Introduction
Since the first human electroencephalography (EEG) recordings in the 1920s, researchers have sought to best understand the significance of these tracings as they relate to the generators (e.g., cortical versus thalamic) of these rhythms and how these may provide insight into brain activity and function in different states (e.g., pathological versus normal and wakefulness versus sleep versus coma). At its core, the voltage fields recorded by the scalp EEG electrodes are the summation of excitatory and inhibitory postsynaptic potentials from pyramidal neurons in the most superficial cortex. The frequencies recorded via scalp EEG that can be visually interpreted (without the assistance of advanced automated software) are limited to a band of frequencies under 30 Hz due to attenuation of recorded signals by skull and interceding tissue layers.
The adult EEG tends to be quite predictable. In polysomnograms (PSGs) there may be mild differences observed in voltages with increasing age, changing distributions of sleep stages, and lower thresholds for arousals, but generally speaking there is not much differentiating the EEG of a 24-year-old from that of a 70-year-old. In stark contrast, there are well-described maturational changes in the EEG in pediatrics; the differences between the EEGs of a full-term newborn and a 6-month-old are profound. The predictable electrographic evolution from the tracé discontinú pattern of an extremely premature neonate to the expected appearance of vertex waves and sleep spindles in infancy reflects the progressive, continued development of the brain from the simplified and smooth gyral pattern of premature neonates to that of a more mature postnatal brain.
This chapter will seek to review the maturational changes of the sleep EEG from prematurity (i.e., <37 weeks gestational age) to term (i.e., 37–42 weeks gestational age) and through infancy and early childhood to the second year of life. To effectively address the stark evolution of the sleep EEG of the extremely premature neonate and a 2-year-old, a broad review of both wake and sleep EEG features is necessary. This will entail a description of core concepts defining the maturing EEG background such as synchrony/asynchrony, continuity/discontinuity, symmetry/asymmetry, and neonatal graphoelements . In this review, key clinical and research correlates to the developmental sleep EEG will be highlighted.
Defining ages and conceptional age
Based on the American Academy of Pediatrics recommendations on age terminology, newborns are considered premature if less than 37 weeks of gestation have been completed, term if 37 to 42 weeks of gestation have been completed, and postterm if they are born after 42 weeks. The term “neonate” refers to a child during the first 28 days after birth and “infant” refers to a child aged 1 to 12 months.
One core concept of neonatal EEG interpretation is establishing the corrected age (CA, sometimes referred to as correct gestational age (GA), conceptional age, postconceptional age, postmenstrual age, or adjusted age). The CA is the estimated GA at birth plus the chronological, actual age (i.e., the number of days and weeks postpartum). The GA is the time elapsed between the first day of the mother’s last menstrual period and day of the neonate’s birth, expressed in completed weeks. Beyond timing of menstrual cycles, fetal ultrasounds and the baby’s postnatal physical examination can be used to modify the GA. A 7-week-old born at 33 weeks’ gestation has a CA of 40 weeks for the purposes of EEG interpretation. Determining the accurate CA is crucial for interpretation of the EEG in neonates as the brain and EEG are expected to develop and mature at a similar rate whether the baby is in utero or ex utero . , For example, the neonate born at 41 weeks’ gestation is expected to share the same EEG features as the 10-week-old neonate born at 31 weeks’ gestation. This example holds true under the assumption that both of these hypothetical neonates are otherwise neurologically normal; there are of course various pathological processes that may negatively influence orderly EEG maturation and development, such as cerebral dysgenesis disorders, hypoxic-ischemic injury, and metabolic and genetic disorders, among others.
Most professionals working with premature infants will continue to use these aforementioned concepts including CA until the chronological age of the child is somewhere around 2 to 2.5 years old, which is luckily within the scope of this chapter. At this age, it is generally felt that most (healthy) premature babies have had an opportunity to “catch up” to age-matched term peers. , Some developmental specialists have estimated a rule of thumb needed to correct the developmental gap from prematurity by multiplying the number of prematurity weeks by 10. For example, a premature infant born at 34 weeks’ gestation (6 weeks early) would be estimated to require 60 weeks (1 year and 2 months) to prospectively catch up to age-matched term-delivered peers.
Practical considerations
Of note, different institutions may utilize different EEG electrode recording montages for the recording of newborns. There are varying opinions on the utility of a full 10–20 EEG montage versus a reduced (“double-distance”) montage for neonatal recordings ( Fig. 1.1 ). The American Society of Clinical Neurophysiology position statement on neonatal EEG recording recommends at least the double distance montage but does not go so far to recommend one montage over the other. The primary argument for the reduced montage relies upon the perspective that the smaller cranial vault is effectively recorded by a reduced complement of electrodes. The contrary concern is that the reduced montage is more likely to miss seizures or be less apt to confidently delineate artifact from ictal recording, which has been borne out in some studies. In institutions that do utilize the double-distance montage, once an infant reaches a CA of approximately 48 weeks, then the standard 10–20 EEG montage is implemented. The double-distance neonatal montage still utilizes a greater array of electrodes than the typical EEG montage recommended by the American Academy of Sleep Medicine (AASM) Manual for the Scoring of Sleep and Associated Events for PSG in children and infants. It is noteworthy that many academic pediatric sleep centers utilize a higher number of electrodes during their PSG recordings, such as midline derivations, than those explicitly recommended by the AASM. Regardless of montage, newborn EEGs have traditionally been recorded and interpreted at “half paper speed” (15 mm/second, or approximately 25–30 seconds per page), that is, a compressed EEG view more similar to what sleep physicians typically review and score on PSG. Once an infant reaches a CA of 48 weeks (i.e., 2 months old), the AASM recommends scoring PSGs based on typical NREM and REM sleep stages.
Premature EEGs (23 weeks to <37 weeks)
It is worthwhile to mention some of the nuances and challenges (both historical and present-day) inherent to the interpretation of premature neonatal EEGs. From a technical perspective, EEG recording is challenging and potentially inadvisable in the extremely premature neonates (i.e., <28 weeks’ gestation) who may have exceptionally fragile and underkeratinized epidermis; the application of EEG electrodes puts these neonates at risk of skin tears and breakdown as well as possibly infection. Thankfully the general indications for EEG monitoring in children at this age are sparse, as seizures are considered generally rare and the baseline biobehavioral states can demonstrate poor reactivity to stimulus. An exceptionally challenging perspective is defining true “normal” in the premature neonate, both in terms of neurological function and EEG activity. Typically neonates who receive any sort of EEG recording have some clinical indication for neurophysiological monitoring, which inherently suggests there is a higher likelihood of neurological abnormality. We have the benefit of more than 50 years of research supporting normative patterns. , , There are several conceptual approaches that may help best understand the early sleep EEG recording: a review of the major background patterns, in particular the expected progression from discontinuity to continuity, and neonatal graphoelements will lay the groundwork to approach our patients’ EEGs based on CA.
Continuity versus discontinuity and synchrony versus asynchrony
There are two essential electrical patterns to recognize in neonatal EEG. Continuity refers to uninterrupted electrical activity with less than 2 seconds of relative voltage attenuation less than 25 μV. Discontinuity is defined by bursts of high-voltage electrical activity separated by interburst intervals (IBIs) of relative voltage attenuation less than 25 to 50 μV for a period of at least 2 seconds. , , , Discontinuity and IBI are a function of age and are best defined in background tracings of tracé discontinú and tracé alternant . Discontinuity may be most easily described as sections of the recording with “on” and “off” periods.
Tracé discontinú (French for “discontinuous tracing”) is a pattern of early prematurity seen up to 30 weeks CA. It is defined by high-voltage (50–300 μV) polymorphic bursts of variable frequencies, often containing spiky or sharply contoured waveforms, separated by dramatically attenuated voltages (<25 μV) in IBIs up to 20 seconds or longer in length ( Fig. 1.2 ). From 30 weeks CA to about 34 weeks CA, tracé discontinú may remain as a marker of quiet sleep.
Tracé alternant (French for “alternating tracing”) is a normal discontinuous EEG pattern in full-term infants representative of quiet (or “N”) sleep. This is characterized by at least three alternating runs of bilateral, symmetrical, synchronous high-voltage (50–150 μV) bursts of 1 to 3 Hz delta activity of 3 to 8 seconds alternating with IBIs of 25 to 50 μV, 4 to 7 Hz, theta activity ( Fig. 1.3 ). Bursts often have polyfrequencies beyond just delta activity. Generally, bursts and IBIs are of similar duration. This pattern becomes apparent around 30 weeks CA, slowly replacing the tracé discontinú pattern. As the CA progresses, this pattern undergoes a transient period of asynchrony between 30 weeks CA and term while IBIs shorten and interburst voltages increase. This pattern persists as a marker of quiet sleep up until about 42 to 44 weeks CA. , , , ,
There are normative values regarding IBI voltages and durations , which help to define an appropriately discontinuous from the excessively discontinuous and dysmature EEG ( Table 1.1 ). While discontinuity is a normal feature (if not the prevailing background) of early neonatal EEGs, EEGs should never be discontinuous beyond 2 months of age (46–48 weeks CA); this would be suggestive of diffuse cerebral dysfunction of nonspecific etiology or, possibly, an inaccurate CA.
| Corrected Gestational Age (Weeks) | Typical Interburst Interval (Seconds) | Maximum Interburst Interval (Seconds) | Voltage of Interburst (μV) |
|---|---|---|---|
| <30 | 6–12 | 35 | <25 |
| 30–33 | 5–8 | 20 | <25 |
| 34–36 | 4–6 | 10 | ∼25 |
| 37–40 | 2–4 | 6 | >25 (typically 50–75) |
| 41–44 | 2–4 | 2–4 | >50 (typically 75–100) |
There is an important caveat regarding discontinuity in that it is expected to be spontaneous and unrelated to specific interventions. It is normal for neonates to have relative voltage attenuation, which may produce a discontinuous-appearing recording during an arousal from sleep or movement and this may best be defined by the concomitant presence of abrupt muscle artifact ( Fig. 1.4 ).
In the extreme perspective of discontinuity is burst suppression. Burst suppression is a markedly abnormal EEG finding of excessively discontinuous background with suppressed (<5 μV) interburst voltages and prolonged IBIs. Burst suppression is invariant, devoid of expected age-appropriate EEG features, absent association to a particular biobehavioral state, and lacking variability and reactivity. The bursts of activity typically contain sharply contoured and spikey-appearing epileptiform discharges. This EEG pattern is associated with severe encephalopathy of nonspecific etiology (e.g., epileptic encephalopathy such as Ohtahara syndrome, severe hypoxic-ischemic encephalopathy) and almost uniformly portends very poor prognosis. ,
The premature neonatal EEG demonstrates a peculiar pattern regarding synchronization of bilateral hemispheric electrical activity. Asynchronous bursts are defined as interhemispheric bursts that are separated by more than 1.5 to 2 seconds of each other. Very early neonatal EEGs (<30 weeks) demonstrate entirely synchronous (often called “hypersynchronous”) bursts in their tracé discontinú background. , , , Around 30 weeks CA the EEG bursts become notably asynchronous. These asynchronous bursts become progressively more synchronous again such that approximately 70% of bursts are synchronous by 34 weeks’ gestation and up to 100% of bursts are synchronous at 40 weeks’ gestation. Some very mild asynchrony may persist in normal neonates until the pattern completes abates to the more continuous sleep backgrounds. While this will be discussed in greater detail later, it is important to note that certain developmental sleep features in infancy through early childhood, such as sleep spindles and vertex waves, are expected to be initially asynchronous and eventually become more synchronous.
Normal continuous background patterns of neonates and defining biobehavioral states
In contrast to the discontinuous tracings of early neonates, an increasingly continuous background develops as neonates approach term. A continuous EEG tracing has no recognizable pauses in activity and there are three primary patterns defined in neonates , , :
- 1.
Low voltage irregular (LVI): The LVI pattern is defined by continuous low voltage (∼15–35 μV) predominantly theta > delta frequencies. This is observed in active (“R”) sleep and wakefulness. See Fig. 1.5 for the LVI pattern observed in active sleep.
Fig. 1.5
Low voltage irregular pattern observed in active sleep.
- 2.
High-voltage slow (HVS): The HVS pattern is defined by continuous, synchronous, and symmetrical high-voltage (∼50–150 μV) delta activity. This activity may be highest amplitude in occipital or central regions. This pattern is observed predominantly in quiet (“N”) sleep, although it may rarely occur in active sleep. The tracé alternant evolves into the HVS pattern as it becomes less discontinuous/more continuous. This pattern is sometimes referred to as a continuous slow wave sleep (CSWS) pattern and is the predecessor of the more mature slow-wave/stage 3 NREM sleep. See Fig. 1.6 for the HVS pattern.
Fig. 1.6
High-voltage slow pattern observed in quiet sleep.
- 3.
Mixed (M): The M pattern is an intermingling of high-voltage delta (albeit of lower-voltage than the true HVS pattern) and lower-voltage polyfrequencies. This pattern may be observed in any biobehavioral state, although it is most frequently representative of wakefulness and active sleep. This pattern is often still referred to by many EEG readers to its French terminology— activité moyenne (“average activity”). See Fig. 1.7 for mixed frequency EEG in active sleep.
Fig. 1.7
Mixed frequency EEG in active sleep.
As is rather apparent in their definitions, these are somewhat nonspecific EEG patterns to state (compared to the expected biobehavioral state of an infant with sleep spindles). It is essential to combine multiple observations to fully assess state as some of these electrophysiological recordings are not standard on every EEG recording (although may be available dependent on the institution). These are the inherent features of a good PSG recording, including visual behavioral assessment, respiratory patterns, EEG, electrooculogram (EOG), and chin electromyogram (EMG). Table 1.2 shows a summary of state characteristics for scoring infants less than 37 weeks CA on PSG :
| Stage | EEG Patterns | Behavior | Respiration | EOG | Chin EMG |
|---|---|---|---|---|---|
| Wake | LVI or M | Eyes open, crying, feeding | Irregular, rapid, shallow | Blinks, rapid or scanning eye movements | Present |
| Quiet Sleep/N | Tracé alternant , HVS, rarely M | Eyes closed, very few movements, some periodic sucking | Deep and regular | No eye movements | Present or may be lower than wake |
| Active Sleep/R | LVI or M | Eyes closed, random small movements (e.g., squirm, grimace) | Irregular, occasional pauses | Rapid eye movements or no eye movements | Low with brief transient phasic EMG bursts |
To add to the confusion of summating multiple parameters to assess sleep state in neonates and infants, there exist definitions of transitional (“T”) and indeterminate sleep. Transitional sleep is, intuitively, a period of transition from one sleep state to another and includes elements of both active and quiet sleep (i.e., a combination of either 2 N and 3 R features or 2 R and 3 N features). Indeterminate sleep is defined by the American Clinical Neurophysiology Society (ACNS) as a state of sleep in which the behavioral state is supportive of sleep (i.e., eye closure) but there is a lack of anticipated features to assign a specific sleep state. The AASM does not recommend scoring of indeterminate sleep and encourages assigning a specific stage of N, R, or T. , , Typical sleep cycling in newborns occurs in approximate 60-minute cycles with initial sleep entry into the active/REM sleep state up through the first 2 to 3 months of life. Newborns tend to have an equal distribution of active and quiet sleep, although the active/REM sleep component slowly decreases with age such that by age 5 years, children have the general composition of sleep architecture and REM distribution of adults.
Neonatal graphoelements
This overview of neonatal graphoelements is helpful in the identification of normal and abnormal EEG features as well as concepts that may provide clues to a wake/sleep EEG epoch. This will not, however, represent an exhaustive review of all neonatal graphoelements. The reader is referred to various pediatric and neonatal EEG atlases for discussion on non–state-defined developmental graphoelements such as temporal sawtooth waves, rhythmic occipital theta activity, and centrotemporal delta activity. , ,
Delta brushes (“ripples of prematurity” or “beta-delta complexes”) and monorhythmic occipital delta
Delta brushes are composed of a combination of a delta frequency transient with superimposed 8 to 22 Hz, typically beta, frequency activity. These tend to be symmetrically represented between the two hemispheres and appear in awake and sleeping infants. These are primarily observed in posterior brain regions and rarely frontally ( Fig. 1.8 ). These first appear in the central/rolandic regions at 26 to 28 weeks CA and peak at 32 to 34 weeks, at which point they are most prominent in occipital electrodes.
Delta brushes may occur in synchronous bilateral hemispheric runs of monorhythmic occipital delta activity (MROD). This is another neonatal graphoelement defined by stereotyped runs of monomorphic high-amplitude 0.5 to 1 Hz delta waves in the occipital regions lasting 2 to 60 seconds. This MROD element appears as early at 23 to 24 weeks, peaks between 31 and 33 weeks, and should entirely abate by 35 weeks. After 34 weeks CA, the delta brushes become a prominent pattern in tracé alternant (i.e., quiet sleep) and are rarely observed in active sleep or wakefulness. Delta brushes are rare by term gestation and should be absent by 44 weeks CA.
Anterior dysrhythmia and encoches frontales
Anterior, or frontal, dysrhythmia is a paroxysm of frontally dominant 50 to 150 μV semirhythmic delta (usually 2–4 Hz) activity which may have some subtle evolution over several (generally <6) seconds to acquire a frontal sharp transient morphology. Although the term “dysrhythmia” may appear to suggest abnormality, this is a normal neonatal graphoelement that arises symmetrically and synchronously between frontal regions in any behavioral state but is most obvious in transition from active to quiet sleep. This tends to appear at 32 weeks CA and persists until about 40 to 44 weeks CA.
Encoches frontales are high amplitude (>150 μV) frontal sharp wave transients with biphasic morphology (negative to positive) best observed in the frontopolar electrodes in symmetric, bilateral, synchronous solitary bursts or admixed in runs of anterior dysrhythmia, again showing a predilection toward the transition from active to quiet sleep and occurring most often in quiet sleep ( Fig. 1.9 ). These may appear as early as 26 weeks but may be asynchronous at inception. They are maximal between 34 and 35 weeks and tend to abate after about 44 to 48 weeks CA.
