The electroencephalogram (EEG) is an important tool in the evaluation of an infant or child with symptoms referable to the central nervous system (CNS). EEGs are used to assess seizure disorders, monitor the progression of a disease, and determine the prognosis for recovery or the development of long-term sequelae. Many EEGs are obtained to “rule out neurologic disorder” or simply to verify a diagnosis that has been well established clinically without much thought from the clinician as to the potential value or yield of such a test. However, the EEG may be overinterpreted by an electroencephalographer with little pediatric experience or training. The clinician, faced with a report of an abnormal EEG, may then feel obliged to undertake additional diagnostic tests, such as magnetic resonance imaging (MRI), or even to advise unnecessary therapy.
This chapter is an overview of pediatric EEG with particular emphasis on newborns and young infants. It illustrates the value of the EEG in the era of neuroimaging, and provides a source of references for those interested in a particular topic. Not all facets of pediatric EEG are reviewed. Rather, the normal patterns seen during infancy and childhood are discussed in detail, with emphasis on those that are often misinterpreted or “over-read.” Most EEG abnormalities are nonspecific and, in the absence of a clinical history, are of little diagnostic value, except to indicate the possibility of a pathologic process involving the CNS. Therefore, neurologic disorders that are often associated with specific and, in some cases, pathognomonic EEG patterns are emphasized.
The EEG is an important adjunct to the neurologic evaluation of the sick newborn infant. It provides an excellent, noninvasive method of assessing at-risk newborns and formulating a prognosis for long-term neurologic outcome. EEGs are commonly utilized in the United States for the evaluation of premature and full-term infants. Improvements in the obstetric management of high-risk pregnancies and major advances in neonatal medicine have decreased the mortality and lessened the morbidity of small premature infants. Although neonatal mortality has declined since the advent of neonatal intensive care units (NICUs), morbidity statistics have been changed less dramatically. The neurologic assessment of critically ill newborns is an important aspect of their initial care because preservation of cerebral function is the goal of the intensive supportive care. The major long-term sequelae of surviving newborn infants are neurologic in nature: cerebral palsy, intellectual disability, and other more subtle motor and cognitive deficits.
The major areas in which the EEG can provide unique information in the assessment of newborn full-term and preterm infants are:
Diagnosis and treatment of seizures.
Evaluation of infants with compromised cerebral function caused by primary neurologic disorders (e.g., hypoxic-ischemic encephalopathy and cerebral infarction) and those with significant systemic disease (e.g., severe respiratory distress syndrome or sepsis) who are at risk for secondary encephalopathies.
Estimation of the conceptional age (CA), which is defined as the estimated gestational age (EGA) plus the legal age. The EGA is the age in weeks of the infant at birth calculated from the date of the mother’s last menstrual period or by ultrasound study or a standardized physical examination. Legal age is that of the infant since birth.
Identification of specific neurologic entities (e.g., intraventricular hemorrhage, periventricular leukomalacia, congenital brain malformations, viral encephalitis, and metabolic encephalopathies).
Determination of prognosis and long-term neurologic outcome.
The EEG is also used in neonatal research concerned with the study of infant behavior and the assessment of physiologic functions that are dependent on the sleep state. For example, studies of neonatal apnea have demonstrated marked differences in the quantity and quality of apneas in active and quiet sleep.
The neurologic disorders of newborn infants are often unique to this period of life. Asphyxia is a common cause of neurologic compromise in the newborn and leads to hypoxic-ischemic encephalopathy. Placental dysfunction, particularly infection, can cause periventricular leukomalacia in premature infants. Metabolic encephalopathies (e.g., hypoglycemia) may cause transient EEG abnormalities, yet permanently impair cerebral function. Transient and more benign metabolic abnormalities (e.g., hypocalcemia) cause seizures but are becoming less common with improvements in neonatal intensive care. Meningitis, traumatic lesions (including subarachnoid hemorrhage), congenital malformations of the CNS, drug withdrawal, and inherited disorders of CNS metabolism (e.g., amino or organic acidurias) constitute the bulk of the remaining neurologic syndromes of relevance to the electroencephalographer.
Many technical aspects of EEG recording are unique to the newborn infant. It is beyond the scope of this chapter to detail the nuances of EEG recording in the newborn nursery, which is reviewed elsewhere. Table 4.1 is a summary of some major technical areas in which EEG recording in neonates differs from that in older children and adults. This recording method is used for preterm and full-term infants until approximately 1 to 2 months after term. The American Clinical Neurophysiology Society (formerly the American Electroencephalographic Society) has published guidelines for the recording of EEGs in infants.
|Electrodes and placement||Electrodes and placement|
|Paste or collodion attachment||Needle electrodes are never used; collodion may not be allowed in some nurseries|
|Minimum of 9 scalp electrodes applied in premature infants and entire 10–20 array in term infants||Small head limits electrodes to frontal, central, occipital, midtemporal, and Cz in premature infants|
|Fp1 and Fp2 placements replaced by Fp3 and Fp4 (halfway between 10–20 placements Fp and F)||Frontal sharp waves, delta, and other activity are of higher amplitude in prefrontal than frontopolar region|
|At least 16-channel recording is preferred||Many polygraphic parameters must be measured; all brain areas can be monitored with one montage|
|Polygraphic (non-EEG) variables recorded routinely—respiration (thoracic, with or without nasal thermistor); extraocular movements (primarily in infants older than 34 weeks CA); and electrocardiogram||These variables are essential in the determination of the behavior state (awake, active sleep, or quiet sleep) and the recognition of artifacts|
|Single montage used for entire recording, particularly with 16-channel recording||Generalized changes in background activity and state-related changes are more important than exact localization of focal abnormalities|
|Screen display of 15 to 20 sec for entire record. Long time constant—between 0.25 and 0.60 sec||It is easier to recognize interhemispheric synchrony and slow background activity|
|Frequent annotations by the technologist of baby’s body movement and, in small premature infants, eye movement||Important information for the determination of behavioral state and possible artifact|
|Technologist attempts to record active and quiet sleep, particularly in older premature and term infants. Duration of record may exceed 60 min||Presence or absence of well-developed sleep states is important for interpretation; some pathologic patterns are seen primarily in quiet sleep|
|Accurate notation of EGA, CA, recent drug administration, recent changes in blood gases||Interpretation is dependent on knowledge of CA of infant. The EEG is very sensitive to abrupt changes in blood gases and certain medications|
The technologist should have additional training in a laboratory that is already recording from neonates, should become familiar with nursery procedures, and should develop a good rapport with the nursery staff. The technologist should annotate normal and abnormal body and facial movements of newborn infants. In conjunction with the respirogram, electro-oculogram, and EEG, the infant’s movements provide the necessary data to determine the behavioral or sleep state. An EEG lacking such clinical observations and polygraphic (non-EEG) variables is extremely difficult, if not impossible, to interpret unless it is grossly abnormal. The use of digital EEG and simultaneous video recordings of the newborn is helpful for detecting movements and artifacts, but does not obviate the need for good annotation and physiologic monitors.
There is extensive literature on the normal cerebral electrical activity of the premature and full-term newborn. Only a summary of the salient characteristics of the normal EEG at each conceptional age is given here.
CNS Maturation and Ontogenic Scheduling
Spontaneous electrocerebral activity in premature newborn infants evolves more rapidly than at any other time during human life. A close and consistent relationship exists between the changes in the EEG and the maturational changes of the nervous system. To understand and interpret neonatal EEGs, it is important to recognize the rapid maturational changes that take place in the brain during the last trimester. Such development takes place in an extrauterine environment in the premature infant. There is a rapid enlargement of the brain, with the weight increasing fourfold from 28 weeks to 40 weeks CA. Its appearance changes from a relatively primitive-looking structure before 24 weeks when the surfaces of the cerebral hemispheres are smooth, to 28 weeks when the major sulci make their appearance. The sulci and gyri continue to develop, until ultimately the complex brain of the full-term infant is achieved. Rapid maturational changes are also apparent in the neurochemical milieu; in interneuronal connectivity, with the development of dendritic trees and synaptogenesis; and in the myelination of axons.
Despite the apparent simplicity of the anatomy of the premature infant’s brain, its repertoire of function is rather extensive. The premature infant of 28 weeks CA is capable of complex, spontaneous motor activity, vigorous crying, and response to stimuli. Behavioral states are also relatively well developed at 28 weeks CA.
In parallel with the developmental maturation of the brain, there is maturation of the bioelectric cerebral activity. According to the rule of ontogenic scheduling, the bioelectric maturation of cerebral activity of the healthy infant occurs in a predictable, time-linked manner. This maturational process is dependent primarily on the age of the brain (i.e., CA), and is independent of the number of weeks of extrauterine life. Therefore, the EEG of a premature infant born at 30 weeks EGA whose legal age is 10 weeks (CA 40 weeks) is similar to that of a 38-week EGA infant who is 2 weeks old. Some minor differences exist in the EEGs of premature infants who mature to term and those of full-term newborns, but these differences appear to be of little clinical significance. Consequently, an accurate estimate of the CA is essential for the correct interpretation of neonatal EEGs.
The EEG patterns of the newborn infant are dependent not only on the CA, but also on the behavioral states of the infant during the recording. Distinct EEG patterns of active (rapid eye movement or REM) and quiet (non-REM) sleep can be identified easily in normal infants by 35 weeks CA and in many infants as early as 27 to 28 weeks CA ( Fig. 4-1 ). It is important to record both sleep states whenever possible, as abnormalities may be found only in one sleep state.
The full-term newborn has easily recognizable waking and sleeping behavioral states that are very similar to those of older children and adults. These states are generally classified as waking, active sleep (REM sleep), quiet sleep (non-REM sleep), and indeterminate or transitional sleep (a sleeping state that cannot be classified definitely as either active or quiet sleep). Active sleep is the most common sleep-onset state in newborn infants and remains so for the first 2 to 4 months of post-term life ; it constitutes approximately 50 percent of the sleeping time in term infants and a slightly higher proportion in premature infants. The duration of active sleep at onset is usually 10 to 20 minutes but may exceed 40 minutes in some newborns.
A normal infant is considered awake if the eyes are open. Behavior may vary from quiet wakefulness to crying with vigorous motor activity. Transient eye closures may accompany crying and also occur during quiet wakefulness. If the eyes remain closed for an extended period of time (usually for more than 1 minute), the infant is considered asleep.
Active sleep in infants older than 29 weeks CA is characterized by eye closure, bursts of rapid horizontal and vertical eye movements, irregular respirations, and frequent limb and body movements ranging from brief twitches of a limb to gross movements of one limb or the entire body, grimacing, smiling, frowning, and bursts of sucking.
Quiet sleep in infants older than 29 weeks CA is characterized by the infant lying quietly with eyes closed, regular respiration, and the absence of rapid eye movements. There is a paucity of body and limb movements, although occasional gross body movements, characterized by brief stiffening of the trunk and limbs, may occur and sometimes are associated with brief clonic jerks of the lower extremities.
Distinctive Patterns in Neonatal EEGs
Certain distinct EEG patterns are common in the newborn, especially in the premature infant. These specific patterns serve as useful findings for determining the CA of the infant’s nervous system. Figure 4-2 depicts some of these waveforms.
Theta Bursts (Sawtooth Waves)
Occipital theta bursts consist of rhythmic, medium- to high-amplitude, 4- to 7-Hz activity located in the occipital regions. These occipital theta bursts are present as a physiologic pattern in very premature infants (maximal between 24 and 26 weeks CA) and disappear by 32 weeks CA. This pattern serves as a useful hallmark in determining CA.
Temporal theta burst activity is seen commonly in the EEGs of slightly older premature infants. This activity appears at approximately 26 weeks CA, with the maximal incidence between 27 and 31 weeks, and is seen rarely beyond 35 weeks. The bursts consist of sharply contoured, high-amplitude (50 to 150 μV, occasionally 200 to 300 μV), 4- to 6-Hz activity; they are maximal in the temporal areas, but often become diffuse and, commonly, bilateral and synchronous (see Fig. 4-1 ).
These regular theta rhythms originate in the occipital regions in very premature infants and with increasing CA migrate toward temporal regions. The gradient of occipitotemporal maturation of this pattern appears to coincide with the timing of gyral development in these regions.
Delta brushes are a hallmark of prematurity and are most abundant between 32 and 35 weeks CA (see Fig. 4-2 ). They consist of short bursts of 8- to 20-Hz rhythmic activity, often with spindle morphology, superimposed on high-amplitude slow waves (0.8 to 1.5 Hz). The amplitude of the fast activity may range from 10 to 100 μV (usually 20 to 50 μV), and that of the slow waves from 25 to 200 μV. Typically, delta brushes occur asynchronously in the homologous areas of the two hemispheres. Delta brushes predominate in the central (rolandic) region in very young premature infants (younger than 32 weeks CA), and extend to the occipital and temporal regions in older premature infants. Brushes are more abundant in active sleep in younger infants (up to 32 weeks CA) and in quiet sleep in infants older than 33 weeks CA. Delta brushes peak in abundance between 31 and 33 weeks CA, and decrease in number with increasing CA. In full-term infants, they are infrequent during quiet sleep and virtually absent in active sleep.
Frontal Sharp Transients ( Encoches Frontales )
Frontal sharp transients are biphasic sharp waves (initially a negative deflection, followed by a wider positive deflection); they are of maximal amplitude in the prefrontal regions (Fp3, Fp4) and occur primarily during sleep, particularly during the transition from active to quiet sleep. The typical amplitude is 50 to 150 μV and the duration of the initial surface-negative component is 200 msec. They usually appear bilaterally and synchronously, although they may be asymmetric in amplitude. Sometimes they appear unilaterally, shifting from one side to the other during the record. Typical frontal sharp transients appear at 35 weeks CA and persist until several weeks after term (see Fig. 4-2 ).
Rhythmic Frontal Delta Activity (Anterior Slow Dysrhythmia)
Rhythmic frontal delta activity consists of bursts and short runs of 1.5- to 4-Hz, 50- to 200-μV delta activity that is often monomorphic and of maximal amplitude in the frontal regions (see Fig. 4-2 ). Like frontal sharp transients, rhythmic frontal delta activity is most prominent during transitional sleep. This activity appears at approximately 37 weeks CA and lasts until approximately 6 weeks after term. Rhythmic frontal delta activity is often intermixed with frontal sharp transients. This activity should be distinguished from prolonged runs of monorhythmic bifrontal delta activity that persists in all sleep states, which is an abnormal finding.
In very young premature infants, the electrical activity of the brain is interrupted by long periods of quiescence. This pattern, called tracé discontinu , may be present in all states of sleep in very premature infants and persists to some degree in infants of 34 to 36 weeks CA. During periods of quiescence, the EEG is very low in amplitude (less than 30 μV) and may even be flat at standard amplification. The active periods increase in duration as the CA increases ( Fig. 4-3 ). They are composed of bursts containing monomorphic delta, theta, and faster activity, delta brushes, and temporal theta bursts. The mean duration of interburst intervals in the healthy premature infant decreases with increasing gestational age; the mean interval at 27 to 29 weeks CA is approximately 6 seconds, and at 30 to 34 weeks is approximately 4 to 5 seconds. The maximum interburst intervals should not exceed approximately 30 seconds at any age beyond 27 weeks CA ( Fig. 4-4 ). In healthy, very preterm infants, the maximum interburst interval duration can be much longer: 126 seconds at 21 to 22 weeks CA, 87 seconds at 23 to 24 weeks CA, and 44 seconds at 25 to 26 weeks CA.
As the CA increases, the periods of relative electrocerebral inactivity become shorter in duration, and the interburst intervals display generalized amplitude attenuation rather than quiescence. This pattern, called tracé alternant , is the discontinuous pattern that characterizes the quiet sleep of the full-term newborn ( Fig. 4-5 ). Tracé alternant consists of 3- to 6-second bursts of high-amplitude delta and theta activity (1 to 6 Hz, 50 to 100 μV) admixed with lower-amplitude beta and theta activity, which occur at intervals of 3 to 6 seconds. The bursts may also contain scattered isolated sharp transients. The interburst intervals contain diffuse moderate-amplitude (25 to 50 μV), mixed-frequency (usually 4- to 12-Hz) activity, similar to that occurring during wakefulness and active sleep following quiet sleep.
A gradual transition from tracé discontinu to tracé alternant occurs as the premature infant approaches term, although even in the healthy full-term infant, occasional short, very-low-amplitude interburst intervals are seen. Tracé discontinu , therefore, differs from tracé alternant primarily on the basis of the amplitude of the activity during the interburst interval.
Interhemispheric synchrony is defined as the relatively simultaneous appearance in both hemispheres of bursts of cerebral activity during discontinuous portions of the record. In the very young premature infant (younger than 29 weeks CA) the EEG activity during tracé discontinu is hypersynchronous, with the degree of synchrony approaching 90 to 100 percent. The degree of interhemispheric synchrony decreases to 60 to 80 percent for the next 4 to 5 weeks (30 to 35 weeks CA) and gradually approaches 100 percent as the infant reaches term.
Age-Specific Background Patterns
Normal Patterns at 23 to 26 Weeks CA
The EEG in very preterm infant is consistently discontinuous ( tracé discontinu ), although some variability in the organization of discontinuous patterns can be observed. In some phases bursts can be longer and intervals shorter, whereas at other times the discontinuity is more marked. In a study of five subjects in this age group, the maximum interburst interval duration ranged between 20 and 62 seconds and the minimum burst length was 1 to 2 seconds. Another study found that the maximum interburst interval duration was 87 seconds at 23 to 24 weeks CA and 44 seconds at 25 to 26 weeks CA. Occipital theta bursts are prominent (maximal at 25 weeks CA). Temporal theta bursts can be seen, but are less common than in older preterm infants. Some delta brushes are observable, but the incidence is generally low (between 9 and 21 percent of bursts).
Normal Patterns at 27 to 29 Weeks CA
The EEG at 27 to 29 weeks CA is characterized by a discontinuous background, tracé discontinu , that consists of mixed-frequency activity, primarily in the delta range. Bursts contain runs of high-amplitude, very slow, rhythmic occipital delta activity; and somewhat lower-amplitude, more arrhythmic central and temporal delta activity. The maximum interburst intervals are generally less than 30 seconds.
Beginning as early as 27 weeks CA, there may be behavioral and EEG differentiation of active and quiet sleep states. During active sleep, REMs are observed with burst of slow waves admixed with theta activity. During quiet sleep, REMs are absent and the background becomes more discontinuous.
High-amplitude temporal theta bursts (see Fig. 4-1 ) become prominent during this period. Isolated frontopolar slow waves with superimposed faster activity (so-called “delta crests”) are also common in this age group. Delta brushes are present primarily in the central (rolandic) and occipital regions.
Normal Patterns at 30 to 32 Weeks CA
Two distinct types of background predominate at 30 to 32 weeks CA. A discontinuous background ( tracé discontinu ) composed of an admixture of occipital delta (1 to 2 Hz, 25 to 100 μV) and centrotemporal delta-, theta-, and alpha-range activity occurs when the child is in quiet sleep. More continuous background occurs with eye movements and more active body movements, consistent with active sleep. Synchrony during tracé discontinu reaches a nadir during this period (approximately 60 to 75 percent).
Delta brushes are very abundant at this age; they are located in the occipital, central, and temporal regions with a higher incidence during the continuous portions of the record (active sleep). There are fewer temporal theta bursts than at earlier ages, and the delta crests and occipital theta bursts are less common.
Normal Patterns at 33 to 35 Weeks CA
Sleep states at 33 to 35 weeks CA are becoming more clearly defined. Discontinuous activity ( tracé discontinu ) is associated with quiet sleep. During wakefulness and active sleep, continuous background slow activity (primarily 1 to 2 Hz, 25 to 100 μV) that is maximal in the temporal, central, and occipital regions is more abundant. In general, an increasing amount of lower-amplitude theta and faster activity is also present in all states. Synchrony approaches 80 percent at 35 weeks CA.
Delta brushes are still abundant but fewer than at an earlier age, and they are more frequent during quiet sleep. Typical frontal sharp transients ( encoches frontales ) appear at 34 to 35 weeks CA. Temporal theta activity should be much less common.
Normal Patterns at 36 to 38 Weeks CA
Three different EEG patterns predominate at 36 to 38 weeks CA. (1) Continuous, diffuse, low-amplitude, 4- to 6-Hz activity (usually less than 50 μV) characterizes quiet wakefulness ( activité moyenne ) . This activity is often admixed with 50- to 100-μV delta (2- to 4-Hz) activity and is the predominant rhythm of sleep-onset active sleep. Delta brushes are rare in active sleep. (2) A discontinuous pattern ( tracé alternant ) with bursts of mixed-frequency slow activity and occasional delta brushes, separated by low- to medium-amplitude background, typifies quiet sleep. During this period, there is a transition from tracé discontinu to tracé alternant . The tracé alternant pattern persists until 4 to 8 weeks after term. (3) The active sleep that follows quiet sleep is characterized by low- to moderate-amplitude theta activity with occasional delta waves at lower amplitude than that occurring during sleep-onset active sleep. Synchrony approximates 90 percent during the discontinuous portions of the EEG.
Delta brushes are less abundant during quiet sleep and are virtually absent during active sleep. Frontal sharp transients and rhythmic frontal delta activity are prominent, particularly during the transition from active sleep to quiet sleep.
Normal Patterns at 38 to 42 Weeks CA (Full-Term Newborn)
Sleep and waking cycles are well established at 38 to 42 weeks CA. Four distinctive patterns occur during sleep. (1) Low- to medium-amplitude theta activity with superimposed delta activity (2 to 4 Hz, less than 100 μV), the latter appearing as continuous activity or in short runs or bursts, characterizes active sleep, particularly at sleep onset. (2) Diffuse continuous delta activity (0.5 to 2 Hz, 25 to 100 μV) is found at the beginning and end of quiet sleep periods and occasionally is found during long portions of such sleep (slow-wave quiet sleep). (3) Tracé alternant is characteristic of well-established quiet sleep. (4) Activité moyenne is a continuous low-amplitude activity (25 to 50 μV) at 1 to 10 Hz (predominantly 4 to 7 Hz) that characterizes wakefulness and active sleep, particularly following a period of quiet sleep. Interhemispheric synchrony during tracé alternant approaches 100 percent, but transient asynchrony may occur at its onset.
Delta brushes are rare in the EEGs of term infants; they occur primarily during quiet sleep. Frontal sharp transients and rhythmic frontal delta activity are abundant, particularly during the transitional period between active sleep and quiet sleep. The amplitude of the background activity is relatively symmetric over the two hemispheres, although transient asymmetries are common, particularly in the temporal regions. Rhythmic theta-alpha bursts, which consist of bursts or short runs (1 to 3 seconds) of sharply contoured, primarily surface-negative, rhythmic theta- and alpha-range activity, are common in the central and midline frontocentral areas, particularly during quiet sleep. Scattered isolated sharp waves are common in the temporal regions and are less common in the central and midline regions. These sharp transients usually are incorporated within the bursts of the tracé alternant .
Figure 4-6 summarizes the various EEG patterns that are seen in the neonatal EEG and how these patterns change as the premature infant matures to term. By knowing which maturational patterns become prominent at a particular CA, an estimation of the CA can be determined.
Pitfalls in Neonatal EEG Interpretation
Inexperienced electroencephalographers often mistakenly characterize the frontal sharp transients and rhythmic frontal delta activity of the older premature infant and the temporal theta bursts of the very premature infant as “epileptiform” or “paroxysmal.” The abundant brushes in the EEGs of younger premature infants may also give the background a “spiky” or “paroxysmal” appearance, especially when a compressed screen display (20 seconds) is used. Multifocal sharp waves are also noted, particularly in the temporal and central regions of healthy infants.
The various discontinuous background patterns of healthy premature infants must be distinguished from the abnormal paroxysmal patterns discussed in the next section. Transient interhemispheric amplitude asymmetries are common in all age groups, as is asynchrony of the bursts during the discontinuous portions of the tracings in premature infants (approximately 31 to 35 weeks CA). There may be significant asymmetries in the onset of the tracé alternant pattern in normal infants, with tracé alternant appearing in one hemisphere 1 to 5 minutes before it emerges in the other.
It is beyond the scope of this chapter to discuss all the deviations from normal that are encountered in the EEGs of newborn infants. Rather, emphasis is placed on the patterns most commonly seen, those associated with certain specific neurologic disorders, and those that appear to be of prognostic value.
Abnormal Background Patterns
Only the most common EEG abnormalities that occur in the neonatal EEG are discussed in this section. Additional discussions of the prognostic value of the EEG in full-term newborns can be found elsewhere. One of the more common causes of a severely abnormal EEG background is hypoxic-ischemic encephalopathy. An EEG performed early can be valuable in predicting the severity of the encephalopathy and prognosis. The following section discusses the various abnormal EEG patterns and their prognostic significance ( Table 4-2 ).
|EEG Background Patterns||Percentage with Favorable Outcome|
|Diffuse slow activity||15–20|
Before interpreting the EEG, it is important to consider any acute changes in metabolic status, administration of neuroactive medications, or changes in respiratory status (e.g., oxygenation or pH) that may affect the background EEG acutely. For example, the EEG may become transiently abnormal and even isoelectric during acute hypoxemia. Many drugs used in nurseries (e.g., intravenous diazepam, lorazepam, midazolam, and opioids) may cause an acute but reversible depression of the EEG. The synthetic opioid, sufentanil, increases the discontinuity of the background in premature infants. Intravenous morphine infusions may cause excessive discontinuity, sharp waves, and even burst-suppression patterns. Therefore if these drugs are being administered at the time of the EEG, the record should be interpreted with caution and another record should be obtained after their discontinuance.
Electrocerebral Inactivity (Isoelectric)
Electrocerebral inactivity implies that there is no discernible cerebral electrical activity, even at high sensitivities. Such an EEG pattern is seen in various clinical settings, most commonly following severe asphyxia, circulatory collapse, and massive intracerebral hemorrhage. It can also be seen in bacterial meningitis, encephalitis, severe malformations of the brain (e.g., hydranencephaly or massive hydrocephalus), and inborn errors of metabolism. The same technical requirements for recording isoelectric records in adults are applied to infants. High sensitivities (equal to or exceeding 2 μV/mm), long time constants (0.1 seconds or more, or low-frequency filter of 1 Hz), and long interelectrode distances should be used. The technologist should also perform various auditory and nociceptive stimulations to confirm lack of reactivity.
In the absence of drug intoxication, acute hypoxemia, hypothermia, and postictal state, this EEG pattern carries a grave prognosis in neonates; however, it does not necessarily indicate “brain death.” The vast majority of neonates with inactive EEGs either die in the neonatal period or survive with severe neurologic deficits. The neuropathology of the brains of neonates with inactive EEGs reveals widespread encephalomalacia and ischemic neuronal necrosis involving the cerebral cortex, corpus striatum, thalamus, midbrain, pons, and medulla.
In newborns, the return of EEG activity may be seen after an isoelectric EEG recorded in the first 24 hours of life or immediately after an acute hypoxic-ischemic event. If recovery occurs, the isoelectric EEG usually is followed by other abnormal patterns (particularly, diffusely slow backgrounds), but occasionally a flat EEG persists for many months following the acute neurologic insult.
Inactive EEG and Brain Death
There are no universally agreed criteria for brain death determination in newborn infants. The Task Force for the Determination of Brain Death in Children established the criteria for determining brain death in children, but did not provide guidelines for diagnosing brain death in infants under the age of 7 days. For infants between the ages of 7 days and 2 months, the Task Force recommended that two physical examinations and EEGs separated by 48 hours be performed routinely.
During the first week after birth, an isoelectric EEG is not a reliable test for determining brain death. However, some believe that brain death can be diagnosed in the full-term newborn, even when less than 7 days of age. If the initial EEG in the newborn shows electrocerebral inactivity in the absence of barbiturates, hypothermia, or cerebral malformations (e.g., hydranencephaly or hydrocephalus), and if the findings on neurologic examination remain unchanged after 24 hours, electrocerebral inactivity is confirmatory of brain death. Based on available data, the risk of misdiagnosis appears exceedingly low. Most neurologists believe that the EEG is a useful adjunct to the determination of brain death if it is performed by an experienced technologist using the standards of the American Clinical Neurophysiology Society (formerly the American Electroencephalographic Society ) and is interpreted by an experienced electroencephalographer.
Burst-Suppression (Paroxysmal) Pattern
The burst-suppression pattern (see Fig. 4-5, A ) is characterized by an isoelectric background interrupted by nonperiodic bursts of abnormal activity: delta and theta activities with admixed spikes, beta activity, or both; less commonly, bursts or short runs of diffuse or focal alpha or theta activity that is occasionally rhythmic. The bursts, which are usually highly synchronous between hemispheres, contain no age-appropriate activity. In the most severe form, this pattern is invariant and minimally altered by stimuli, and persists throughout waking and sleeping states. This abnormal pattern must be differentiated from the normal discontinuous patterns seen in the quiet sleep of premature infants.
Burst-suppression patterns are seen following a variety of severe brain insults (e.g., asphyxia, severe metabolic disorders, CNS infections, and cerebral malformation). A burst-suppression pattern can be induced pharmacologically with high doses of barbiturates and other neuroactive medications. In the absence of significant concentrations of neuroactive medications, a burst-suppression pattern usually is associated with a very poor prognosis (85 to 100 percent unfavorable prognosis, depending on the study). Infants who have a burst-suppression pattern that changes with stimulation have a somewhat better prognosis. A burst-suppression pattern in the first 24 hours of life that is replaced rapidly by a less severely abnormal EEG is followed occasionally by a normal neurologic outcome.
Variants of Burst-Suppression Pattern
In a small group of full-term newborns with neonatal hypoxic-ischemic encephalopathy, Sinclair and co-workers compared different types of burst-suppression patterns with outcome. Burst-suppression pattern was defined as a background with bursts lasting 1 to 10 seconds alternating with periods of marked background attenuation (amplitude consistently less than 5 μV). A modified burst-suppression pattern was defined as a burst-suppression pattern that was not constantly discontinuous throughout the recording, had periods of attenuation that contained activity higher than 5 μV, or both. Those newborns with a burst-suppression pattern had poor outcome: death in 6 of 15, severe disability in 4, and normal outcome in 2 of 9 survivors. The outcome in those with a modified burst-suppression pattern was more favorable: neonatal death in 1 of 8, severe disabilities in 1, and normal outcome in 3 of 7 survivors. The EEGs were performed within the first week of life, but the timing of the EEG in regard to the timing of the insult was not provided. The study did not assess whether the burst-suppression pattern could be modified with stimulation.
Instead of rigidly distinguishing between burst-suppression and other constantly discontinuous patterns based on the amplitude of activity during the suppressions, Biagioni and colleagues examined EEG parameters of discontinuity and correlated these scores with the outcome. In 32 full-term infants with hypoxic-ischemic encephalopathy who had an EEG with a constantly discontinuous pattern, they noted the minimum burst duration (activity greater than 45 μV), maximum interburst interval duration (interburst interval was defined as a period of activity less than 45 μV), and mean interburst interval amplitude. The best indicators predictive of a normal outcome were maximum interburst interval duration shorter than 10 seconds, mean interburst interval amplitude greater than 25 μV, and minimum burst duration longer than 2 seconds. The maximum interburst interval duration correlated with the severity of the hypoxic-ischemic encephalopathy. It was also lengthened significantly (more than double) in children who received phenobarbital. The timing of the EEG was also an important factor in determining prognosis; of the nine subjects who had a constantly discontinuous EEG after the eighth day from birth none had a normal outcome. All of the 32 EEGs in this study would be compatible with the definition of modified burst-suppression pattern, as defined by Sinclair and colleagues, because there were no records with interburst interval amplitudes less than 5 μV.
Menache and co-workers also examined several EEG parameters of discontinuity in 43 term or near-term infants who had a variety of neurologic disorders. They included EEGs with constantly or transiently discontinuous patterns. Only 7 had a burst-suppression pattern as defined above. A predominant interburst interval duration (defined as the duration of more than 50 percent of all interburst intervals with amplitudes less than 25 μV) of longer than 30 seconds correlated with the occurrence of both unfavorable neurologic outcome and subsequent epilepsy. Infants with this finding had 100 percent probability of severe neurologic disabilities or death (all had hypoxic-ischemic encephalopathy) and an 86 percent chance of developing subsequent epilepsy.
Excessively Discontinuous Background
The background EEG is normally discontinuous in very premature infants, with periods of total absence of cerebral activity lasting many seconds. With the maturation of the brain, the duration of these flat periods, or interburst intervals, decreases as the preterm infant approaches term. Conversely, the duration of the longest period of continuous EEG increases with CA (see Figs. 4-3 and 4.4 ). In healthy, very preterm infants the maximum interburst interval duration in one study was 126 seconds at 21 to 22 weeks CA, 87 seconds at 23 to 24 weeks CA, and 44 seconds at 25 to 26 weeks CA.
There are some differences among studies regarding the duration of normal interburst intervals, owing in part to the various criteria used to define them. Nevertheless, a conservative statement is that the maximum duration of interburst intervals should not exceed 60 seconds at 24 to 27 weeks CA, 30 seconds at 28 to 29 weeks CA, and 20 seconds at 30 to 31 weeks. In full-term infants the maximum interburst interval should not exceed 6 seconds. Excessively long interburst intervals are indicative of encephalopathy at all CAs, and generally have unfavorable prognostic implications.
Unlike the burst-suppression pattern, there is some reactivity to tactile stimulation and often preservation of sleep-state transitions in an EEG with an excessively discontinuous background. Furthermore, in the premature infant, periods of EEG activity separated by excessively long interburst intervals may contain many of the normal transients that are abundant at this age (e.g., delta brushes and temporal theta bursts), whereas the bursts within the burst-suppression pattern of term infants are composed of abnormal EEG activity.
Low-Voltage Undifferentiated Pattern
A low-voltage pattern is usually defined as activity between 5 and 15 μV during all states. Faster frequencies tend to be depressed or obliterated. Differentiation of sleep state in low-voltage records may be difficult, although some amplitude differences may exist between sleep states, with amplitudes being slightly higher in quiet sleep than in active sleep.
Low-voltage records are seen in a variety of severe CNS disorders, including hypoxic-ischemic encephalopathy; toxic-metabolic disturbances; congenital hydrocephalus; and severe intracranial hemorrhage, including large subdural hematomas. The prognostic value of the low-voltage EEGs depends strongly on the timing of recording after a presumptive brain injury. The pattern is ominous, especially when it persists beyond the first week after the insult; there is high probability of neurologic sequelae or death. Therefore, EEGs obtained shortly after an acute event should be interpreted with caution, and a follow-up study should be performed.
Several caveats about the interpretation of low-voltage EEGs are required. It is important that the recording is long enough to include periods of quiet sleep (which are generally higher in amplitude than active sleep). Neurodepressive medications and surfactant treatment may depress EEG amplitude. Severe scalp edema, subgaleal hematomas, subdural hematomas, and extra-axial fluid collections may also attenuate the EEG amplitude artifactually.
Low-Voltage Background with Theta Rhythms
The low-voltage background with theta rhythms is a variant of the low-voltage pattern. Continuous low-voltage background (5 to 25 μV) is accompanied by low-voltage (5 to 15 μV) theta activity that occurs in bursts of varying lengths or sometimes continuously. The theta activity may be diffuse but is more often focal or multifocal. The EEG shows no reactivity to stimulation and, usually, no discernible sleep states. This pattern is usually seen in neonates with hypoxic-ischemic encephalopathy and is highly associated with unfavorable outcomes.
Diffuse Slow Activity
The pattern of diffuse slow activity, also called the monomorphic medium-voltage pattern, consists of widespread amorphous delta activity that persists throughout the recording and is not altered significantly by sensory stimuli ( Fig. 4-7 ). The faster patterns in the theta and beta range, which are normally abundant in the EEGs of term infants, are absent. The background is devoid of the normal patterns seen in the premature infant, such as delta brushes, temporal theta bursts, and occipital theta bursts. This pattern may emerge during the recovery phase after hypoxic-ischemic encephalopathy, replacing more severe patterns such as electrocerebal inactivity, low-voltage undifferentiated pattern, or burst-suppression. When a pattern of diffuse slow activity persists beyond 1 month in full-term neonates after hypoxic-ischemic encephalopathy, the prognosis appears to be poor, with two-thirds of the subjects having neurologic sequelae.
This type of abnormal background must be distinguished from the high-amplitude, slow-wave pattern that occurs during a portion of the normal quiet sleep of preterm and term infants, and that gradually replaces the tracé alternant pattern during the first 4 to 6 weeks after term. This pattern should also be distinguished from frontal rhythmic delta activity (anterior slow dysrhythmia), which is a normal pattern seen in transitional sleep that appears at 37 weeks CA and persists for several weeks after term.
Grossly Asynchronous Records
The degree of interhemispheric synchrony is dependent on the CA. A record is considered to be grossly asynchronous if, during the discontinuous state, all background rhythms are persistently asynchronous between the hemispheres (estimated less than 25 percent synchrony for infants older than 30 weeks CA). The majority of records with gross interhemispheric asynchrony have paroxysmal backgrounds ( Fig. 4-8 ). Records with markedly asynchronous burst-suppression patterns are often seen in infants with severe hypoxic-ischemic encephalopathy, congenital malformations (e.g., agenesis of corpus callosum or Aicardi syndrome), and periventricular leukomalacia. A grossly asynchronous pattern usually is associated with an unfavorable outcome.
Amplitude Asymmetry Pattern
A persistent amplitude asymmetry in the background activity between the hemispheres ( Fig. 4-9 ) that exceeds 50 percent and is present in all states is thought to be significant. This pattern commonly correlates with lateralized structural pathologies (e.g., intraparenchymal hemorrhages, strokes, or congenital malformations). It is important to exclude the presence of asymmetric scalp edema or cephalohematomas and technical pitfalls (e.g., electrode “salt bridges” or asymmetric electrode placement).
Transient amplitude asymmetries have nonspecific prognostic significance. Transient unilateral attenuation of background EEG activity, usually lasting about 1 minute, may occur rarely during slow-wave quiet sleep in normal newborns. These episodes usually occur within minutes of the time the infant first enters slow-wave quiet sleep and are accompanied by normal background activity. Transient asymmetries also may occur after subclinical electrographic seizures.
Focal abnormalities usually consist of localized amplitude attenuation of background activity, with or without spikes (see Fig. 4-9 ). Again, it is important to exclude electrode placement errors, salt-bridge formation, localized scalp edema, and cephalohematoma, any of which can cause localized attenuation of amplitudes. Less commonly seen is focal high-amplitude slowing, often accompanied by spikes and sharp waves; it is seen more commonly in older premature and term infants. This pattern may correlate with focal cerebral lesions (e.g., hemorrhage, cerebral infarction, and periventricular leukomalacia). However, the sensitivity and specificity of the EEG for diagnosing focal cerebral abnormalities seem to be poor.
Disturbance of Sleep States
Sleep-state differentiation should be readily apparent after 34 weeks CA. EEGs that lack distinct sleep states despite a long period of recording (1 hour) usually occur in infants with encephalopathies from a variety of causes. The background is usually persistently low in amplitude or is excessively discontinuous. It is important to ensure that this lack of change in sleep state is not caused by excessive environmental stimulation in the NICU, hypothermia, toxic factors, or administration of neuroactive medications. If these causes can be eliminated, an EEG that contains no recognizable states generally is associated with a poor prognosis. If the EEG is performed within the first 24 hours, it should be repeated after several days.
Several types of abnormalities of sleep states can be encountered in an EEG. They include the lack of well-developed active and quiet sleep; poor correlation (concordance) between the behaviors characterizing a particular state and the EEG; excessive transient or “indeterminate” sleep; excessive lability of sleep states; and deviations from the normal percentages of specific behavioral states. Much variability exists in the interpretation of these patterns, and few normative data are available. These patterns occur in babies with mild hypoxic or metabolic encephalopathies or subarachnoid hemorrhage, or following complicated pregnancies or deliveries. As the abnormal clinical state resolves, the EEG rapidly returns to normal. These abnormalities are etiologically nonspecific and apparently have little predictive value.
Other Nonspecific Background Disturbances
Many abnormalities are of little prognostic value in isolation but often accompany the severe abnormalities just discussed. These disturbances include excessive amounts of anterior slow dysrhythmia; increased incidence of encoches frontales (frontal sharp transients); excessive amounts of fast background rhythms, particularly beta activity; and a transiently dysmature background. These abnormalities are etiologically nonspecific and have little predictive value.
EEGs are considered dysmature if they contain patterns that are at least 2 weeks immature for the CA. EEG patterns that have been used as benchmarks for determining maturity include the degree of interhemispheric synchrony, number of delta brushes during active and quiet sleep, number of temporal theta bursts, and duration and morphology of the interburst intervals during discontinuous sleep. Normative scales for these patterns have been developed by several investigators however, the determination of dysmaturity usually is performed subjectively. The determination of dysmaturity is made more easily when the infant’s CA is near term. Dysmature patterns that resolve on serial EEGs (transient dysmaturity) are not significant with respect to outcome; however, persistently dysmature patterns on serial EEGs may indicate a poor prognosis (e.g., neurologic sequelae or early death).
A variety of exogenous and endogenous factors may disturb the maturational schedule of the EEG. Seizures may produce a temporary dysmaturity or regression. When dysmaturity results from seizures, it does not consistently correlate with a poor outcome. Dysmature EEG patterns may be caused by prolonged physiologic disturbances, such as chronic lung disease or patent ductus arteriosus, that cause an arrest or delay in brain maturation. Dysmature EEG patterns are often observed in infants with severe bronchopulmonary dysplasia and cystic periventricular leukomalacia.
Biagioni and co-workers obtained EEGs within the first 2 weeks in 63 preterm infants (28 to 34 weeks EGA) and scored the degree of dysmaturity using precise maturational criteria. They found that a normal EEG was associated with a favorable prognosis (in 25 of 26 infants), but a highly dysmature EEG was not necessarily associated with a poor prognosis (only 2 of 9 infants had severe neurologic sequelae). The fact that some infants had a normal evolution despite very dysmature EEGs may reflect transient effects on the EEG of metabolic or circulatory disturbance early in the course of the preterm infant.
Value of Serial EEGs in Newborn Infants
Serial EEG recordings beginning shortly after birth are useful to assess the timing and mode of brain injuries and also may assist in elucidating their pathogenesis in preterm infants. Serial EEGs may help to distinguish between acute and chronic pathologic processes. The former are characterized by EEG findings of acute depression (e.g., increased discontinuity, decreased faster frequencies, and low amplitudes); the latter may consist of dysmature and disorganized EEG patterns. The timing of brain insults may be assessed by considering the stage of EEG abnormalities in relation to the time of birth. For example, serial EEGs in preterm infants may be useful for determining the timing of injury. Hayakawa and colleagues performed serial EEGs and categorized background EEG abnormalities into acute- or chronic-stage abnormalities in infants with cystic periventricular leukomalacia. The timing of injury was judged to be postnatal if the EEG was normal during the early neonatal course but afterward developed acute-stage followed by chronic-stage abnormalities. Insults just before or around birth would result in acute-stage abnormalities during the early neonatal period, whereas antenatal insults would result in chronic-stage abnormalities during this period. In infants whose initial EEG displayed chronic-stage abnormalities, the mean age of cystic degeneration on ultrasonography was 4 days earlier than those with acute-stage abnormalities, presumably because the injury occurred several days before birth.
Serial EEGs in Preterm Infants
The recording of serial EEGs is an important aspect of the evaluation of premature infants. The EEG should be obtained at the time of acute neurologic insult and repeated 1 to 2 weeks later, particularly if the initial EEG is normal or only moderately abnormal. In premature infants, abnormal outcomes have been associated with patterns including isoelectric EEG; positive rolandic sharp waves; a burst-suppression pattern that is distinct from the tracé discontinu seen in normal, early premature infants (see Fig. 4-1 ); excessive interhemispheric asynchrony (see Fig. 4-8 ); persistent and significant (greater than 50 percent) interhemispheric amplitude asymmetries; and an excessively slow background of variable amplitude (10 to 100 μV) that is unresponsive to stimulation and devoid of normal rhythms such as delta brushes (see Fig. 4-7 ). Other less severe EEG abnormalities, such as mild asymmetries and asynchrony for the conceptional age, transient excessively discontinuous backgrounds, and alterations in the amount of background faster rhythms, do not seem to have prognostic value. The worsening of background pattern on serial recordings is associated with higher probability of neurologic sequelae.
Tharp and colleagues prospectively studied all premature infants (birth weight less than 1,200 g) admitted to an NICU and reported that neurologic sequelae occurred in all infants whose neonatal EEGs were markedly abnormal and in the majority of those with two or more moderately abnormal tracings (recorded at weekly intervals). The EEG was more sensitive than the neurologic examination in predicting poor outcome (72 percent vs. 39 percent), whereas both were equally effective in predicting normal outcome. The EEG also proved more sensitive than cranial computed tomography (CT) and ultrasonography in establishing the severity of the encephalopathy. The combination of EEG and ultrasonography may be particularly useful in detecting brain injury in preterm infants. In a recent study, EEG abnormalities (within 72 hours) in conjunction with abnormal ultrasonography detected periventricular leukomalacia with a sensitivity of 94 percent and a specificity of 64 percent.
Maruyama and colleagues evaluated acute EEG background abnormalities in 295 preterm infants (EGA 27 to 32 weeks) on the basis of continuity, frequency spectrum, and amplitude, and graded them on a five-point scale. The EEGs were performed within the first week of life. The maximal grade of the acute background abnormalities correlated with the subsequent development of cerebral palsy (mostly because of periventricular leukomalacia) and its severity, but the presence of significant acute background abnormalities also had a high false-positive rate.
Serial EEGs in Term Infants
Serial EEGs have also been utilized in the assessment of full-term newborns. Takeuchi and Watanabe assessed 173 high-risk, full-term infants with hypoxic-ischemic encephalopathy (defined as an episode of fetal distress or an Apgar score of 5 or less at 1 or 5 minutes after delivery). The severity of the depression of EEG background activity and its persistence correlated with the neurologic outcome. Infants with normal EEGs and with only minimal or mild background depression that disappeared during the first few days of life had good outcomes. By contrast, neurologic sequelae occurred in infants with a major depression of the background at any time (burst-suppression, or nearly isoelectric background); with moderate depression (abnormal tracé alternant , a discontinuous background, or a very-low-voltage irregular pattern) lasting longer than 4 days; or with mild depression present after 9 days. Similarly, in a study of nine term infants with hypoxic-ischemic encephalopathy, Pressler and colleagues found that an early EEG was an excellent prognostic indicator for a favorable outcome if normal within the first 8 hours after birth. The outcome was unfavorable if major background abnormalities persisted beyond 8 to 12 hours. However, an inactive or very depressed EEG within the first 8 hours correlated with a good outcome if the EEG activity recovered within 12 hours.
Zeinstra and colleagues confirmed that serial EEGs are better than a single study performed early. They performed two EEGs, the first 12 to 36 hours after birth and the second at 7 to 9 days in 36 term infants with acute neonatal asphyxia. Several infants with a burst-suppression pattern on the initial EEG showed a significant improvement on the second EEG, and had a favorable outcome. If the first EEG was normal or mildly abnormal, the second EEG did not add substantially to the prognostic value.
Abnormal EEG Transients
Spikes and sharp waves are seen commonly in neonatal EEGs and must be interpreted conservatively. In older infants and children, interictal spikes and sharp waves are signatures of an epileptogenic disturbance. In term and premature newborns, however, such sharp or fast transients may be seen in asymptomatic newborns with a normal outcome. Even when seen in “excessive” amounts, they tend to be relatively nonspecific and do not necessarily imply an epileptogenic abnormality. Unless they are repetitive, periodic, or confined to specific regions, pathologic significance should be assigned with caution. This section focuses on some EEG transients that may have a pathologic significance. The relationship between sharp transients and neonatal seizures is discussed later.
Positive Rolandic Sharp Waves
Positive rolandic sharp waves are moderate- to high-amplitude (50 to 200 μV), surface-positive transients lasting 100 to 250 msec ( Fig. 4-10 ). The morphology is variable: simple, notched, or with superimposed fast rhythms. They may occur in the central regions (C3, C4), either unilaterally or bilaterally, or in the central vertex (Cz) region. Positive rolandic sharp waves correlate with deep white matter lesions, particularly periventricular leukomalacia. They may be an early marker of white matter injury, often preceding the ultrasonographic detection of cystic changes. They may be associated with intraventricular hemorrhage, if there is a component of white matter injury. Positive rolandic sharp waves usually are associated with an unfavorable outcome. However, infants with these sharp waves often have other EEG background abnormalities that may confound determination of their prognostic significance. In studies of premature infants, the occurrence of positive rolandic sharp waves with a frequency exceeding 2 per minute was highly associated with motor disabilities or early mortality.
In late premature infants (CA exceeding 34 weeks), low-amplitude positive rolandic sharp waves that are sometimes difficult to distinguish from the background may be present. They occur in short bursts lasting 3 to 7 seconds, with a repetition rate of 1 to 4 Hz. This type of positive rolandic sharp waves has unclear prognostic significance.
Temporal Sharp Transients
Abnormal temporal sharp transients must be distinguished from normal, usually sharply contoured theta bursts of activity that are seen in the temporal areas during the discontinuous ( tracé discontinu ) background in normal premature infants (26 to 32 weeks CA), with highest incidence between 29 and 31 weeks CA. These temporal theta bursts or “sawtooth” waves are normal patterns and are not associated with seizures.
Positive Temporal Sharp Waves
Positive temporal sharp waves have a morphology and polarity similar to those of positive rolandic sharp waves but occur over the midtemporal regions (T3 and T4). In a study of premature infants (31 to 32 weeks CA) they were seen in approximately half of asymptomatic infants and in three-quarters of children with various disorders (asphyxia, metabolic disorders, and cystic periventricular leukomalacia). Their incidence was higher in the asphyxia group when compared with the asymptomatic group, but not in the other disorders. In the asymptomatic group, their frequency tended to decrease rapidly on subsequent EEGs, whereas they persisted in the pathologic groups. Scher and colleagues found that positive temporal sharp waves were present in a population of healthy, asymptomatic infants, with the peak at 33 to 36 weeks CA. They were uncommon in term infants but persisted in premature infants who had matured to term. These authors postulated that the persistence of these sharp waves in the latter group may represent an electrographic pattern of dysmaturity. Others have found a correlation with hypoxic-ischemic insult in the newborn period. Full-term infants who have positive temporal sharp waves appear to have a high incidence of focal or diffuse structural lesions on neuroimaging studies and a high incidence (80 percent) of other EEG background abnormalities. From this study, the prognostic implication of positive temporal sharp waves is unclear. There is no consistent association with neonatal seizures.
Excessive Frontal Sharp Transients
Frontal sharp transients are physiologic patterns seen in infants between 35 and 45 weeks CA. These sharp waves are seen more abundantly in mild encephalopathies and are absent in severe encephalopathies. Nunes and colleagues found increased density and asynchrony of frontal sharp transients in symptomatic hypoglycemic neonates compared with normal controls. However, it is not clear whether hypoglycemia was responsible because serum glucose concentrations were not recorded around the time of the EEG and the infants had encephalopathies of various etiologies, including asphyxia, hydrocephalus, and sepsis.
Focal periodic discharges and periodic lateralized epileptiform discharges (PLEDs) are pathologic patterns that occur in various encephalopathic conditions, but do not necessarily imply ictal events. The distinction between these two patterns is sometimes difficult to define but may relate to their duration and persistence. Focal periodic discharges consist of broad-based, often biphasic discharges that may occur focally or independently at various locations ( Fig. 4-11 ). They may last from a few seconds to several minutes and sometimes become faster in frequency. The usual lack of evolution in the morphology, frequency, or field of the discharge differentiates them from electrographic seizures. However, focal periodic discharges sometimes may represent a “slow” ictal pattern, often associated with focal clonic seizures.
Focal periodic discharges may occur in a variety of CNS disorders, such as cerebrovascular insults (strokes), hypotension, bacterial meningitis, viral encephalitis, and brain malformation, and in the course of metabolic disorders. Scher and Beggarly studied 34 infants with focal periodic discharges in EEGs recorded in the neonatal period. Cerebral infarction was the most common underlying pathology. Most (75 percent) of these infants had an unfavorable prognosis (death or neurodevelopmental sequelae).
Focal periodic discharges may occur in neonates without clinical seizures. However, they sometimes follow a high-frequency EEG discharge (with focal clinical seizures occurring during the high-frequency discharge). In preterm infants, focal periodic discharges are associated less commonly with electrographic seizures and a demonstrable cerebral lesion.
PLEDs are defined as stereotyped, repetitive, paroxysmal complexes occurring with a regular periodicity (between 1 and 10 seconds). The morphology may be variable: slow waves, sharp waves, or spikes that are bi-, tri-, or polyphasic, lasting 200 to 400 msec. Duration is at least 10 minutes. They often last much longer. PLEDs exhibit no evolution in morphology, frequency, or field, and they are not associated with ictal manifestations. PLEDs have been seen in patients with focal pathology (e.g., cerebral infarcts) or more diffuse encephalopathies (e.g., perinatal asphyxia).
In infants with severe encephalitis, the periodic discharges may be multifocal and often are located in the temporal, frontal, and central regions. The periodic slow waves or sharp slow waves have a periodicity of 1 to 4 seconds and persist throughout the tracing, interrupted only by focal seizures.
Focal periodic discharges and PLEDs may represent a continuum of a similar pathophysiologic process, appearing in brain-injured infants who sometimes exhibit clinically detectable seizures and who usually display other EEG background abnormalities and disorganized states. The prognosis appears poor for either pattern, although it may depend more on the underlying etiology.
Rhythmic Theta-Alpha Activity
The EEG of preterm and term infants sometimes shows excessive amounts of rhythmic theta- or alpha-frequency activity. In a study of term infants (37 weeks CA or older) by Hrachovy and O’Donnell, such patterns were found in a variety of conditions, including congenital heart diseases, congenital brain abnormalities, and hypoxemia, as well as in infants receiving neuroactive medications. They concluded that this pattern was diagnostically nonspecific and may be seen occasionally in infants without overt CNS disease.
Rhythmic theta-alpha bursts , which consist of bursts or short runs (1 to 3 seconds) of sharply contoured, primarily surface-negative, rhythmic theta- and alpha-range activity, are common in the central and frontocentral midline areas, particularly during quiet sleep ( Fig. 4-12 ). These are called wickets or rhythmic sharp theta burst by some authors. Such bursts occur in healthy infants, although some authors believe that they are often seen in infants who have suffered various CNS insults, particularly when clearly defined sharp waves or spikes are intermixed with these bursts. However, in a study of newborns with electrographic seizures, the presence of this activity seemed to have a favorable prognostic value.