27
Cerebral Function Monitoring During the First Week of Life in Extremely Small Low Birthweight (ESLBW) Infants By L. Hellström-Westas1, 1. Rosen2 and N. W. Svenningsen1 Neonatal Intensive Care Unit, Departments of 1Pediatrics, and 2Clinical Neurophysiology, University Hospital, Lund, Sweden
Abstract In order to evaluate the usefulness of early continuous EEG-monitoring in very preterm neonates, recordings with a Cerebral Function Monitor (CFM) were made prospectively in 31 ESLBW infants with birthweights below 901 grams, during their first week of life. The CFM background activity was, as expected from EEG studies, dominated by a suppression-burst pattern in 94 % of the infants. Some infants had periods with more continuous EEG activity or suppression-burst changing into continuous. Patterns similar to sleep-wake cycling (SWC) were identified in infants with gestational ages as low as 24 weeks. The level of the CFM-background activity was mainly influenced by the presence and severity of intracranial hemorrhage (ICH), but also by medications such as phenobarbital. Epileptiform activity (EPA) was only found in infants with ICH, and was identified in 75 % of these infants. Of the infants with EPA 87 % had periods with subclinical EPA, although 47 % had both clinical and subclinical seizures. The presence of more continuous activity and SWC were indicators of a favourable outcome, whereas ele.ctrocerebral inactivity predicted an unfavorable outcome. The prognostic estimates of mortality and neurologic outcome were similar for early CFM recording (positive predictive value 69-100 %) and cranial ultrasound scan (positive predictive value 71-100 %). The monitoring of cerebral electrical activity also provided immediate and clinically useful information during the intensive care of these ESLBW infants. Further studies on the causal relation between EPA and the development of ICH should be performed before definite conclusions can be drawn concerning any preventive effect from anticonvulsive treatment of clinicaVsubclinical seizures.
Keywords EEG-monitoring - Cerebral Function Monitor - Very low birthweight - Preterm - Infants
Received January 28, 1990; accepted February 16, 1990 Neuropediatrics 22 (1991) 27-32 © Hippokrates Verlag Stuttgart
During the last decade the prognosis for ESLBW babies has changed significantly and an increasing number of these infants are now surviving. However, this group of neonates still carries a high risk of developing cerebral complications such as intracranial hemorrhage (ICH) and periventricular or subcorticalleucomalacia (PVL) leading to permanent brain damage. Infants with ICH and PVL also carry a higher risk of developing seizures, both clinical and subclinical (4), which in turn increases the risk of subsequent neurologie sequelae (19). Although brain function is often compromised during the first days of life in siek newborn infants requiring neonatal intensive care, experience with continuous cerebral monitoring is limited. Several studies with conventional electroencephalography (EEG) in both term and moderately preterm infants have shown correlation between EEG features during the first week of life, and later neurological prognosis (20, 26). In an earlier study with CFM-monitoring of preterm infants below 32 weeks of gestation, we observed an early increase of continuous background activity (within 40-60 hours of life) in infants without ICH, in contrast to infants who developed ICH (9). The infants were all treated with phenobarbital and receiving mechanical ventilation. In a later study, with 2-channel tape recorded EEG, Connell et al (5) had similar findings. In the 1980s different electrophysiological techniques for uninterrupted surveillance of brain function have been applied for detection of seizures and for observation of cerebral recovery after birth asphyxia (1, 2, 3, 7, 14). Various methods for displaying modified EEG data in an easily understandable and interpretable way have been developed e. g. selective filtering, spectral analysis and amplitude-integration (8, 23). For clinical use we prefer the Cerebral Function Monitor (2, 13) since it provides an immediately accessible and feasible method for the supervision of siek neonates. The technique is simple and bedside interpretation possible also for neonatologists (for confirmation of CFM-findings, and support with conventional EEG close collaboration with clinical neurophysiologists is recommended). The design with only one channel may of course lead to underestimation of e. g. seizure activity, however this is in many cases counteracted by the simplicity and 24-hour availability of the monitor. The aim of the present study was to investigate the clinical value and characteristics of early postnatal continuous amplitude-integrated EEG-monitoring in very preterm ESLBW neonates, especially in relation to certain neonatal complications and later outcome.
Downloaded by: University of Pennsylvania Libraries. Copyrighted material.
Introduction
N europediatrics 22 (1991)
Methods and material As part of a long-term follow-up program on ESLBW neonates born during a 24-month-period in 1984-86, all infants with birthweights less than 901 g, admitted to our Neonatal Intensive Care Unit (NICU) during their first four days of life, were planned to have continuous monitoring of cerebral electrical activity. The NICU at the University Hospital in Lund, Sweden, is a regional level 111 perinatal unit covering around 16,000 deliveries annually. The patients were both inborn and outborn (some mothers were transferred before the delivery). During this 24-months-period, 42 ESLBW infants were admitted to the NICU, immediately after birth or during their first 96 hours of life. All infants were nursed in incubators and treated according to our routines as previously described (25). The 0-1 year survival rate was 55 % (= 23 of 42). Twenty ofthe 23 survivors were CFM-recorded during their first week of life. The infants were consecutively chosen for EEG-monitoring when the CFM apparatus was available. A few infants were selectively not included - considered either nonviable and very immature or with lethaI malformations. Gf the 11 non-monitored infants five died within 24 hours, and another three within 60 hours of postnatal age. In total 31 of the 42 ESLBW infants were continuously monitored with a Cerebral Function Monitor (CFM) during their first week of life. As shown in Table 1, the gestational ages (GA) ranged from completed 23 to 33 weeks (mean 26.4), according to intrauterine ultrasound during midgestation. Birth-weights (BW) ranged from 540 to 900 grams (mean 772), four infants were small for gestational age (SGA). For sedation during mechanical ventilation, phenobarbital was given to 24 infants, with a loading dose of 10 mg/kg followed by a daily dosage of 2.5-5 mg/kg. Bloodconcentrations of phenobarbital were measured in 13 babies during CFM-registration and were within or just slightly above therapeutical (65-130 Jlmol/l) levels in 10, and between 200 and 288 Jlmol/l in three infants. Diazepam for additional sedation or treatment of seizures was given to nine infants, two of these infants also received lidocaine because of severe seizures. For recurrent apneas theophyllamine or coffeine was given. A loading dose of 5-10 mg/kg theophyllamine or 20 mg/kg coffeine was followed by a daily dosage of 3-6 mg/kg and 10 mg/kg, respectively. Six infants were treated with theophyllamine or coffeine (two infants were monitored with CFM both during theophyllamine and phenobarbital treatment). Three infants did not receive any neuroactive medication. Besides CFM surveillance cranial ultrasound investigations were made repeatedly (2-3 times) during the first week of life in order to detect ICH. Neurodevelopmental followup examinations have presently been made in all survivors at term i. e. expected time of birth (24), 14-16 months and two years of age. The Cerebral Function Monitor (CFM) uses the filtered and selectively amplified EEG-signal, obtained from a pair of bi-parietal electrodes. The CFM has a special filter which fairly sharply attenuates frequencies below 2 and above 15 Hz, giving an amplitude-integrated recording containing the main EEG frequencies but with little disturbances from artifacts. The recording is written out on a heat sensitive paper, on a semilogarithmic scale. The low paper speed of either 6 or 30 cm/ hour, makes it possible to monitor even for prolonged periods
L. Hellström- Westas et al
and to obtain an overview of the cerebro-electrical pattern during the whole registration. The technical design of the CFM, with the semilogarithmic write-out, makes the CFM more sensitive to cerebral activity in the low voltage region, compared to the EEG. The technique is fully described by Prior (23). We used a Cerebral Function Monitor 4640 (Lectromed, Ltd.) in 30 infants, and a Cerebral Function Monitor (Critikon) in one infant. In parallel with the amplitude curve an impedance curve records the reliability of the signal, and shows artifacts from movement, nursing procedures, or loose electrodes. The CFMtracings from the first 168 hours (i. e. first week) of life were independently and blindly investigated by two members of the investigation team, who also were without knowledge of the infants clinical conditions. The CFM recordings were classified as folIows: 1) Main type of background activity, i. e. mainly continuous, or discontinuous with suppression-burst pattern (SB). During SB presence or absence of "reactivity to handling" was estimated, i. e. the CFM-pattern transiently becomes more continuous during handling of the baby e. g. suctioning in the endotracheal tube or diaper changes (Fig. 1 C). The classification of background activity in CFM recordings is based mainly on pattern recognition, hence the comparison between the two independent investigators. However, the basis for the pattern recognition is derived from published studies on CFM in normal neonates (29, 30, 31). No previous study has been published with extremely preterm infants.
A
1h
Fig. 1 Types of CFM-traces in ESLBW infants: A) Saw-tooth pattern of recurrent epileptogenic activity (EPA), without clinical correlate i. e. silent seizures (infant No. 16). B) Continuous cerebral activity, after phenobarbital and mechanical ventilation was started, changing into suppressionburst pattern (infant No. 14). G) Suppression-burst pattern with increased amplitude and reactivity to handling (t) during nursing care (infant No. 10). 0) Sleep-wake cycling (SWC). Bars underline periods with deep sleep (infant No. 14).
Downloaded by: University of Pennsylvania Libraries. Copyrighted material.
28
Cerebra1Monitoring in ESLBW Infants
recordings and constant clinical observation together with recordings of heart-rate and respiratory patterns, we could identify sleep-wake cycling corresponding to certain patterns in the CFM (10). In the present study, cyclical changes similar to these patterns were denotated sleep-wake cycling (SWC) (Fig. 1 D). 3) Presence or absence of electrocerebral inactivity. Electrocerebral inactivity was defined as maximum cerebral activity continuously below 5 J,lVolt for at least two hours. 4) Presence or absence of epileptiform activity (EPA). Patterns with EPA are usually clearly distinguished in the CFM. letal seizure activity in the EEG, with repetitive high voltage activity is corresponded in the CFM by a rather abrupt onset of increase of both maximum and minimum voltage activity, followed by a decrease when the ictal activity ceases. A period with repetitive EPA in CFM (corresponding to status epilepticus or serial seizures) resembles a saw-tooth and has been designated "saw-tooth pattern". The diagnosis of EPA in CFM may in some cases be discussed, especially when not accompanied by clinical symptoms. We have diagnosed as EPA only clinical seizures accompanied by corresponding changes in the CFM, or recurrent CFM-activity with saw-tooth-pattern with or without clinical signs. Solitary seizures without clinical correlates, appearing as single peaks in the CFM record, would therefore not be classified as EPA. Reactivity to handling may in some instances resemble single EPA, with increased voltage. It is therefore essential that all kinds of handling with the patient is notated simultaneously on the CFM recording and impedance curve. 5) The percentage of time during which the minimum EEG voltage was above 3 J,lVolts, here called "% Continuous Activity", or "% CA", was calculated for every 2-hourperiod in the 30 records from CFM-Lectromed. This level was chosen empirically for semi-quantitative estimation of the CFM activity in ESLBW infants. One recording (from the CritikonCFM) could be classified into main type of tracing and presence of EPA but could not be compared when counting % CA.
29
year follow-up, in each case. Fifteen infants (75 %) survived without, and five infants (25 %) with neurologie handicap at 2 year follow-up. The neurological sequelae were spastic diplegia in infants No. 18, 22 and 27; and psychomotor retardation in infant No. 26. One infant (No. 22) had both spastic diplegia and psychomotor retardation as well as a shunt-operated hydrocephalus. Infant No. 27 also had epilepsia and No. 26 had retinopathy of prematurity. Ten infants died in the neonatal period and one at four months of age. In seven infants 10 EEG recordings were made during the CFM registrations. There was complete agreement between all 10 EEG and the CFM recordings. Interobserver agreement between the two CFM-investigators (LHW and IR), was total except for one case of EPA (No. 26). After re-scrutinizing the CFM-record it was decided that this infant had EPA.
CFM-tracings
Different typical CFM patterns from cases No. 10, 14 and 16 are illustrated in Fig. 1. 1. Background activity: In 29 of 31 infants (94 %) suppression-burst (SB) was the initially dominating background pattern. Of these 29 infants, 13 were weIl at 2 year follow-up, 5 had neurologie sequelae and 11 had died. "Reactivity" was identified in 20 of 29 infants (69 %). Among the 29 infants with SB-pattern, "reactivity" was found in 9 of the 13 infants (69 %) who did weIl on follow-up, but also in all 5 surviving with sequelae (total 78 % in all surviving infants). In non-survivors, 6 of the 11 (55 %) showed this reactivity pattern. The difference in "reactivity" between surviving and nonsurviving infants did not reach statistic significance (p :::;: 0.18). Periods with continuous activity was present in 11 of the 31 ESLBW infants (34 %), either intermixed with the SB, or as SB changing into more continuous activity. At 2 year follow-up 10 of these 11 infants were weIl, and only one had neurologie handicap (No. 18). Eight of these infants also had SB-pattern in the CFM (Table 1). 2. Sleep-wake cycling (SWC): was found in 16 ESLBW babies - 11 with gestational age 26 weeks or less, the In 7 infants one or more conventional EEGs most immature being a baby born after 24 weeks of gestation. were made as well. Cranial ultrasound investigations were made In 15 surviving infants healthy at 2 year follow-up 11 had SWC through the anterior fontanelle using a 7.5 MHz transducer (73 %). In contrast only 3/11 (27 %) ofthe infants who died had (Technicare Autosector) and graded according to Papile (21). For statistical calculations Wilcoxon rank sum test for two SWC, and 2/5 (40 %) surviving with sequelae had SWC. The groups and Fisher exact probability test were used. The study difference between healthy survivors versus survivors with neurwas approved by the Ethical Committee of the University of ologie handicap and non-survivors was statistically significant (p:::;: 0.02). Lund and performed with institutional and parental consent. 3. Electrocerebral inactivity: was found in six infants - five died and one survived with severe neurologie Results handicap. However, three of these six infants had simultaneous very high serum concentrations of phenobarbital (200-288 In the 31 ESLBW infants of this study, the J,lmoVI). CFM-registrations were started at median 18 hours (range 4. Epileptiform activity (EPA): EPA was iden1-96) after birth i. e. 19 recordings were started within 24 tified in 75 % (15/20) of infants with ICH but in none of 11 hours, and 8 between 25 and 48 hours of life. The recordings without ICH (p
Cerebral Function Monitoring During the First Week of Life in Extremely Small Low Birthweight (ESLBW) Infants By L. Hellström-Westas1, 1. Rosen2 and N. W. Svenningsen1 Neonatal Intensive Care Unit, Departments of 1Pediatrics, and 2Clinical Neurophysiology, University Hospital, Lund, Sweden
Abstract In order to evaluate the usefulness of early continuous EEG-monitoring in very preterm neonates, recordings with a Cerebral Function Monitor (CFM) were made prospectively in 31 ESLBW infants with birthweights below 901 grams, during their first week of life. The CFM background activity was, as expected from EEG studies, dominated by a suppression-burst pattern in 94 % of the infants. Some infants had periods with more continuous EEG activity or suppression-burst changing into continuous. Patterns similar to sleep-wake cycling (SWC) were identified in infants with gestational ages as low as 24 weeks. The level of the CFM-background activity was mainly influenced by the presence and severity of intracranial hemorrhage (ICH), but also by medications such as phenobarbital. Epileptiform activity (EPA) was only found in infants with ICH, and was identified in 75 % of these infants. Of the infants with EPA 87 % had periods with subclinical EPA, although 47 % had both clinical and subclinical seizures. The presence of more continuous activity and SWC were indicators of a favourable outcome, whereas ele.ctrocerebral inactivity predicted an unfavorable outcome. The prognostic estimates of mortality and neurologic outcome were similar for early CFM recording (positive predictive value 69-100 %) and cranial ultrasound scan (positive predictive value 71-100 %). The monitoring of cerebral electrical activity also provided immediate and clinically useful information during the intensive care of these ESLBW infants. Further studies on the causal relation between EPA and the development of ICH should be performed before definite conclusions can be drawn concerning any preventive effect from anticonvulsive treatment of clinicaVsubclinical seizures.
Keywords EEG-monitoring - Cerebral Function Monitor - Very low birthweight - Preterm - Infants
Received January 28, 1990; accepted February 16, 1990 Neuropediatrics 22 (1991) 27-32 © Hippokrates Verlag Stuttgart
During the last decade the prognosis for ESLBW babies has changed significantly and an increasing number of these infants are now surviving. However, this group of neonates still carries a high risk of developing cerebral complications such as intracranial hemorrhage (ICH) and periventricular or subcorticalleucomalacia (PVL) leading to permanent brain damage. Infants with ICH and PVL also carry a higher risk of developing seizures, both clinical and subclinical (4), which in turn increases the risk of subsequent neurologie sequelae (19). Although brain function is often compromised during the first days of life in siek newborn infants requiring neonatal intensive care, experience with continuous cerebral monitoring is limited. Several studies with conventional electroencephalography (EEG) in both term and moderately preterm infants have shown correlation between EEG features during the first week of life, and later neurological prognosis (20, 26). In an earlier study with CFM-monitoring of preterm infants below 32 weeks of gestation, we observed an early increase of continuous background activity (within 40-60 hours of life) in infants without ICH, in contrast to infants who developed ICH (9). The infants were all treated with phenobarbital and receiving mechanical ventilation. In a later study, with 2-channel tape recorded EEG, Connell et al (5) had similar findings. In the 1980s different electrophysiological techniques for uninterrupted surveillance of brain function have been applied for detection of seizures and for observation of cerebral recovery after birth asphyxia (1, 2, 3, 7, 14). Various methods for displaying modified EEG data in an easily understandable and interpretable way have been developed e. g. selective filtering, spectral analysis and amplitude-integration (8, 23). For clinical use we prefer the Cerebral Function Monitor (2, 13) since it provides an immediately accessible and feasible method for the supervision of siek neonates. The technique is simple and bedside interpretation possible also for neonatologists (for confirmation of CFM-findings, and support with conventional EEG close collaboration with clinical neurophysiologists is recommended). The design with only one channel may of course lead to underestimation of e. g. seizure activity, however this is in many cases counteracted by the simplicity and 24-hour availability of the monitor. The aim of the present study was to investigate the clinical value and characteristics of early postnatal continuous amplitude-integrated EEG-monitoring in very preterm ESLBW neonates, especially in relation to certain neonatal complications and later outcome.
Downloaded by: University of Pennsylvania Libraries. Copyrighted material.
Introduction
N europediatrics 22 (1991)
Methods and material As part of a long-term follow-up program on ESLBW neonates born during a 24-month-period in 1984-86, all infants with birthweights less than 901 g, admitted to our Neonatal Intensive Care Unit (NICU) during their first four days of life, were planned to have continuous monitoring of cerebral electrical activity. The NICU at the University Hospital in Lund, Sweden, is a regional level 111 perinatal unit covering around 16,000 deliveries annually. The patients were both inborn and outborn (some mothers were transferred before the delivery). During this 24-months-period, 42 ESLBW infants were admitted to the NICU, immediately after birth or during their first 96 hours of life. All infants were nursed in incubators and treated according to our routines as previously described (25). The 0-1 year survival rate was 55 % (= 23 of 42). Twenty ofthe 23 survivors were CFM-recorded during their first week of life. The infants were consecutively chosen for EEG-monitoring when the CFM apparatus was available. A few infants were selectively not included - considered either nonviable and very immature or with lethaI malformations. Gf the 11 non-monitored infants five died within 24 hours, and another three within 60 hours of postnatal age. In total 31 of the 42 ESLBW infants were continuously monitored with a Cerebral Function Monitor (CFM) during their first week of life. As shown in Table 1, the gestational ages (GA) ranged from completed 23 to 33 weeks (mean 26.4), according to intrauterine ultrasound during midgestation. Birth-weights (BW) ranged from 540 to 900 grams (mean 772), four infants were small for gestational age (SGA). For sedation during mechanical ventilation, phenobarbital was given to 24 infants, with a loading dose of 10 mg/kg followed by a daily dosage of 2.5-5 mg/kg. Bloodconcentrations of phenobarbital were measured in 13 babies during CFM-registration and were within or just slightly above therapeutical (65-130 Jlmol/l) levels in 10, and between 200 and 288 Jlmol/l in three infants. Diazepam for additional sedation or treatment of seizures was given to nine infants, two of these infants also received lidocaine because of severe seizures. For recurrent apneas theophyllamine or coffeine was given. A loading dose of 5-10 mg/kg theophyllamine or 20 mg/kg coffeine was followed by a daily dosage of 3-6 mg/kg and 10 mg/kg, respectively. Six infants were treated with theophyllamine or coffeine (two infants were monitored with CFM both during theophyllamine and phenobarbital treatment). Three infants did not receive any neuroactive medication. Besides CFM surveillance cranial ultrasound investigations were made repeatedly (2-3 times) during the first week of life in order to detect ICH. Neurodevelopmental followup examinations have presently been made in all survivors at term i. e. expected time of birth (24), 14-16 months and two years of age. The Cerebral Function Monitor (CFM) uses the filtered and selectively amplified EEG-signal, obtained from a pair of bi-parietal electrodes. The CFM has a special filter which fairly sharply attenuates frequencies below 2 and above 15 Hz, giving an amplitude-integrated recording containing the main EEG frequencies but with little disturbances from artifacts. The recording is written out on a heat sensitive paper, on a semilogarithmic scale. The low paper speed of either 6 or 30 cm/ hour, makes it possible to monitor even for prolonged periods
L. Hellström- Westas et al
and to obtain an overview of the cerebro-electrical pattern during the whole registration. The technical design of the CFM, with the semilogarithmic write-out, makes the CFM more sensitive to cerebral activity in the low voltage region, compared to the EEG. The technique is fully described by Prior (23). We used a Cerebral Function Monitor 4640 (Lectromed, Ltd.) in 30 infants, and a Cerebral Function Monitor (Critikon) in one infant. In parallel with the amplitude curve an impedance curve records the reliability of the signal, and shows artifacts from movement, nursing procedures, or loose electrodes. The CFMtracings from the first 168 hours (i. e. first week) of life were independently and blindly investigated by two members of the investigation team, who also were without knowledge of the infants clinical conditions. The CFM recordings were classified as folIows: 1) Main type of background activity, i. e. mainly continuous, or discontinuous with suppression-burst pattern (SB). During SB presence or absence of "reactivity to handling" was estimated, i. e. the CFM-pattern transiently becomes more continuous during handling of the baby e. g. suctioning in the endotracheal tube or diaper changes (Fig. 1 C). The classification of background activity in CFM recordings is based mainly on pattern recognition, hence the comparison between the two independent investigators. However, the basis for the pattern recognition is derived from published studies on CFM in normal neonates (29, 30, 31). No previous study has been published with extremely preterm infants.
A
1h
Fig. 1 Types of CFM-traces in ESLBW infants: A) Saw-tooth pattern of recurrent epileptogenic activity (EPA), without clinical correlate i. e. silent seizures (infant No. 16). B) Continuous cerebral activity, after phenobarbital and mechanical ventilation was started, changing into suppressionburst pattern (infant No. 14). G) Suppression-burst pattern with increased amplitude and reactivity to handling (t) during nursing care (infant No. 10). 0) Sleep-wake cycling (SWC). Bars underline periods with deep sleep (infant No. 14).
Downloaded by: University of Pennsylvania Libraries. Copyrighted material.
28
Cerebra1Monitoring in ESLBW Infants
recordings and constant clinical observation together with recordings of heart-rate and respiratory patterns, we could identify sleep-wake cycling corresponding to certain patterns in the CFM (10). In the present study, cyclical changes similar to these patterns were denotated sleep-wake cycling (SWC) (Fig. 1 D). 3) Presence or absence of electrocerebral inactivity. Electrocerebral inactivity was defined as maximum cerebral activity continuously below 5 J,lVolt for at least two hours. 4) Presence or absence of epileptiform activity (EPA). Patterns with EPA are usually clearly distinguished in the CFM. letal seizure activity in the EEG, with repetitive high voltage activity is corresponded in the CFM by a rather abrupt onset of increase of both maximum and minimum voltage activity, followed by a decrease when the ictal activity ceases. A period with repetitive EPA in CFM (corresponding to status epilepticus or serial seizures) resembles a saw-tooth and has been designated "saw-tooth pattern". The diagnosis of EPA in CFM may in some cases be discussed, especially when not accompanied by clinical symptoms. We have diagnosed as EPA only clinical seizures accompanied by corresponding changes in the CFM, or recurrent CFM-activity with saw-tooth-pattern with or without clinical signs. Solitary seizures without clinical correlates, appearing as single peaks in the CFM record, would therefore not be classified as EPA. Reactivity to handling may in some instances resemble single EPA, with increased voltage. It is therefore essential that all kinds of handling with the patient is notated simultaneously on the CFM recording and impedance curve. 5) The percentage of time during which the minimum EEG voltage was above 3 J,lVolts, here called "% Continuous Activity", or "% CA", was calculated for every 2-hourperiod in the 30 records from CFM-Lectromed. This level was chosen empirically for semi-quantitative estimation of the CFM activity in ESLBW infants. One recording (from the CritikonCFM) could be classified into main type of tracing and presence of EPA but could not be compared when counting % CA.
29
year follow-up, in each case. Fifteen infants (75 %) survived without, and five infants (25 %) with neurologie handicap at 2 year follow-up. The neurological sequelae were spastic diplegia in infants No. 18, 22 and 27; and psychomotor retardation in infant No. 26. One infant (No. 22) had both spastic diplegia and psychomotor retardation as well as a shunt-operated hydrocephalus. Infant No. 27 also had epilepsia and No. 26 had retinopathy of prematurity. Ten infants died in the neonatal period and one at four months of age. In seven infants 10 EEG recordings were made during the CFM registrations. There was complete agreement between all 10 EEG and the CFM recordings. Interobserver agreement between the two CFM-investigators (LHW and IR), was total except for one case of EPA (No. 26). After re-scrutinizing the CFM-record it was decided that this infant had EPA.
CFM-tracings
Different typical CFM patterns from cases No. 10, 14 and 16 are illustrated in Fig. 1. 1. Background activity: In 29 of 31 infants (94 %) suppression-burst (SB) was the initially dominating background pattern. Of these 29 infants, 13 were weIl at 2 year follow-up, 5 had neurologie sequelae and 11 had died. "Reactivity" was identified in 20 of 29 infants (69 %). Among the 29 infants with SB-pattern, "reactivity" was found in 9 of the 13 infants (69 %) who did weIl on follow-up, but also in all 5 surviving with sequelae (total 78 % in all surviving infants). In non-survivors, 6 of the 11 (55 %) showed this reactivity pattern. The difference in "reactivity" between surviving and nonsurviving infants did not reach statistic significance (p :::;: 0.18). Periods with continuous activity was present in 11 of the 31 ESLBW infants (34 %), either intermixed with the SB, or as SB changing into more continuous activity. At 2 year follow-up 10 of these 11 infants were weIl, and only one had neurologie handicap (No. 18). Eight of these infants also had SB-pattern in the CFM (Table 1). 2. Sleep-wake cycling (SWC): was found in 16 ESLBW babies - 11 with gestational age 26 weeks or less, the In 7 infants one or more conventional EEGs most immature being a baby born after 24 weeks of gestation. were made as well. Cranial ultrasound investigations were made In 15 surviving infants healthy at 2 year follow-up 11 had SWC through the anterior fontanelle using a 7.5 MHz transducer (73 %). In contrast only 3/11 (27 %) ofthe infants who died had (Technicare Autosector) and graded according to Papile (21). For statistical calculations Wilcoxon rank sum test for two SWC, and 2/5 (40 %) surviving with sequelae had SWC. The groups and Fisher exact probability test were used. The study difference between healthy survivors versus survivors with neurwas approved by the Ethical Committee of the University of ologie handicap and non-survivors was statistically significant (p:::;: 0.02). Lund and performed with institutional and parental consent. 3. Electrocerebral inactivity: was found in six infants - five died and one survived with severe neurologie Results handicap. However, three of these six infants had simultaneous very high serum concentrations of phenobarbital (200-288 In the 31 ESLBW infants of this study, the J,lmoVI). CFM-registrations were started at median 18 hours (range 4. Epileptiform activity (EPA): EPA was iden1-96) after birth i. e. 19 recordings were started within 24 tified in 75 % (15/20) of infants with ICH but in none of 11 hours, and 8 between 25 and 48 hours of life. The recordings without ICH (p
