Atlas of Neonatal Electroencephalography, 3rd Edition

Chapter 1

Approach to Visual Analysis and Interpretation

The basic principles of visual analysis and interpretation of the electroencephalogram (EEG) that apply to older patients (Kellaway, 2003) also generally apply to neonates, although with some additional special considerations. For example, because of the rapid rate of cerebral development in the neonatal period, the age-dependent features of the EEG become critically important. The interpretation of the neonatal EEG requires the recognition of EEG changes from conceptional ages less than 28 weeks through 44 weeks. In addition, types of abnormalities that are age-dependent also must be identified. Further, because of the special clinical problems of neonates, it is critical to understand how specific etiologic factors may affect cerebral function and, in turn, the neonatal EEG.

Nearly 40 years ago it was pointed out that the characteristics of the EEG known to be normal and, to a lesser degree, abnormal in neonates had not been established (Kellaway and Crawley, 1964); in 2003, the situation has not changed significantly. One problem is that an assumption of normality cannot be made in a newborn with the same degree of confidence as it can in older children. This is related to the brevity of the period of life available for study and the limitations of the neurologic examination. Traditionally, in studies of neonatal EEG, infants have been considered normal if they had no abnormal neurologic signs at birth and were clinically normal at the time of discharge from hospital. Some manifestations of cerebral dysfunction may not become clinically evident until a certain level of brain maturation has been achieved. In addition, to date, systematic, long-term serial studies that correlate neonatal EEG with neurologic, psychological, and behavioral development from birth to adolescence have not been carried out. As a consequence of these limitations, the significance of certain specific features of the EEG has not been established.

Despite these limitations, the imperatives of clinical practice require a clear presentation of the current knowledge of neonatal EEG. The data presented in this atlas reflect more than 50 years of our experience in the study of normal and abnormal neonates. In addition, this atlas reflects the work of other investigators (Blume and Dreyfus, 1982; Clancy, Huang, and Temple, 1993; Dreyfus-Brisac, 1957; Dreyfus-Brisac, 1959; Dreyfus-Brisac, 1962; Dreyfus-Brisac, 1964; Dreyfus-Brisac, 1968; Dreyfus-Brisac, 1970; Dreyfus-Brisac, 1978;Dreyfus-Brisac and Blanc, 1956; Dreyfus-Brisac et al., 1957; Dreyfus-Brisac et al., 1961; Dreyfus-Brisac et al., 1962; Dreyfus-Brisac and Monod, 1970; Ellingson, 1958; Ellingson, 1979;Engel and Butler, 1963; Lombroso, 1979; Lombroso, 1982; Lombroso, 1985; Monod et al., 1960; Monod and Pajot, 1965; Monod et al., 1972; Sainte-Anne-Dargassies et al., 1953; Tharp et al., 1981; Torres and Anderson, 1985; Torres and Blaw, 1968; Watanabe and Iwase, 1972; Watanabe et al., 1974). In this regard, we owe a large debt to the pioneering French group led by Dreyfus-Brisac. The group's early studies of the EEGs of premature infants were conducted in a rich institutional environment that was unique in those days. They provided the basic insights that facilitated consistent progress in the development of knowledge and rational interpretation of the EEG of the newborn.


In the interpretation of the EEGs of children and adults, we believe that the only data that should be considered before visual analysis is initiated are the age and the state of consciousness of the patient (Kellaway, 2003). When interpreting EEGs of neonates, the first of these should not be considered or known, because the determination of the EEG conceptional age (CA) is a critical part of the analysis and assessment of the record. It is recognized that the determination of the


state of consciousness of the neonate may be difficult to determine, but as will be shown later, it may have an impact on the eventual interpretation of the recording (see Chapter 6).

Analysis of the neonatal EEG is initially directed toward the determination of CA based on detection and the recognition of the various developmental features occurring in the record (see Chapter 4). If the EEG lacks recognized developmental features that would permit the determination of CA, then this, in itself, is evidence of significant brain dysfunction. Developmental EEG features may suggest a specific CA, but a discrepancy may exist between the clinically determined CA and the EEG-derived CA; referred to as external dyschronism. This may arise simply from a miscalculation of the clinically determined age, and this is the most likely if the EEG is normal in all other respects. The developmental characteristics of the EEG in deep non-rapid eye movement (REM) sleep may be more immature than those of the EEG awake and in light sleep; referred to as internal dyschronism. In this instance, the most immature features of the deep non-REM sleep findings reflect the age at or before which a cerebral insult may have occurred.

The final step in the process of analysis is the detection and characterization of any abnormal features. These features also may be age-dependent, are described in Chapters 5, 6 and7, and include characterization of background activity and focal features.

As in older children and adults, visual analysis of the EEG of newborns should be an orderly process, involving a series of logical steps that result in the technical analysis and on whichinterpretation is based. Then, and only then, should a correlation be made with the clinical history and findings to derive a clinical impression. Figure 1-1 summarizes the steps of this intellectual process.


The neonatal EEG can be a powerful tool when applied to specific clinical questions (Table 1-1). However, the usefulness of the EEG in these situations will depend on the scope and quality of the information provided by the referring physician and also on the clinical neurophysiologist's understanding of the neurologic disorders of newborn.

What Is the Conceptional Age?

The CA-dependent features of the neonatal EEG are the character of the background activity, the presence of wake-sleep stages, specific waveforms and patterns (collectively referred to as graphoelements), and reactivity (see Chapter 4). Recognition of these features allows the clinical neurophysiologist to determine the CA of an infant within a 2-week epoch. In some clinical circumstances, the CA of the infant is unknown; indeterminate; or based on inconsistent data from maternal history, infant physical examination, or prenatal head ultrasound. However, determination of an accurate EEG-derived CA may add to the understanding of ongoing clinical problems and assessment of the potential risk of future difficulties.

Is There Evidence of Focal Brain Dysfunction?

As in older children and adults, the EEG in neonates may indicate the presence of a consistent focal brain abnormality. Findings such as persistent voltage asymmetries, focal slow activity, and recurrent and persistent sharp waves, either in isolation or in combination, may indicate focal intracranial abnormalities such as subdural fluid effusion, subarachnoid hemorrhage, intracranial hemorrhage, cystic or atrophic lesions, cortical infarction, cerebral malformation, and rarely, a spaceoccupying lesion.

Sharp waves may suggest brain injury that is either focal when sharp waves are persistent and unifocal, or diffuse when sharp waves are multifocal (see Chapters 5 and 6). However, focal sharp waves are rarely indicative of epileptogenicity in the newborn (see Chapter 7) (Mizrahi and Kellaway, 1998). Positive rolandic sharp (PRS) waves were initially associated with the presence of intraventricular hemorrhage (IVH) in the premature infant (Dreyfus-Brisac and Monod, 1964). Subsequently, PRS waves have been shown to be associated with periventricular leukomalacia, a condition that may be a consequence of periventricular brain injury including IVH (Clancy and Tharp, 1984; Novotny et al., 1987). Focal sharp waves that recur in a periodic fashion have been associated with herpes simplex virus encephalitis (Mizrahi and Tharp, 1982; Mikati et al., 1990). It should be noted that the significance of some types of sharp waves in the neonatal EEG is unknown and is discussed further in Chapter 5.

Is Evidence Found of Diffuse Brain Dysfunction?

The neonatal EEG may indicate the presence and degree of diffuse brain dysfunction. Although it may provide evidence of an encephalopathy and its severity, the EEG is less likely to provide information concerning etiology. Thus, similar findings of background activity may be present in infants with hypoxic-ischemic encephalopathy, central nervous system (CNS) infections, bilateral cerebral hemorrhage or infarctions, some metabolic disturbances, and other etiologic factors that may cause diffuse CNS injury.

The characterization of abnormalities of the background activity that describe the continuum of diffuse dysfunction are, from the least to the most severe: depressed and undifferentiated, suppression-burst, and isoelectric (see Chapter 6). The timing of these findings in relation to injury may suggest both severity of the encephalopathy and its prognosis, because some abnormal findings obtained in the acute period after injury may be only transient. Additional findings that suggest diffuse brain dysfunction are those characterized by the term internal dyschronism.


FIG. 1-1. Flow diagram showing the process of visual analysis and interpretation of the neonatal EEG.


TABLE 1-1. Clinical questions that can be addressed by the neonatal EEG

·   What is the conceptional age?

·   Is there evidence of focal brain dysfunction?

·   Is there evidence of diffuse brain dysfunction?

·   When did the brain insult occur?

·   Are there clinical or electrical seizures?

·   What is the neurologic prognosis?

·   Is there an indication that a specific disease entity is present?

When Did the Brain Insult Occur?

No formal studies link EEG findings with the determination of the time of occurrence of diffuse brain injury. However, a number of logical assumptions have been made in the application of EEG in relation to this clinical question. As discussed earlier, internal dyschronism occurs when the developmental features of the EEG in deep non-REM sleep are more immature than those in wakefulness and light sleep and indicates that the infant has had a brain insult. The CA derived from the features in deep sleep indicates that an insult occurred at or before the developmental age associated with those specific features.

Are Clinical or Electrical Seizures Present?

The clinical problem of neonatal seizures is discussed in Chapter 7 and in more detail elsewhere (Mizrahi and Kellaway, 1998). Recording seizures is dependent on the duration of the EEG recording and its timing. Seizures may be classified according to the temporal relation of clinical and EEG events. Electroclinical seizures are those in which clinical and EEG seizure activities overlap in time, typically with close correlation of limb, body, or facial movements to electrical discharges. Electrical-only seizures are those that occur without clinical events. Clinical-only seizures are those without any EEG correlate.

Interictal findings, such as the character of the background, may be helpful in assessing the degree and distribution of CNS dysfunction and may even suggest the degree of risk of seizure occurrence (Laroia et al., 1998). However, interictal focal sharp waves do not provide reliable markers of potential epileptogenesis. Focal sharp waves or even spikes do not always have the same implications in the neonate as they do in older children or adults and therefore may not be considered epileptiform.

What Is the Prognosis?

Some features of the EEG that indicate focal or diffuse injury also may suggest the long-term neurologic prognosis. For example, the frequency of occurrence of PRS waves may predict the occurrence of neurologic sequelae (Blume and Dreyfus-Brisac, 1984), and a severely abnormal background activity also may suggest a poor prognosis. However, the overriding factor in the use of neonatal EEG in determination of prognosis is the evolution of findings over time in sequential studies. A single normal EEG near the time of suspected injury usually predicts a good outcome. An initial EEG with an abnormal background, even when severely abnormal, may, over time, evolve to a less abnormal or even a normal recording, depending on the nature of the brain insult. The rate of resolution, if any, will be the most predictive of outcome, rather than a single EEG at one time.

Is an Indication of a Specific Disease Entity Present?

In general, it is unusual for the neonatal EEG to provide data that will identify specific diseases. Only a few patterns are diagnostic. PRS waves traditionally have been associated with the presence of IVH, but now are more consistently associated with periventricular leukomalacia (Clancy and Tharp, 1984). Herpes simplex virus encephalitis has been associated with the finding of periodic lateralized epileptiform discharges (Mizrahi and Tharp, 1982), although this finding also may be seen in other conditions (Hrachovy et al., 1990). The condition of holoprosencephaly is associated with a specific pattern of rapidly changing background activity (DeMyer and White, 1964). A pattern of periodic hypsarrhythmia in term infants has been associated with nonketotic hyperglycinemia and other inborn errors of metabolism (Aicardi, 1985; Ohtahara, 1978). All of these patterns are discussed in Chapter 6.