Hypoxic-ischemic encephalopathy has an incidence of

Transcription

1 Original Article Amplitude-Integrated EEG Is Useful in Predicting Neurodevelopmental Outcome in Full-Term Infants With Hypoxic-Ischemic Encephalopathy: A Meta-Analysis Journal of Child Neurology Volume 22 Number 9 September Sage Publications / hosted at R. Edwin Spitzmiller, DO, Tonya Phillips, MD, Jareen Meinzen-Derr, PhD, and Steven B. Hoath, MD Hypoxic ischemic encephalopathy is a common cause of neurological complications resulting in chronic handicapping conditions, such as cerebral palsy. Amplitude-integrated electroencephalography (EEG) has been used in many European countries for more than a decade in the evaluation of infants with hypoxic ischemic encephalopathy but has not been widely used in the United States. The objective of this study was to evaluate the evidence supporting use of amplitude-integrated EEG as a quantitative predictor of neurodevelopmental outcome in full-term infants with hypoxic ischemic encephalopathy. To assess efficacy, the authors performed a meta-analysis of the literature evaluating the use of the amplitude-integrated EEG or cerebral function monitor in full-term infants with hypoxic ischemic encephalopathy and their neurodevelopmental outcome. A total of 8 studies were eligible for the primary meta-analysis. There was an overall sensitivity of 91% (95% CI 87-95) and a negative likelihood ratio of 0.09 (95% CI ) for amplitude-integrated EEG tracings to accurately predict poor outcome. Amplitude-integrated EEG is a valuable bedside tool for predicting long-term neurodevelopmental outcome in term infants with hypoxic ischemic encephalopathy. This information is useful in structuring communication and care plans for physicians and parents. Early assessment techniques such as amplitude-integrated EEG provide objective means for determining inclusion in clinical studies evaluating therapies for hypoxic ischemic encephalopathy and for predicting which patients are most likely to respond to treatment. Keywords: amplitude-integrated EEG; full-term newborn; hypoxic ischemic encephalopathy; meta-analysis Hypoxic-ischemic encephalopathy has an incidence of approximately 1 to 6 out of 1000 full-term infants. 1-4 Of the infants with encephalopathy, 15% to 20% die during the newborn period and an additional 25% have permanent neurological deficits. Hypoxic ischemic encephalopathy is one of the most common causes of neurological complications resulting in chronic handicapping conditions, the most common of which is cerebral palsy. 5 The Centers for From the Department of Pediatrics, Divisions of Neonatology (RES, SBH), Neurology (TP), Epidemiology and Biostatistics (JM-D), and the Skin Sciences Institute (SBH), University of Cincinnati and the Cincinnati Children s Hospital Medical Center, Cincinnati, OH. None of the authors (RES, TP, JM-D, and SBH) has a commercial entity that has an interest in the subject of this manuscript. Address correspondence to: R. Edwin Spitzmiller, DO, Pediatrix Medical Group of Ohio, One Wyoming Street, Dayton, OH 45409; Edwin_Spitzmiller@pediatrix.com. Spitzmiller RE, Phillips T, Meinzen-Derr J, Hoath SB. Amplitudeintegrated EEG is useful in predicting neurodevelopmental outcome in full-term infants with hypoxic-ischemic encephalopathy: a meta-analysis J Child Neurol. 2007;22: Disease Control and Prevention estimates the total lifetime costs of cerebral palsy to be around $11.47 million. 6 Hypoxic-ischemic encephalopathy is diagnosed based on a combination of signs of intrauterine distress and abnormal postnatal findings. Intrauterine distress may be evidenced by late decelerations on fetal electrocardiogram monitoring and/or meconium-stained amniotic fluid. Postnatal findings include Apgar scores of 5 at 5 minutes, arterial cord ph 7.1, delayed spontaneous respirations, multiorgan failure, base deficit > 15 or various combinations of the above. Hypoxic ischemic encephalopathy is classified based on the above laboratory, physical, and electroencephalographic (EEG) findings in a well-known system first described by Sarnat. 7,8 The prognostic value of the scoring system and the EEG is most useful in those infants with mild or severe encephalopathy. 8 The amplitude-integrated electroencephalogram is a cerebral function monitor first introduced in Lund, Sweden, more than a decade ago and is currently used primarily in European countries. Amplitude-integrated EEG is used to monitor infants who are paralyzed for mechanical ventilation, 1069

2 1070 Journal of Child Neurology / Vol. 22, No. 9, September 2007 severely asphyxiated infants, infants with intraparenchymal hemorrhages, and infants with seizures. These infants are monitored for hours to days, depending on their clinical condition. 9 The amplitude-integrated EEG is not widely used in the United States of America at this time. To our knowledge, there is no meta-analysis of the prognostic value of the amplitude-integrated EEG in full-term infants with hypoxic ischemic encephalopathy and their neurodevelopmental outcome. In this article, we review the literature regarding the use of the amplitude-integrated EEG in full-term infants with hypoxic ischemic encephalopathy and test the hypothesis that amplitude-integrated EEG is predictive of neurodevelopmental outcome. Methods Search Strategy All studies of the amplitude-integrated EEG and its use in the prediction of neurodevelopmental outcome were identified by a literature search of PubMed and Medline. The search was limited to human studies and clinical trials performed between January 1966 and February Using the search fields of the abstract, MeSH subject heading, exploded subject heading, publication type, subject heading word, text word, and title, the search strategy involved the following keywords: amplitude integrated EEG, cerebral function monitor, hypoxic-ischemic encephalopathy, neurodevelopmental follow up, clinical trial, full term infant, and electroencephalography. Articles were initially identified based on their title and references, and bibliographies from identified studies were also searched. The outcomes of interest primarily included an abnormal neurological exam as performed by a pediatric neurologist at the time of inclusion, at discharge, and at regular intervals after discharge from the neonatal intensive care unit. At the time of inclusion, hypoxic ischemic encephalopathy was defined according to Sarnat. 7 Cerebral palsy as defined by Hagberg, 10 or documented developmental quotient 85 on Griffiths Developmental Scale performed at follow-up, was considered abnormal. The primary investigator (RES) identified all published studies in peer-reviewed journals and abstracts submitted at national conferences reporting amplitude-integrated EEG and hypoxic ischemic encephalopathy and follow-up. The full text of the article was reviewed in detail, and a decision to include the study was made by the primary investigator. A second reviewer (TP) then reviewed the articles in detail. Discrepancies were discussed, and a final decision was made regarding which studies to include. Quality Assessment The quality of the trials was evaluated for blinding of the individuals interpreting the tracings to the clinical condition of the infant, the rapidity at which the amplitude-integrated EEG was applied after birth, the scales with which neurodevelopment follow-up was assessed, and the consistency of abnormal development definitions used by the various authors. A data collection form (Table 1) was used to aid extraction of relevant information on clinical outcomes, timing of the amplitude-integrated EEG, and the number of infants studied. Statistical Analysis The data from each of the studies were extracted as 2-by-2 tables. From these tables, we conducted our primary analysis, computing sensitivities (true positive rate), specificities (true negative rate), and likelihood ratios, which incorporate both sensitivity and specificity. The likelihood ratio for a positive result states how much the odds of the disease/ outcome increase when the test is positive, whereas the likelihood ratio for a negative result states how much the odds of the disease/outcome decrease when the test is negative. All of the tests were reported with 95% confidence intervals. Confidence intervals for sensitivities and specificities were determined based on the confidence interval for proportions, and the confidence interval for the likelihood ratios used methods presented by Simel and colleagues. 11 We used a meta-analysis to pool results for the likelihood ratios utilizing a random effects model. The random-effects model assumes that the population effect size is expected to vary from study to study, with differences caused by experimental error and differences in populations (between-study variability). Because of this additional variability, the 95% confidence interval for the pooled result is wider (less precise) than for a fixed-effect model, giving a more conservative estimate of association. The meta-analysis results include a graphic display of results to aid interpretation of the findings. The Forest plot, a widely used form of presentation that plots point estimates from different studies along with their error bars, has been created for a visual representation of individual study point estimates. Forest plots were produced for the sensitivities, specificities, and the loglikelihood ratios (positive and negative). A test for heterogeneity was performed by using the Cochran Q statistic to evaluate whether the pooled studies represent a homogeneous distribution of effect size. Missing effect sizes pose a particular problem in meta-analysis, because this jeopardizes the generalization of the results. However, all included studies had sufficient information about effect sizes to contribute to the analysis. One trial (van Rooij) used a subject inclusion criterion that differed from the other included studies. Results in this study were published in abstract form and were included after personal communication with the lead author. The following definitions were used for each tracing: continuous normal voltage is a continuous background activity with voltage of 10 to 25 microvolts; discontinuous normal

3 Amplitude-Integrated EEG / Spitzmiller et al 1071 Table 1. List and Description of Included Studies Amplitude-Integrated Reference EEG Timing Number of Infants Outcomes Studied Results Eken 31 Within 6 hours 31 3,9,18,24 months Griffiths 12 normal voltage 10 mild encephalopathy, Developmental <85 abnormal 2 moderate encephalopathy 19 abnormal tracing 15 died, 4 minor handicap Toet 13 3 and 6 hours old 73 At 3,9,18,24 months, and at 3 and 68 infants followed up to 12 mos. 5 years. Griffith s mental Predictive value of flat tracing, burst developmental scale. Adverse suppression, continuous low voltage, and outcome = death in neonatal poor outcome at 3 hours positive period, cerebral palsy, or predictive value 78%, negative predictive global delay value 84%, sensitivity 85%, specificity 77%; at 6 hours positive predictive value 86%, negative predictive value 91%, sensitivity 91%, specificity 86% ter Horst 12 Median age 4 hours 39 At 24 months: 3 groups Burst suppression or worse if before 6 hours 35 minutes (1) normal; (2) mildly likelihood ratio+ 2.7 (95% CI ) for abnormal neurologic adverse outcome; between 24 and abnormalities present other 46 hours likelihood ratio + 19 (95% CI than cerebral palsy or infantile ); after 24 hours not significant. spasms; (3) severely abnormal Normal pattern likelihood ratio for severe severe mental/motor delay, neurodeficits at follow-up <.3 through infantile spasms, or cerebral first 48 hours palsy Hellstrom- First 6 hours mos follow-up; 3 groups Normal pattern: sensitivity 89%, specificity Westas 33 (1) Healthy (including minor 94.7% positive predictive value 96.2%; visual and hearing deficits (2) Abnormal pattern: sensitivity 94.7%, neurological sequelae (cerebral specificity 89.3%, positive predictive palsy, psychomotor retardation) value 85.7% (3) death al Naqeeb 35 Median age 14 full term Follow-up at 18 and 24 mos. Positive predictive value of normal pattern 18 hours (range, 56 encephalopathic Normal = no abnormal with good outcome 96.2%; abnormal 2 hours-21 days) neurological signs or pattern with poor outcomes 85.7%, developmental quotient 85, prognostic efficacy 91.5% abnormal = neuromotor abnormalities and/or GQ < 85 Thornberg 32 Upon arrival to 38 Denver Development at 5, 10, 21- burst suppression or paroxysmal neonatal intensive 18 mos. If normal at 18 months activity 20 death or severe care unit; 30 no further follow-up. 15 infants neurodisability; 1 severe dyskinetic within first followed to 2.5 years severe syndrome but normal mental development; 2 hours, 8 in neurological disability 17- Continuous activity; 1 died at 6 mos first 1-2 days from SIDS but normal development at that of life point, the rest normal at 18 mos. van Rooij 34 Within 6 hours 160 Griffiths mental developmental 65 infants had initial flat tracing/continuous scale and full neuro exam at low-voltage pattern: 6 recovered to postnatal age of at least continuous normal voltage, and all survived 12 months neonatal period, but 1 had severe disability and 5 were normal at follow-up 25 burst suppression as initial pattern: 12 improved to normal: of those 12, 1 died, 5 moderate to severe disability, 2 mild disability, 4 were normal; if no recovery was seen in first 24 hours death or severe disability Shalak 30 Within 12 hours 50 No follow-up after discharge; 35 infants had normal tracings, and 3 of detailed neurological these developed persistent encephalopathy; examination while in the 4 with moderately abnormal tracings, and neonatal intensive care unit 3 of these developed persistent and immediately prior to encephalopathy; 11 with severely discharge; infants characterized suppressed tracings, and 8 of these into modified Sarnat stages developed persistent encephalopathy. NOTE: EEG = electroencephalography.

4 1072 Journal of Child Neurology / Vol. 22, No. 9, September 2007 voltage is a discontinuous trace with voltage predominately >5 microvolts; burst suppression is discontinuous trace with periods of low cortical activity (<5 microvolts) with bursts of higher amplitude; continuous low voltage is continuous background pattern of very low voltage, around or less than 5 microvolts; flat tracing is a mainly isoelectric tracing of <5 microvolts; and epileptiform activity is single or repetitive events with sustained cortical activity. 12,13 Potentially relevant studies identified through search (n = 24) Studies excluded that only used conventional EEG (n = 9) Results A total of 24 studies were identified using the above search strategy. Figure 1 briefly summarizes the reasons studies were excluded. Nine studies were excluded because they used conventional EEG One study was excluded because it utilized amplitude-integrated EEG in evaluating the outcome of infants with intraventricular hemorrhage. 23 Three studies were excluded because the infants were premature ; one was excluded because it evaluated amplitude-integrated EEG for seizure detection, 27 and in 2 studies, long-term outcome data were difficult to interpret. A total of 8 studies were used in the final meta-analysis. 12,13,30-35 The pooled results of the meta-analysis, indicating the accuracy of severe amplitude-integrated EEG tracings to predict moderate/severe disability or death, are summarized in Table 2 and Figures 2-5. The sensitivities of severe amplitude-integrated EEG tracings to predict a poor outcome (as defined by this study) ranged from 73% to 100%; the specificities ranged from 73% to 100% as well. The chi-square test for negative likelihood ratios shows no significant heterogeneity among the studies (P =.19). Overall, severe amplitude-integrated EEG tracings appeared to be accurate in predicting a poor outcome, with a 91% ( % CI) pooled sensitivity and a 0.09 ( % CI) pooled negative likelihood ratio. Eken et al 31 enrolled 31 infants with a gestational age between 37 and 42 weeks and treated for hypoxic ischemic encephalopathy as defined by Sarnat. 7 The infants were diagnosed with hypoxic ischemic encephalopathy when they met clinical criteria in the first 24 hours and met at least 3 of the following: (1) signs of intrauterine asphyxia, (2) arterial cord blood ph <7.10, (3) delayed onset of spontaneous respirations, (4) Apgar score <5 at 5 minutes, and (5) multiorgan failure. The amplitude-integrated EEG was graded as normal if background activity was continuous and of normal voltage and without seizure activity; abnormal if background activity showed burst suppression, continuous extremely low voltage defined as activity below 5 µv, flat, or when seizure patterns were present. Neurodevelopmental assessments were done at 6 to 24 months of age. Twelve of the 31 had continuous background activity, and all survived with a good outcome except 1 infant who required extra corporeal membrane oxygenation. Nineteen of the 31 had an abnormal Studies excluded that evaluated outcome of intraventricular hemorrhage (n = 1) Studies excluded that evaluated infants less than 901 grams (n = 1) or premature (n = 2) Studies excluded that used amplitude integrated EEG for seizure detection (n = 1) Studies excluded that did not report outcomes (n = 3) Studies included in the meta-analysis which met outlined criteria (n = 8) Figure 1. Flow chart for inclusion/exclusion of studies. EEG = electroencephalography. amplitude-integrated EEG tracing. Three of the infants had continuous but extremely low voltage; all 3 died. Eleven infants had a burst suppression pattern; 7 of the 11 died. Of the surviving 4 infants, 1 had a minor handicap and 3 were normal at follow-up. Five of the 19 had an isoelectric pattern, and all died. Toet et al 13 enrolled 73 infants admitted to the neonatal intensive care unit within 3 hours after birth and treated for perinatal asphyxia if they met at least 3 of the following criteria: (1) signs of intrauterine asphyxia, (2) arterial ph<7.10, (3) Apgar score <5 at 5 minutes, or (4) multiorgan failure. Sixty-eight were followed at least to 12 months of age. The amplitude-integrated EEG was applied at 3 and 6 hours of life. At 3 hours, 36 infants had an abnormal tracing (flat tracing, continuous low voltage,

5 Amplitude-Integrated EEG / Spitzmiller et al 1073 Table 2. Accuracy of Severe Amplitude-Integrated Electroencephalography Tracings to Predict Moderate/Severe Disability or Death Likelihood Likelihood True False True False Sensitivity Specificity Ratio + Ratio Positive Positive Negative Negative (95% CI) (95% CI) (95% CI) (95% CI) Thornberg (92-100) 100 (89-100) 36.2 ( ).023 ( ) Hellstrom-Westas (85-100) 89 (77-100) 8.8 ( ).06 ( ) Eken (83-100) 79 (58-100) 4.4 ( ).075 ( ) Toet (77-99) 73 (57-89) 3.3 ( ).17 ( ) al Naqeeb (45-100) 100 (92-100) 30.8 ( ).26 ( ) Shalak (69-100) 92 (83-100) 11 ( ).12 ( ) ter Horst (51-95) 100 (82-100) 15.4 ( ).28 ( ) van Rooij (88-98) 85 (77-93) 6.3 ( ).08 ( ) Summary (87-95) 88 (84-92) 10.1 (5.5-18).09 ( ) Figure 2. Forest plot illustrating the sensitivities of severe amplitudeintegrated electroencephalography in identifying infants who will have moderate/severe disability or death. Sensitivities (along x-axis) are reported for each study (along y-axis) as well as the overall pooled sensitivity. burst suppression) and 24 had abnormal outcomes including death, major handicap, or global delay. At 6 hours, 35 infants had abnormal tracings, and 31 of these had abnormal outcomes. There were 32 normal tracings at 3 hours, and 27 of these infants had normal outcome. At 6 hours, there were 33 normal tracings, and 30 of those infants had normal outcome. Ter Horst et al 12 retrospectively identified 39 full-term infants who were treated for severe perinatal asphyxia as defined by the presence of at least 2 of the following: (1) signs of fetal distress, (2) umbilical ph or first capillary ph <7.00, (3) delayed spontaneous respiration requiring artificial ventilation at 5 minutes of life, and (4) signs of multiorgan failure. Nine infants were excluded secondary to improper data collection or death. In the remaining 30 infants, amplitude-integrated EEG was recorded in the first 72 hours of life. These infants were followed until 2 years of age. Seven infants had a normal tracing throughout the 72 hours; 6 were normal, and 1 had mild deficits at follow-up. Two infants transitioned from a normal tracing to burst suppression/epileptiform activity then back to a normal tracing. These infants had mild deficits. Five infants transitioned from an abnormal tracing to normal before 12 hours of age; 4 were normal, and 1 had mild deficits. One infant transitioned before 36 hours and had mild deficits. Eight infants had a persistently abnormal tracing, and all 8 either died or had severe deficits. There was no differentiation between death and abnormal outcome in these infants. Hellstrom-Westas et al 33 enrolled 47 infants, and all amplitude-integrated EEGs were recorded in the first 6 hours of life. The infants enrolled had at least 1 sign of intrapartum distress, and they all needed resuscitation with positive pressure ventilation. Thirty-three of the infants required mechanical ventilatory support. Twenty-four of the 33 had signs of hypoxic ischemic encephalopathy. The presence of hypoxic ischemic encephalopathy was not possible to evaluate in 9 infants because of sedation. Outcome was assessed at 12 to 18 months. Twenty-six infants had continuous background activity of normal voltage, and 25 of these had normal outcome; 1 required special training for delayed psychomotor development at age 4 years. Twenty-one infants had an abnormal amplitude-integrated EEG; 9 died, 9 survived with handicap, and 3 were normal. al Naqeeb et al 35 enrolled 14 healthy full-term infants who served as controls and 56 with encephalopathy. Encephalopathy was diagnosed if there was rapid appearance of abnormal neurological signs. Neurologic assessment was performed at 18 and 24 months of age. Twenty-four infants were studied within 12 hours of birth, 9 with normal amplitude and all were normal; 4 with moderately abnormal amplitude and 1 with normal outcome; 1 abnormal outcome and 2 died. Four had suppressed amplitude; 1 with abnormal outcome, and 3 deaths. One infant had normal amplitude with seizures, and that infant had a

6 1074 Journal of Child Neurology / Vol. 22, No. 9, September 2007 Figure 3. Forest plot illustrating the specificities of severe amplitudeintegrated electroencephalography (EEG) in identifying infants who will have moderate/severe disability or death. Specificities (along x-axis) are reported for each study (along y-axis) as well as the pooled specificity. Figure 5. Forest plot illustrating the log-likelihood of a positive test (ln LR+) of severe amplitude-integrated electroencephalography (EEG) in identifying infants who will have moderate/severe disability or death. Figure 4. Forest plot illustrating the log-likelihood of a negative test (ln LR ) of severe amplitude-integrated electroencephalography (EEG) in identifying infants who will have moderate/severe disability or death. normal outcome. Eight infants had moderately abnormal amplitude with seizures; 2 had normal outcome, 3 had abnormal outcome, and 3 died; 6 infants had suppressed amplitude with seizures, 3 had abnormal outcome and 3 died. Thornberg et al 32 enrolled 38 infants with a clinical diagnosis of perinatal asphyxia. The infants included in the study were all transferred into the neonatal intensive care unit. All infants were > 36 weeks gestational age and required resuscitation at delivery with a mean time to spontaneous respiration of 18 minutes. Each infant fulfilled at least 2 of the following criteria for asphyxia: (1) abnormal fetal cardiotocography, (2) meconium-stained amniotic fluid, (3) Apgar score 6 at 5 minutes, (4) ph 7.2 and/or bicarbonate concentration less than 15mEq /L. A follow-up neurologic exam was performed at 5, 10, and 18 months of age by a pediatric neurologist according to standardized programs including the Denver Developmental Screening Test. Children who were considered normal at 18 months of age did not usually have further follow-up. Seventeen infants had continuous activity during the first and/or second day of life, and all had a normal neurological exam at follow-up; 21 infants had burst suppression or paroxysmal activity, and all either died or developed severe disability. van Rooij et al 36 retrospectively reported 160 infants with the most severely abnormal tracings and evaluated their time course of recovery in relation to neurophysiological and neuroimaging findings and neurodevelopmental outcome. A total of 160 infants with hypoxic ischemic encephalopathy were evaluated, 70 with continuous normal voltage/discontinuous normal voltage pattern and 90 with flat tracing/continuous low voltage/burst suppression within first 6 hours of birth. All of the survivors were seen in an outpatient clinic and were assessed using the Griffith s mental developmental scale at postnatal ages of at least 24 months. Neuromotor assessment was performed by a pediatric physiotherapist who was blinded to the amplitude-integrated EEG. Of the 70 infants with either continuous normal voltage or discontinuous normal voltage, 64 were normal at follow-up and 6 had moderate to severe handicap. Of the 90 with either a flat tracing, continuous background of very low voltage, or burst suppression pattern, 11 were normal and 79 either died or had moderate to severe handicap at follow-up. Shalak et al 30 enrolled 50 infants who met all the criteria for hypoxic ischemic encephalopathy. Specifically, delivery at 36 weeks gestation, admission to the neonatal intensive care unit, intrapartum distress defined as fetal heart rate abnormalities or meconium staining of the amniotic fluid, and either an Apgar score of 5 at 5 minutes or severe fetal acidemia defined as either an umbilical cord ph

7 Amplitude-Integrated EEG / Spitzmiller et al or base deficit 16 milliequivalents per liter. The amplitude-integrated EEG was applied within the first 12 hours of birth. A normal tracing was identified in 35 infants, and 3 of these developed persistent encephalopathy in the neonatal period. Four had a moderately abnormal amplitude-integrated EEG tracing, and 3 of these had persistent encephalopathy. Eleven infants had a severely suppressed amplitude-integrated EEG, and 8 of these developed persistent encephalopathy in the neonatal period. Discussion To our knowledge, this is the first meta-analysis of amplitudeintegrated EEG as a predictor of neurodevelopmental outcome. This analysis clearly shows amplitude-integrated EEG to be an excellent screening tool in this population of asphyxiated full-term infants. Conventional electroencephalography has been used in the evaluation of infants with hypoxic ischemic encephalopathy for many years. Although not the primary goal of this meta-analysis, we also reviewed studies that analyzed the usefulness of conventional EEG in the outcome of infants with hypoxic ischemic encephalopathy. We found 6 studies that analyzed the predictive value of conventional EEG in regard to neurodevelopmental outcome in fullterm infants with hypoxic ischemic encephalopathy. 15,18-21 Three studies 15,17,18 analyzed either continuous or serial conventional EEG recordings. All 3 studies concluded that early conventional EEG is predictive of neurological outcome. In addition, and perhaps more important, all 3 concluded that infants who progressed to a normal tracing by 8 to 12 hours of age had better outcomes than those infants whose tracings were abnormal throughout or worsened during the monitoring period. Three studies analyzed 1 conventional EEG either during the first 48 hours of life 20,21 or during the first week of life. 19 The Sinclair study was a retrospective study that analyzed infants born at 37 gestational weeks who had Sarnat stage I-III hypoxic ischemic encephalopathy and at least 1 conventional EEG that demonstrated burst suppression pattern or modified burst suppression. The long-term outcome was then correlated with the conventional EEG pattern, and those infants with modified burst suppression had a better outcome than those with burst suppression. These authors also concluded that conventional EEG has excellent prognostic ability in infants with hypoxic ischemic encephalopathy. As evidenced from these studies, the conventional EEG is useful in predicting the neurodevelopmental outcome in this select group of infants. The conventional EEG, however, is cumbersome and requires expert interpretation and many years of training. Conversely, the amplitude-integrated EEG can be attached to an intravenous (IV) pole at the bedside and does not interfere with day-to-day care. Its utility and practicality have been previously described in numerous articles. 13,27,33,35 Briefly, it records a single-channel EEG from biparietal electrodes and the signal is then filtered, rectified, smoothed, and amplitude integrated. Its interpretation is based primarily on pattern recognition. Because the amplitude-integrated EEG is less cumbersome and is easier to interpret, it is important to determine adequate correlation between amplitude-integrated EEG and conventional EEG, as the latter is considered the gold standard. Our search yielded one study 28 that compared the two. The authors of this study evaluated 36 neonates with gestational age 36 weeks. The amplitude-integrated EEG and conventional EEG were recorded simultaneously and analyzed off-line and classified by 2 investigators who were blinded to the clinical condition and the neurodevelopmental outcome of each infant. They found full agreement between amplitude-integrated EEG and conventional EEG ictal activity. A normal amplitude-integrated EEG corresponded with a normal or depressed conventional EEG in 90% of cases. The positive predictive value for severely abnormal amplitude-integrated EEG to correspond with severely abnormal conventional EEG was 100%. Their conclusion was that amplitude-integrated EEG was a reliable tool for monitoring background patterns and ictal activity. In centers using amplitude-integrated EEG regularly, it is interpreted primarily by clinicians who have no training in conventional EEG analysis and have received minimal training in amplitude-integrated EEG tracing analysis. 27 Interobserver variability of the amplitudeintegrated EEG was studied by al Naqeeb and colleagues. The tracings in this study were analyzed by 2 pediatric residents who had no formal training or research interests in amplitude-integrated EEG and were given brief training in its interpretation. The residents and an amplitudeintegrated EEG expert evaluated the study tracings. The authors found excellent agreement between each resident and the expert with respect to background and seizure activity. 35 This would indicate that a nonexpert can be taught to analyze the amplitude-integrated EEG tracing confidently, although focal low amplitude and very short periods of seizure discharges may be missed. The results shown in Figures 2-5 and Table 2 indicate that the amplitude-integrated EEG is a useful bedside tool in predicting the neurodevelopmental outcome of full-term infants with hypoxic ischemic encephalopathy. There are novel therapies for the treatment and prevention of sequelae of hypoxic ischemic encephalopathy. 1-5,37,38 Whole-body hypothermia has been used in experimental conditions to provide neuroprotection. It has recently been shown that it is safe to use whole-body hypothermia under stringent conditions in full-term infants with hypoxic ischemic encephalopathy. 39 In a recent publication, Gluckman et al 40 reported that selective head cooling with mild systemic hypothermia demonstrated benefit in a population of infants with moderately abnormal amplitude-integrated EEG. Their

8 1076 Journal of Child Neurology / Vol. 22, No. 9, September 2007 inclusion criteria included Apgar scores, continued need for resuscitation, severe acidosis, Sarnat staging, and the severity of amplitude-integrated EEG. Infants were randomly assigned to either cooling (n = 116) or control (n = 118). The authors performed logistic regression analysis controlling for baseline amplitude-integrated EEG amplitude (172 intermediate amplitude-integrated EEG and 46 worst amplitude-integrated EEG group), which revealed that the background amplitude-integrated EEG amplitude and the presence of seizures at the time of enrollment were independently associated with unfavorable outcome. No effect of cooling was found in the infants with the most severe baseline amplitude-integrated EEG (odds ratio 1.8, 95% CI , P =.021); however, a protective effect was seen in the 172 infants with intermediate amplitude-integrated EEG. The outcome was more favorable in the cooled group than in the control group (odds ratio 0.47%; 95% CI ; P =.021). 40 Infants affected by hypoxic ischemic encephalopathy can have devastating outcomes. Amplitude-integrated EEG can be used early in infants at risk for hypoxic ischemic encephalopathy based on history and physical exam. The optimal timing is unclear, although earlier and prolonged monitoring as well as repeated monitorings appear to improve determination of outcome. Five of the studies 12,13,31,33,34 applied amplitude-integrated EEG within 6 hours of delivery. Normalization of the tracing earlier portends a better outcome than persistently severe tracings. Early use of the amplitude-integrated EEG will identify the infants at highest risk for poor outcomes, allowing these infants to receive prompt intervention, such as whole-body hypothermia or selective head cooling. Shalak and her colleagues evaluated the role of coupling the amplitude-integrated EEG and early neurological examination. The neurologic exam was performed within the first 12 hours of birth, and the amplitude-integrated EEG was acquired simultaneously or within 1 hour of the exam. They report a sensitivity of 78%, specificity of 94%, positive predictive value of 85%, and negative predictive value of 92% for abnormal outcome, when there is an abnormal neurologic exam and an abnormal amplitude-integrated EEG. They conclude that the combination of the 2 had the highest predictive capacity with a specificity of 94%, a positive predictive value of 85%, and a positive likelihood ratio of 13, and a significant decrease in the number of false-positive identifications. 30 There was no follow-up after discharge from the neonatal intensive care unit of these infants, but the authors state the presence of moderate or severe encephalopathy at the time of discharge is a strong predictor of poor neurodevelopmental outcome based on a number of large studies 41-43,44(p ) previously published. Amplitudeintegrated EEG is a sensitive and specific modality for determining neurodevelopmental outcome, but the physical exam and possibly additional neurophysiological testing may assist the clinician to determine the severity of a hypoxic-ischemic cerebral insult. The study by van Rooij et al 34 analyzed neuroimaging findings and neurophysiological testing in full-term asphyxiated newborns with severely abnormal amplitude-integrated EEG tracings (flat tracing, continuous low voltage, burst suppression). The study sought to relate the imaging and the neurophysiologic testing with the neurodevelopmental outcome. Only the abstract including amplitude-integrated EEG data was available at the time of this writing, although the data regarding imaging and neurophysiological testing have been submitted for publication (personal communication with LS devries). This large study may lend further credence to the benefit of amplitude-integrated EEG recording in asphyxiated term newborns. Conclusion Based on the empirical results of our meta-analyses (Figures 2-5, Table 2), we conclude that amplitude-integrated electroencephalogram should be part of the initial evaluation in full-term infants with suspected hypoxicischemic encephalopathy. The meta-analysis clearly shows that amplitude-integrated EEG is useful in predicting longterm neurodevelopmental outcome in the population of highrisk term infants. Amplitude-integrated EEG can be used at the bedside without interruption of daily care and can be interpreted without extensive formal training. It also is useful in determining which infants should be included in established neuroprotective protocols. These results apply, thus far, only to full-term infants. More studies are needed to determine the amplitude-integrated EEG patterns in premature infants. Although amplitude-integrated EEG technology allows us to predict which infants will have poor neurological outcomes postnatally and thus to provide early, appropriate neuroprotective measures, we are still handicapped by our limited ability to detect and prevent brain asphyxia prenatally. Further research on the prenatal continuum and prevention of hypoxic ischemic encephalopathy is desperately needed. Further research using amplitudeintegrated EEG should use standard definitions of mild, moderate, or severe abnormalities when reporting to facilitate clarification in the literature. References 1. Vannucci RC. Hypoxic-ischemic encephalopathy. Am J Perinatol. 2000;17: Gluckman PD, Pinal CS, Gunn AJ. Hypoxic-ischemic brain injury in the newborn: pathophysiology and potential strategies for intervention. Semin Neonatol. 2001;6: Ferriero DM. Neonatal brain injury. N Engl J Med. 2004;351:

9 Amplitude-Integrated EEG / Spitzmiller et al Sunshine P. Hypoxic-ischemic encephalopathy: pathophysiology and implications for therapy. Przegl Lek. 2002;59(Suppl 1): Amato M, Donati F. Update on perinatal hypoxic insult: mechanism, diagnosis and interventions. Eur J Paediatr Neurol. 2000; 4: Centers for Disease Control and Prevention. Morbidity and Mortality Weekly Report. In: Ward JW, ed. Morbidity and Mortality Weekly Report. Washington, DC: Department of Health and Human Services, Centers for Disease Control and Prevention; 2004: Sarnat HB, Sarnat MS. Neonatal encephalopathy following fetal distress. A clinical and electroencephalographic study. Arch Neurol. 1976;33: Valkounova I, Maresova D, Trojan S. Perinatal complication: hypoxic-ischemic encephalopathy. Sb Lek. 2001;102: Greisen G. Tape-recorded EEG and the cerebral function monitor: amplitude-integrated, time-compressed EEG. J Perinat Med. 1994; 22: Hagberg B, Hagberg G, Olow I. The changing panorama of cerebral palsy in Sweden I. Analysis of the general changes. Acta Paediatr Scand. 1975;64: Simel DL, Samsa GP, Matchar DB. Likelihood ratios with confidence: sample size estimation for diagnostic test studies. J Clin Epidemiol. 1991;44: ter Horst HJ, Sommer C, Bergman KA, Fock JM, van Weerden TW, Bos AF. Prognostic significance of amplitude-integrated EEG during the first 72 hours after birth in severely asphyxiated neonates. Pediatr Res. 2004;55: Toet MC, Hellstrom-Westas L, Groenendaal F, Eken P, de Vries LS. Amplitude integrated EEG 3 and 6 hours after birth in full term neonates with hypoxic-ischaemic encephalopathy. Arch Dis Child Fetal Neonatal Ed. 1999;81:F19-F Pfenninger J, Bachmann D, Wagner BP. Survivors with bad outcome after hypoxic-ischaemic encephalopathy: full-term neonates compare unfavourably with children. Swiss Med Wkly. 2001;131: Azzopardi D, Guarino I, Brayshaw C, et al. Prediction of neurological outcome after birth asphyxia from early continuous twochannel electroencephalography. Early Hum Dev. 1999;55: Mandel R, Martinot A, Delepoulle F, et al. Prediction of outcome after hypoxic-ischemic encephalopathy: a prospective clinical and electrophysiologic study. J Pediatr. 2002;141: Zeinstra E, Fock JM, Begeer JH, van Weerden TW, Maurits NM, Zweens MJ. The prognostic value of serial EEG recordings following acute neonatal asphyxia in full-term infants. Eur J Paediatr Neurol. 2001;5: Pressler RM, Boylan GB, Morton M, Binnie CD, Rennie JM. Early serial EEG in hypoxic ischaemic encephalopathy. Clin Neurophysiol. 2001;112: Sinclair DB, Campbell M, Byrne P, Prasertsom W, Robertson CM. EEG and long-term outcome of term infants with neonatal hypoxicischemic encephalopathy. Clin Neurophysiol. 1999;110: Selton D, Andre M. Prognosis of hypoxic-ischaemic encephalopathy in full-term newborns value of neonatal electroencephalography. Neuropediatrics. 1997;28: Pezzani C, Radvanyi-Bouvet MF, Relier JP, Monod N. Neonatal electroencephalography during the first twenty-four hours of life in full-term newborn infants. Neuropediatrics. 1986;17: Inder TE, Buckland L, Williams CE, et al. Lowered electroencephalographic spectral edge frequency predicts the presence of cerebral white matter injury in premature infants. Pediatrics. 2003;111: Hellstrom-Westas L, Klette H, Thorngren-Jerneck K, Rosen I. Early prediction of outcome with aeeg in preterm infants with large intraventricular hemorrhages. Neuropediatrics. 2001; 32: Hellstrom-Westas L, Rosen I, Svenningsen NW. Cerebral function monitoring during the first week of life in extremely small low birthweight (ESLBW) infants. Neuropediatrics. 1991;22: Romeo MG, Tina LG, Cilauro S, et al. The importance of using the cerebral function monitor (CFM) in the neurological prognosis of neonates in intensive care. Pediatr Med Chir. 1998; 20: Klebermass K, Kuhle S, Kohlhauser-Vollmuth C, Pollak A, Weninger M. Evaluation of the Cerebral Function Monitor as a tool for neurophysiological surveillance in neonatal intensive care patients. Childs Nerv Syst. 2001;17: Rennie JM, Chorley G, Boylan GB, Pressler R, Nguyen Y, Hooper R. Non-expert use of the cerebral function monitor for neonatal seizure detection. Arch Dis Child Fetal Neonatal Ed. 2004;89:F37-F Toet MC, van der Meij W, de Vries LS, Uiterwaal CS, van Huffelen KC. Comparison between simultaneously recorded amplitude integrated electroencephalogram (cerebral function monitor) and standard electroencephalogram in neonates. Pediatrics. 2002; 109: Archbald F, Verma UL, Tejani NA, Handwerker SM. Cerebral function monitor in the neonate. II: birth asphyxia. Dev Med Child Neurol. 1984;26: Shalak LF, Laptook AR, Velaphi SC, Perlman JM. Amplitudeintegrated electroencephalography coupled with an early neurologic examination enhances prediction of term infants at risk for persistent encephalopathy. Pediatrics. 2003;111: Eken P, Toet MC, Groenendaal F, de Vries LS. Predictive value of early neuroimaging, pulsed Doppler and neurophysiology in full term infants with hypoxic-ischaemic encephalopathy. Arch Dis Child Fetal Neonatal Ed. 1995;73:F75-F Thornberg E, Ekstrom-Jodal B. Cerebral function monitoring: a method of predicting outcome in term neonates after severe perinatal asphyxia. Acta Paediatr. 1994;83: Hellstrom-Westas L, Rosen I, Svenningsen NW. Predictive value of early continuous amplitude integrated EEG recordings on outcome after severe birth asphyxia in full term infants. Arch Dis Child Fetal Neonatal Ed. 1995;72:F34-F Recovery of amplitude-integrated EEG (aeeg) background patterns within 24 hours after perinatal asphyxia. Neonat Neurol. May 3, al Naqeeb N, Edwards AD, Cowan FM, Azzopardi D. Assessment of neonatal encephalopathy by amplitude-integrated electroencephalography. Pediatrics. 1999;103(6 Pt 1): van Rooij LG, Toet MC, Osredkar D, van Huffelen AC, Groenendaal F, de Vries LS. Recovery of amplitude integrated electroencephalographic background patterns within 24 hours of perinatal asphyxia. Arch Dis Child Fetal Neonatal Ed. 2005;90:F245-F Shankaran S, Laptook A. Challenge of conducting trials of neuroprotection in the asphyxiated term infant. Semin Perinatol. 2003;27:

10 1078 Journal of Child Neurology / Vol. 22, No. 9, September Tuor UI, Del Bigio MR, Chumas PD. Brain damage due to cerebral hypoxia/ischemia in the neonate: pathology and pharmacological modification. Cerebrovasc Brain Metab Rev. 1996;8: Lucey J, ed. Whole body hypothermia for hypoxic-ischemic encephalopathy (HIE). Hot Top Neonatol. December 14, 2004, Washington, DC. 40. Gluckman PD, Wyatt JS, Azzopardi D, et al. Selective head cooling with mild systemic hypothermia after neonatal encephalopathy: multicentre randomised trial. Lancet. 2005;365: Nelson KB, Ellenberg JH. Neonatal signs as predictors of cerebral palsy. Pediatrics. 1979;64: Robertson CM, Finer NN. Long-term follow-up of term neonates with perinatal asphyxia. Clin Perinatol. 1993;20: Thornenberg E, Thiringer K, Odelback A, Milson I. Birth asphyxia: incidence, clinical course, and outcomes in a Swedish population. Acta Paediatr. 1995;84: Volpe JJ. Neurology of the Newborn. 4th ed. Philadelphia, PA: W B Saunders; 2001.