp waves, especially midtemporal sharp waves or sharp slow complexes, can also be normal in premature neonates and perhaps as late as 44 weeks of conceptional age. In normal neonates these are rare in number, random in occurrence, and without persistent focality. The following features suggest that focal sharp waves are abnormal in this age range^23,72,95: (a) The amplitude of abnormal sharp waves is higher; (b) normal sharp waves occur randomly over the two hemispheres, whereas abnormal sharp waves occur in repetitive runs; (c) normal temporal sharp waves are typically mono- or diphasic, whereas abnormal sharp waves are typically polyphasic with aftergoing slow waves; (d) sharp waves with positive polarity are abnormal, whereas negative sharp waves may be normal; and (e) whereas normal temporal sharp waves may occur in any wake-sleep state in premature infants, they are more common turning transitional sleep in term infants. Temporal sharp waves present in awake term infants are probably abnormal, especially if they are persistent. Overall, the association between IEDs and epilepsy appears to increase as the neonate transitions into infancy and further. There is no information indicating the approximate age at which the degree of association between IEDs and seizures reaches that found in adults, and this is likely to vary depending on the type of IED.