Electrographic seizures in neonates correlate with poor neurodevelopmental outcome

Subjects.

In August 1994, we defined at-risk criteria for neonatal seizures based on criteria suggested by others1,4,7,10,14 and on our experience with a group of infants with asphyxia.16 The criteria for EEG monitoring were as follows:

• Presence of conditions often associated with seizures

• 1. Birth asphyxia (Apgar score ≤5 at 5 minutes and/or pH < 7.2 in cord blood or in the first hour, and continuing lethargy/obtundation in the first 2 hours

• 2. Meningitis/encephalitis (defined by CSF cell counts or cultures)

• 3. Symptomatic sepsis (ill-appearing or shocky infant with positive blood culture)

• 4. Congenital cerebral anomalies (defined by neuroimaging)

• 5. Grade III or IV intraventricular hemorrhage (if < 27 weeks gestational age)

• 6. Encephalopathy/altered mental status

Clinical suspicion of seizures

• 1. Clinical seizures or movements suggestive of seizures

• 2. Apnea of unknown etiology in full-term infant

Muscle paralysis

• 1. Treatment with pavulon or other muscle paralytics, making it impossible to recognize clinical seizures in an infant with suspected neurologic insult

For the next 2 years, infants brought to our attention who met these criteria underwent continuous EEG monitoring for at least 18 hours. If no ESz were identified, monitoring was stopped. If ESz were identified, the infants were treated with anticonvulsant therapy and the monitoring was continued until no further ESz were identified for 24 hours. After determining that the background of the initial EEG recording in these at-risk infants was highly sensitive in predicting which infants would have ESz,17 since September 1996 continuous EEG monitoring has been instituted for only those at-risk infants with EEG background abnormalities or those already demonstrating ESz during the initial EEG recording. The study group comprises 40 infants who had ESz recorded between August 1994 and September 1997. The control group with regard to outcome comprises 28 infants who met the at-risk criteria and were monitored between August 1994 and September 1996 but found to have no ESz.

Not all infants who met criteria for monitoring were actually monitored. We have previously reported the reasons that 39 infants who met at-risk criteria in the first 2 years were not monitored17: 18 (gestational ages 24 to 39 weeks) were critically ill and too unstable to be monitored (14 of these died within 24 hours of birth); eight infants with gestational age ≤27 weeks and grade III/IV intraventricular hemorrhages were not monitored to avoid the stress and handling of electrode application; eight were missed because we were not notified in a timely manner that they were potential candidates; and four with possible clinical seizures were monitored for less than 18 hours because no ESz were identified, they appeared well, and their caretakers wanted the monitoring stopped. The monitoring record of an additional infant was lost before it had been analyzed.

During the period of study, approximately 1,200 newborns were admitted per year to our neonatal intensive care unit (NICU); 20% of them were outborn. Therefore, approximately 1% of our admissions who met criteria were stable enough to have monitoring and were found to have ESz.

Continuous EEG monitoring procedures.

EEG recording was started as soon as possible after an infant met at-risk criteria. Nineteen scalp electrodes were fixed with collodion and filled with electrode gel in the positions defined by the International 10–20 system. A 21-channel Nihon Kodon (Irvine, CA) machine was used to record 16 channels of EEG data in a standard bipolar montage and five extracerebral monitoring channels, including eye movements, electrocardiogram, nasal thermistor, and thoracic (respiratory) movements. A standard recording was run at a paper speed of 30 mm/sec for 20 to 30 minutes. The paper speed was then reduced to 5 mm/sec, and EEG monitoring continued until 18 to 24 hours without ESz had been recorded, or until electrodes had to be disconnected for the infant’s care. Occasional interruptions in monitoring occurred either for technical reasons or because of clinical needs of the infant (e.g., ultrasound, CT scan). A standard 20- to 40-minute recording at a paper speed of 30 mm/sec was obtained at the conclusion of continuous monitoring.

Analysis of the EEG and electrographic seizures.

The background activity of the initial EEG, the continuous monitoring, and the final EEG were analyzed by a board-certified electroencephalographer (MM), and a background grade was assigned according to previously defined criteria (table 1).17

According to our standard practice during the period of this study, the continuous monitoring records were reviewed visually by the authors to determine the presence of ESz, both at the bedside when seizures were frequent and in sections of the accumulated paper recording at intervals determined by the frequency of the seizures and the availability of the authors. Information obtained about ESz was used to guide treatment of the seizures. When present, the authors, each infant’s nurse, and the EEG technologist wrote their observations directly on the EEG paper while seizurelike movements were occurring. Video recordings were not obtained.

For the purpose of this study, the following data were retrospectively tabulated from all of the monitoring records containing ESz: time of onset, location, duration, and clinical correlates of each ESz; and time of onset and offset, and total duration (defined as the combined duration of all epochs) of status epilepticus. ESz were defined as abnormal rhythmic, stereotyped wave forms evolving in time and space and lasting for at least 10 seconds.18 Status epilepticus was defined as at least 30 minutes of recording during which focal or widespread electrographic seizure activity occurred at least 50% of the time.19 If gaps occurred in the recording, the number and duration of seizures were estimated for that period based on the number and duration of seizures occurring before and after the gap.

Anticonvulsant therapy.

During the first 2 years of the period covered in this analysis, infants were treated initially with either lorazepam (up to three doses of 0.1 mg/kg IV in an 8-hour period, at least 15 minutes apart) or phenobarbital (20 mg/kg followed by 10 mg/kg IV at intervals of at least 1 hour up to a total of 40 mg/kg). If lorazepam failed to control the ESz, phenobarbital was given and vice versa. If seizures continued despite these two drugs, 20 mg/kg of phenytoin was given IV. In the third year, treatment was initiated with phenobarbital. Then, if ESz continued despite 40 mg/kg of phenobarbital, 20 mg/kg of phenytoin was given. Anticonvulsant therapy was initiated by the infant’s caretakers before monitoring began if clinical seizures were obvious or frequent or interfered with the infant’s stability. Once monitoring began, therapy was given for ESz, whether or not clinical seizures continued. ESz were deemed uncontrolled if they persisted or recurred longer than 2 hours after any dose of AED.

Neurodevelopmental outcome.

Infants admitted to the NICU at the Children’s Hospital at Strong take part in a follow-up program that includes two visits to the follow-up clinic in the first year of life or tracking questionnaires sent out to parents and primary care physicians. Information from this follow-up program, notes in the hospital chart for subsequent visits or admissions, and child neurology clinic notes were reviewed retrospectively. Additionally, with the approval of our Human Studies Committee, parents or guardians of the infants were contacted by phone and asked the questions from sections E, H, L, and P of a questionnaire designed to assess the prevalence of disabilities in the general population.20 These sections deal with neurodevelopmental milestones and medications, hospitalizations, or special services the child has received. If parents agreed, the child’s primary care provider was contacted for specific information regarding development and growth data, including head circumference and body weight.

Based on all clinical information obtained, each infant was assigned a degree of cerebral palsy (CP) as defined by Palisano et al.21 Grades of CP severity were as follows:

• Grade I

• Walks without restriction; limitations in more advanced gross motor skills

• Before age 2 years, infants move in and out of sitting, crawl on hands and knees, pull to a stand, and learn to walk without assistive devices between 18 months and 2 years

• Grade II

• Walks without assistive devices; limitation walking outdoors and in the community

• Before age 2 years, infants maintain floor sitting, but may need to use their hands for support to maintain balance; they creep on their stomachs and may crawl and may pull to a stand and cruise holding on

• Grade III

• Walks with assistive devices; limitations walking outdoors and in the community

• Before age 2 years, infants maintain floor sitting when the low back is supported; they roll and creep forward on their stomachs

• Grade IV

• Self-mobility with limitations; children are transported or use power mobility outdoors and in the community

• Before age 2 years, infants have head control, but trunk support is required for floor sitting; they can roll to supine and may roll to prone

• Grade V

• Self-mobility is severely limited, even with the use of assistive technology

• Before age 2 years, infants are unable to maintain antigravity head and trunk postures in prone and sitting positions, and require adult assistance to roll

Head circumferences and weights were plotted to determine if the infant had developed microcephaly (head circumference below the third percentile) or failure to thrive (weight below the third percentile). The occurrence of seizures after discharge from the nursery, the age at independent walking, and scores on the Bayley scale in the first year of life were noted when available.

Statistical analysis.

Demographic characteristics were compared for infants with and without ESz by χ² test for dichotomous variables (or by the Fisher exact test if the content of any cell was 5 or less) or by the Wilcoxon ranked sum test for continuous variables. Characteristics of neurodevelopmental outcome were compared by risk ratios with 95% CI for infants with and without ESz. Where risk ratios are calculated, the p value given is for the risk ratios. Receiver operating characteristic (ROC) curves were calculated to determine three groups for number of ESz and for cumulative time in ESz. The impact of the amount of ESz activity versus the various outcome variables was assessed by logistic regression for these three groups.

Patient and monitoring demographics.

The infants who had ESz (n = 40) and those who did not (n = 28) did not differ in regard to reasons for monitoring and other baseline characteristics (table 2). The primary cause of ESz in the 40 infants, as deemed by clinical judgment after assessing their entire NICU course, included asphyxia (n = 23), stroke (n = 7), and other causes (intraparenchymal bleeding, n = 2; subdural and subarachnoid bleeding, n = 2; meningitis, n = 1; sepsis, n = 1; hyponatremia secondary to renal infarction, n = 1; unknown, n = 3). The reason for monitoring and the final diagnosis for those with ESz are not the same because at the time of initiation of monitoring, the primary reason for the suspected seizures may not have been known. There were no differences between the hours of age at start of monitoring or the percentage of infants who had received AED before monitoring between the infants with and without ESz.

Infants who met our at-risk criteria for asphyxia were not found to have differences in Apgar score at 5 minutes (median 4, range 0 to 6 with ESz; median 3, range 1 to 7 with no ESz); initial pH, defined as pH of cord blood or pH within the first hour of life (median 7.01, range 6.58 to 7.33 versus median 7.05, range 6.57 to 7.24); or initial base excess (median −18.5, range −30 to −6 versus median −12.8, range −20 to −9) as a function of the presence or absence of ESz.

Characteristics of EEG background.

The EEG background was abnormal (EEG background grade 3 to 5) in 92% of the at-risk infants with ESz and 18% of those without (p < 0.001). Fifteen infants (14 with asphyxia and one with sepsis) had no indication of sleep/state cycling or change in state at any time during their monitoring.

Characteristics of electrographic seizures.

ESz occurred within 1 hour of the onset of the recording in 26 (65%), within 6 hours in 11 (28%) and within 16 to 21 hours in the remaining three (7%) infants. Recording time after the last recorded seizure was at least 10 hours (median 24 hours) in 33 of the 40 infants. In the remaining infants, recordings were stopped within 1 hour (n = 3) or 3 to 6 hours (n = 4) of the last recorded ESz because of other clinical needs of the infants.

The infants with ESz spent 8 minutes to 30 hours (median 2 hours) cumulative time in ESz; 31% spent more than 4 hours in ESz. Additionally, 17 (43%) of the infants with ESz spent from 38 minutes to 32 hours (median 3 hours) in electrographic status. Because 65% of these infants were having ESz when the recording began and 7% were still seizing when the recording ended, these figures underestimate the total time spent seizing or in status. In eight infants, none of the ESz were detected clinically; in 31 infants, clinical correlates were noted by the team for 1% or fewer of their ESz. In one infant with idiopathic seizures, all 14 ESz were associated with clinical correlates.

The ESz of each infant were stereotypic in location, frequency, and waveform at onset and during evolution. The ESz usually started in the location that would have the maximum amplitude of rhythmic activity when the electrical activity spread to adjacent areas. For infants with sleep state cycling, there was no clear preponderance of ESz within quiet or active sleep.

Infants with asphyxia tended to have two to four patterns of ESz, occurring bilaterally in different locations, often in the central or temporal regions. Sometimes one of these repeating patterns would start before another stopped or one evolved into another, occasionally involving all 19 electrodes. In 19 of 20 asphyxiated infants whose monitoring continued for at least 6 hours after the last recorded seizure, ESz had stopped by 73 hours of age (median 47 hours).

The infants with strokes (without asphyxia) had ESz isolated to the area of their stroke, and all had preservation of sleep state cycling throughout the recording. The presence of persistent focal ESz in an area of voltage suppression, associated with preservation of sleep state cycling, was seen only in infants with strokes. The last recorded ESz in infants with strokes was at 29 to 131 hours of age (median 49 hours).

Infants with ESz related to asphyxia had many more seizures and longer cumulative time in seizure than infants with ESz related to stroke. There was no difference between these two groups in the average length of ESz (table 3). There tended to be more variability in length of ESz among infants with asphyxia versus those with stroke.

Infants without sleep state cycling in their EEG background had more ESz, spent more time in ESz, were more likely to have status, and were more likely to spend more time in status than those who developed cycling at some time in their recording (table 4).

Response to AED therapy.

The ESz in these neonates were only partially responsive to standard AED. Six infants (15%) stopped having ESz after one to four doses of lorazepam. Only one (3%) of the remaining 33 infants had ESz stop after 20 mg/kg of phenobarbital. An additional 10 infants (25%) had no further ESz after receiving 30 or 40 mg/kg of phenobarbital. ESz in 11 infants (27%) stopped after 40 mg/kg of phenobarbital and 20 mg/kg of phenytoin. Twelve infants (30%) continued to have ESz despite full loading doses of both phenobarbital and phenytoin. No infant required intubation or pressors owing simply to AED therapy. The response of the ESz according to etiology of the seizures is shown in table 3.

In most infants, as additional doses of AED were given, the ESz abated for several minutes to 1 or 2 hours but then resumed. ESz became shorter, more confined, and lower in amplitude before stopping, but they preserved their characteristic onset, frequency of the rhythmic discharges, and pattern of evolution, with some foci disappearing before others. Most of the ESz associated with clinical correlates occurred in the early stages of treatment.

Neurodevelopmental outcome.