You are on page 1of 10

A Scoring System for Early Prognostic Assessment After Neonatal Seizures Francesco Pisani, Lisa Sisti and Stefano

Seri Pediatrics published online Sep 14, 2009; DOI: 10.1542/peds.2008-2087

The online version of this article, along with updated information and services, is located on the World Wide Web at: http://www.pediatrics.org

PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly publication, it has been published continuously since 1948. PEDIATRICS is owned, published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2009 by the American Academy of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.

Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on September 16, 2009

A Scoring System for Early Prognostic Assessment After Neonatal Seizures
WHAT’S KNOWN ON THIS SUBJECT: Only 1 article has reported a scoring system to predict the outcome of neonates with seizures; however, this previous score was developed on clinical detection of the seizures and did not consider ultrasound examination of the brain. WHAT THIS STUDY ADDS: We developed a scoring system to predict neurologic outcome in neonates with seizures. A composite score accurately predicted favorable outcome in 29 of 36 and unfavorable outcome in 60 of 70 of the infants.
AUTHORS: Francesco Pisani, MD,a Lisa Sisti, MD,a and Stefano Seri, MD, FRCPb
Child Neuropsychiatric Unit, University of Parma, Parma, Italy; and bSchool of Life and Health Sciences, Clinical Neurophysiology Unit, Aston University, Birmingham, United Kingdom KEY WORDS neonatal seizures, neurodevelopmental outcome, newborns, preterm infants ABBREVIATIONS EEG— electroencephalography GA— gestational age ROC—receiver operating characteristic AUC—area under the curve NPV—negative predictive value PPV—positive predictive value IVH—intraventricular hemorrhage CI— confidence interval www.pediatrics.org/cgi/doi/10.1542/peds.2008-2087 doi:10.1542/peds.2008-2087 Accepted for publication May 29, 2009 Address correspondence to Francesco Pisani, MD, University of Parma, Via Gramsci 14, 43100 Parma, Italy. E-mail: francesco.pisani@unipr.it PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275). Copyright © 2009 by the American Academy of Pediatrics FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
a

abstract
OBJECTIVE: The aim of this study was to devise a scoring system that could aid in predicting neurologic outcome at the onset of neonatal seizures. METHODS: A total of 106 newborns who had neonatal seizures and were consecutively admitted to the NICU of the University of Parma from January 1999 through December 2004 were prospectively followed-up, and neurologic outcome was assessed at 24 months’ postconceptional age. We conducted a retrospective analysis on this cohort to identify variables that were significantly related to adverse outcome and to develop a scoring system that could provide early prognostic indications. RESULTS: A total of 70 (66%) of 106 infants had an adverse neurologic outcome. Six variables were identified as the most important independent risk factors for adverse outcome and were used to construct a scoring system: birth weight, Apgar score at 1 minute, neurologic examination at seizure onset, cerebral ultrasound, efficacy of anticonvulsant therapy, and presence of neonatal status epilepticus. Each variable was scored from 0 to 3 to represent the range from “normal” to “severely abnormal.” A total composite score was computed by addition of the raw scores of the 6 variables. This score ranged from 0 to 12. A cutoff score of 4 provided the greatest sensitivity and specificity. CONCLUSIONS: This scoring system may offer an easy, rapid, and reliable prognostic indicator of neurologic outcome after the onset of neonatal seizures. A final assessment of the validity of this score in routine clinical practice will require independent validation in other centers. Pediatrics 2009;124:e580–e587

e580

PISANI et al

Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on September 16, 2009

ARTICLES

Neonatal seizures are considered an epiphenomenon of the underlying neurologic pathology. Despite improved perinatal care, mortality and incidence of neurologic sequelae in the seizing newborn remain high. The identification of early predictors of short-term and longterm neurodevelopmental outcome after neonatal seizures could be a valuable tool in providing supportive indicators for an early referral for long-term follow-up and habilitative intervention. Clinical studies have suggested that the cause of neonatal seizures is the most important factor in influencing outcome.1–5 Other prognostic factors reported in the literature are the ictal semeiology and an early onset of seizures.6,7 Reliability of both of these characteristics can be limited and strongly depends on the seizure type (clinical, electroclinical, or only electrical); the gestational age of the newborn (in the preterm newborn, seizures are usually difficult to detect); and the modality of seizure identification, whether it is only clinical or with temporally related video electroencephalography (EEG).7,8 The latter consideration is particularly important, given that in some newborns the clinical manifestation of the seizures can be subtle or even only electrical and therefore clinically silent. Evidence on the prognostic significance of specific seizure types is not univocal6,7,9–11 because multiple seizure types often coexist in the same patient, and duration of seizures can also play a nonnegligible role.2,11–14 Clinical variables such as gestational age, Apgar score, birth weight, the need for resuscitation maneuvers, neurologic examination,2,4,15,16 and instrumental findings such as EEG and cerebral ultrasonography—now routinely available in most NICUs— have been attributed a significant prognostic role.11,17–20 Development of a neonatal scoring system has been attempted in the past,
PEDIATRICS Volume 124, Number 4, October 2009

but the diagnosis of neonatal seizures was based on clinical criteria only and the studies did not include neuroimaging data.21,22 To address these issues, we conducted a hospital-based study to devise a clinically viable risk scoring system and assess its properties in early prediction of adverse outcome of newborns with neonatal seizures.

started at a mean age of 6.36 days of postnatal life (median: 2.00; SD: 12.17): in preterm infants at a mean age of 6.98 of postnatal days (median: 2.00; SD: 8.98) and in term newborns at a mean age of 5.8 days of postnatal life (median: 2.00; SD: 14.57). A clinical scoring system and prognostic cutoff values were defined by assigning a value from 0 to 3 to the clinical variables by increasing level of severity. The composite score for all newborns was subsequently evaluated in relation to developmental outcome. General development was assessed using the Griffiths’ Mental Development Scale33 and classified as normal when the development quotient was 80 and abnormal when 80.34 The neurodevelopmental outcome was classified as favorable or adverse and was based on data that were systematically collected by a multidisciplinary team that was responsible for the longitudinal follow-up program under the auspices of the local health authority. The definition of favorable outcome included normal neurologic development or mild muscle tone and/or reflex abnormalities, whereas an adverse outcome included death, cerebral palsy, developmental delay, chronic epilepsy, blindness, or deafness. The relationship between scores on the chosen variables and neurodevelopmental outcome at 2 years of age was investigated. Receiver operating characteristic (ROC) curves were constructed to measure the performance of our scoring system in predicting the outcome at 2 years of age. The ROC curve was chosen to show how sensitivity (vertical axis) changed with respect to the false-positive estimate (horizontal axis; 1-specificity), because the decision criterion was varied. The area under the curve (AUC) is considered a better indicator of predictive
e581

METHODS
Sample The study was conducted on data from a neonatal seizure database that consisted of 106 newborns who were consecutively admitted to the NICU of University of Parma from January 1999 through December 2004 and followed up by the Department of Child Neurology. Details on the clinical characteristics of the sample have been described in a previous publication.14 Candidate variables were selected on the basis of a literature review and of their routine availability in the initial days after hospital admission (Table 1).23–27 Data collected included gestational age (GA), type of delivery, birth weight, Apgar score at 1 and 5 minutes, and the need for and type of resuscitation immediately after birth. Time of seizure onset was divided in before or after the first 24 or 48 hours of life. On the basis of seizure semeiology, patients were classified as having 1 type of seizure or multiple seizure types.28 Electroclinical and/or electrographic only seizures were considered. Status epilepticus was recorded as present or absent and defined as continuous seizure activity for at least 30 minutes or recurrent seizures that lasted a total of 30 minutes without definite return to the baseline neurologic condition between seizures.14 Neurologic examination was evaluated clinically. Video EEG and the cerebral ultrasound, which are considered important prognostic indicators,29–32 were included in the analysis. The first video EEG monitoring was

Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on September 16, 2009

TABLE 1 Sample Characteristics
Variable Delivery Spontaneous Cesarean delivery GA, wk 29 29–33 34–36 37 Birth weight, g 1000 1000–1499 1500–2499 2500 Apgar score at 1 min 0–3 4–7 8–10 Apgar score at 5 min 0–3 4–7 8–10 Reanimation maneuver Ordinary assistance Oxygen supplementation Resuscitation 1 min with positivepressure ventilation Endotracheal intubation Cardiac massage and/or drug therapy Etiologyb II I 0 Seizure onset, h 48 48 Seizure type, n 1 1 Anticonvulsant therapy efficacy Immediate response Partial response No response Neurologic examination Normal/mild abnormal Moderately abnormal Severely abnormal Ultrasound brain scanc I II III EEG findingsd I II Status epilepticus Present Absent
a b

Favorable Outcome, n (%) 18 (41.0) 18 (29.0) 4 (14.0) 1 (5.0) 6 (27.3) 26 (47.0) 2 (9.0) 3 (30.0) 3 (15.0) 28 (52.0) 7 (18.0) 5 (18.0) 24 (60.0) 1 (8.0) 10 (26.0) 25 (45.0) 23 (55.0) 1 (25.0) 2 (17.0) 7 (18.0) 3 (30.0) 12 (17.0) 12 (54.5) 12 (86.0) 17 (41.0) 19 (29.0) 15 (41.0) 21 (30.0) 30 (53.0) 4 (19.0) 2 (7.0) 24 (56.0) 10 (29.4) 2 (6.9) 18 (67.0) 17 (55.0) 1 (2.0) 27 (60.0) 9 (15.0) 1 (4.0) 35 (44.0)

Unfavorable Outcome, n (%) 26 (59.0) 44 (71.0) 25 (86.0) 18 (95.0) 16 (72.7) 29 (53.0) 20 (91.0) 7 (70.0) 17 (85.0) 26 (48.0) 31 (82.0) 23 (82.0) 16 (40.0) 11 (92.0) 29 (74.0) 30 (55.0) 19 (45.0) 3 (75.0) 10 (83.0) 31 (82.0) 7 (70.0) 58 (83.0) 10 (45.5) 2 (14.0) 24 (59.0) 46 (71.0) 22 (59.0) 48 (70.0) 27 (47.0) 17 (81.0) 26 (93.0) 19 (44.0) 24 (70.6) 27 (93.1) 9 (33.0) 14 (45.0) 47 (98.0) 18 (40.0) 52 (85.0) 25 (96.0) 45 (56.0)

Total

P .219

44 62 .007a 27 19 22 55 .001a 22 10 20 54 .001a 38 28 40 .019a 12 39 55 .007a 42 4 12 38 10 .001a 70 22 14 .213 41 65 .390 37 69 .001a 57 21 28 .001a 43 34 29 .001a 27 31 48 .001a 45 61 .001a 26 80

accuracy than fixed sensitivity and specificity because it yields an index that is independent of the cutoff point. In addition, ROC curves were used to determine the cutoff values with the best sensitivity and specificity in discriminating between patients with a good outcome from those who had an unfavorable outcome. The negative predictive value (NPV) and the positive predictive value (PPV) were also analyzed. The Student’s t test for unpaired data was used to compare differences in the mean value of continuous variables of subcategories of patients. Nominal data were analyzed by using 2 test and, when necessary, Fisher’s exact test for 2-by-2 comparisons. The variables with a P value of .01 on univariate analysis were included in a multiple logistic regression analysis. A multiple logistic regression model was applied to determine independent risk factors for adverse neurodevelopmental outcome. Statistical analysis was performed by using SPSS 15.0.1 for Windows.35 For the statistical analysis, EEG findings were grouped into 2 categories: 1 normal or mildly abnormal when there was excess sharp activity, absence or decreased frequency of normal patterns, excessively long lowvoltage periods, or overall slightly decreased voltage; and 2 moderately or severely abnormal when there was asymmetries in voltage or frequencies, asynchrony for age, lowvoltage invariant activity, burstsuppression pattern, or permanent discontinuous activity. Ultrasound findings were grouped into 3 categories: I normal; II intraventricular hemorrhage (IVH) of degree 1 or 2, transient periventricular echodensities, borderline ventricular dilatation; and III IVH of degree 3 or 4, intraparenchymal hemorrhage, periventricular leukomalacia, and brain malforma-

Significant on univariate analysis. 0 indicates unknown or transient metabolic disorder; I, mild or moderate hypoxic-ischemic encephalopathy or IVH of degree 1 or 2; II, meningitis, severe hypoxic-ischemic encephalopathy, brain malformation, sepsis, IVH of degree 3 or 4, or intracerebral hemorrhage. c I indicates normal; II, IVH of degree 1 or 2, transient periventricular echodensities, or borderline ventricular dilation; III, IVH of degree 3 or 4, intraparenchymal hemorrhage, periventricular leukomalacia, or brain malformation. d I indicates normal or mildly abnormal; II, moderately or severely abnormal.

e582

PISANI et al

Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on September 16, 2009

ARTICLES

tions. Etiologies were grouped in the following 3 categories: 0 unknown or transient metabolic disorder; 1 mild or moderate hypoxic-ischemic encephalopathy or IVH of degree 1 and 2; and 3 meningitis, severe hypoxicischemic encephalopathy, brain malformation, sepsis, degree 3 to 4 IVH, or intracerebral hemorrhage.

TABLE 2 Potential Predictors of Adverse Outcome (Univariate Analysis)
Variable GA, wk 37 30–36 29 Birth weight, g 2500 1500–2499 1000–1499 500–999 Apgar score at 1 min 8–10 4–7 0–3 Etiologya 0 I II Reanimation maneuver 0 ordinary assistance 1 oxygen supplementation 2 resuscitation 1 min with positivepressure ventilation 3 endotracheal intubation 4 cardiac massage and/or drug therapy Anticonvulsant therapy efficacy 0 immediate response 1 partial response 2 no response Cerebral ultrasound findingsb I II III EEG findingsc I II Neurologic examination Normal Moderately abnormal Severely abnormal Status epilepticus Absent Present
a

OR 1.000 2.391 5.603 1.000 6.103 2.513 10.769 1.000 6.900 6.643 1.000 5.000 29.000 1.000 3.632 6.053 5.361 2.825 1.000 4.722 14.444 1.000 1.647 94.000 1.000 8.667 1.000 3.032 17.053 1.000 19.444

95% CI

P .00700

0.814–7.021 1.720–18.250

.11300 .00400 .00100 .00800 .21400 .00300 .00006 .00100 .00000 .00600 .06000 .00005 .00700 .28100 .03100 .00100 .17000 .00004 .01200 .00100 .00000 .36000 .00000 .00000 .00000 .00000 .02200 .00000 .00000 .00400

1.600–23.269 0.587–10.756 2.289–50.661

RESULTS
Birth weight, GA, Apgar scores at 1 minute, need for resuscitation, cause of seizure, neurologic examination, presence of status epilepticus, efficacy of the anticonvulsant therapy, the background EEG activity, and the cerebral ultrasound scans were the most significant risk factors in the univariate analysis (Tables 1 and 2). Exploratory factor analysis (Principal Axis Factorialization, eigenvalues 1) showed that the variable pairs GA and birth weight, as well as Apgar score at 1 minute and resuscitation had unifactorial saturation on a unique latent component (0.743 and 0.648, and 0.831 and 0.853, respectively). Birth weight is an objective and easily available measure, unlike GA, which is anamnestic and can be less reliable. In light of these considerations, GA was excluded from additional analysis. The variables Apgar score at 1 minute and resuscitation showed a non-Gaussian bimodal distribution. When both variables were simultaneously considered, this led to inadequate ROC curves, with loss of the concavity and therefore a lesser accuracy of the cutoff. Given the statistical properties highlighted by factor analysis and that Apgar score, when added to other clinical variables, led to the construction of more accurate ROC curves, we decided to use this variable in the final analysis and to exclude the variable resuscitation. Seizure onset did not reveal any prognostic power with both a cutoff age of 24 (P .495) and 48 hours (P .213).
PEDIATRICS Volume 124, Number 4, October 2009

2.173–21.914 2.358–18.715

0.899–27.815 5.734–146.666

0.349–37.826 1.180–31.055 1.932–14.878 0.641–12.442

1.412–15.787 3.130–66.661

0.566–4.792 11.101–795.933

3.435–21.865

1.170–7.855 3.593–80.933

2.511–150.591

0 indicates unknown or transient metabolic disorder; I, mild or moderate hypoxic-ischemic encephalopathy or IVH of degree 1 or 2; II, meningitis, severe hypoxic-ischemic encephalopathy, brain malformation, sepsis, IVH of degree 3 or 4, or intracerebral hemorrhage. b I indicates normal; II, IVH of degree 1 or 2, transient periventricular echodensities, or borderline ventricular dilation; III, IVH of degree 3 or 4, intraparenchymal hemorrhage, periventricular leukomalacia, or brain malformation. c I indicates normal or mildly abnormal; II, moderately or severely abnormal.

The best outcome predictors on multiple logistic regression were birth weight, Apgar score at 1 minute, preictal neurologic examination and ultrasound at seizure onset, efficacy of the anticonvulsant therapy, and presence of status epilepticus. These were further used to devise 2 scoring systems in which the variable background EEG activity was added only to the second (Tables 3 and 4).

In the first scoring system, the minimum possible total score was 0 and the maximum was 12 (Tables 4 and 5). This score was highly accurate, with an AUC corresponding to 0.917 (95% confidence interval [CI]: 0.858 – 0.975; P .001) and with a cutoff of 4 that provided the best compromise with a sensitivity of 85.7%, a specificity of 80.6%, and a PPV of 89.6%. Results indicated that the model accurately predicted outcome in 74.4% (29 of
e583

Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on September 16, 2009

TABLE 3 Final Multiple Logistic Regression Model to Predict Adverse Outcome
Variable Birth weight, g 1500–2499 500–999 Apgar score at 1 min 4–7 Anticonvulsant therapy efficacy No response Cerebral ultrasound findings IVH of degree 3 or 4, intraparenchymal hemorrhage, periventricular leukomalacia, or brain malformation Neurologic examination Moderately abnormal Status epilepticus Present OR 42.003 202.731 29.725 63.637 108.905 95% CI 2.710–651.002 3.822–10 752.230 1.842–479.688 2.329–1738.554 5.930–2000.080 P .008 .009 .017 .014 .002

30.201 51.787

2.123–429.688 1.256–2135.550

.012 .038

AUC of 0.919 (95% CI: 0.864 – 0.974; P .001). Considering a cutoff of 5, the score presented a sensitivity of 81.4%, a specificity of 83.3%, and a PPV of 90.5%. Results indicate that 69.8% (30 of 36) of the infants with favorable outcome and 90.5% (57 of 70) of those with unfavorable outcome were correctly predicted. Moreover, the comparison of the AUC of both scores did not show any difference (nonsignificant).

DISCUSSION
Despite the improvement in perinatal care,36 mortality and above all the incidence of neurologic sequelae in newborns with seizures remain high.5 It is therefore paramount for neonatologists to have early and accurate prognostic indicators of short-term and long-term outcome to plan diagnostic, therapeutic, and habilitative intervention. Given the paucity of available studies, we tried to devise a scoring system based on a set of clinical and instrumental variables that are easily available in the NICU and have been shown in previous studies to be strongly correlated with neurodevelopmental outcome.21,22,37 The clinical score was designed to be easily applicable in the first days after the onset of the neonatal seizures and accurate in identifying as early as possible newborns who have seizures and will have an unfavorable outcome. In our cohort, only 10 newborns with a score 4 had neurologic sequelae. A score of 4 provided a sensitivity of 85.7%, a specificity of 80.6%, and a PPV of 89.6%, confirming its potential usefulness as a prognostic indicator. The variables chosen for this study had 3 main properties: they were available at the earliest stages after seizure onset; had “a priori” an early prognostic value2,4,9,14–16; and, for the instrumental investigations, were routinely available in NICUs. Comparing our scoring

TABLE 4 Clinical and Instrumental Variables Scored
Variable Birth weight, g 1000 1000–1499 1500–2499 2500 Apgar at 1 min 0–3 4–7 8–10 Neurologic examination Severely abnormal Moderately abnormal Normal or mildly abnormal Background EEG activity Asymmetries in voltage or frequencies, asynchrony for age, isoelectric or low-voltage invariant activity, burst-suppression pattern, permanent discontinuous activity Normal or excessive sharp activity, absence or decreased frequency of normal patterns, excessively long low-voltage periods or overall slightly decreased voltage Anticonvulsant therapy efficacy No response Partial response Immediate response Ultrasound brain scan IVH of degree 3 or 4, intraparenchymal hemorrhage, periventricular leukomalacia, or brain malformation IVH of degree 1 or 2, transient periventricular echodensities, or borderline ventricular dilation Normal Status epilepticus Present Absent Total Score 1 3 2 1 0 2 1 0 2 1 0 — Score 2 3 2 1 0 2 1 0 2 1 0 1

0

2 1 0 2 1 0 1 0 0–12

2 1 0 2 1 0 1 0 0–13

36) of the newborns with favorable outcome and in 89.6% (60 of 70) of those with unfavorable outcome. When we included background EEG activity, not significant on the multiple
e584 PISANI et al

logistic regression but used in the study as a criterion for case definition, the maximum possible total score reached 13 (Tables 4 and 6). This score had a comparable accuracy, with an

Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on September 16, 2009

ARTICLES

TABLE 5 Score 1
Cutoff 3 4 5
AUC: 0.917; SE: 0.030; P

Sensitivity 0.957 0.857 0.757

Specificity 0.694 0.806 0.917

Accuracy 0.825 0.831 0.837

PPV 0.859 0.896 0.946

NPV 0.893 0.744 0.660

.001; 95% CI: 0.858 to 0.975.

TABLE 6 Score 2
Cutoff 4 5 6
AUC: 0.919; SE: 0.028; P

Sensitivity 0.900 0.814 0.743

Specificity 0.778 0.833 0.917

Accuracy 0.839 0.823 0.830

PPV 0.887 0.905 0.945

NPV 0.800 0.698 0.647

.001; 95% CI: 0.864 to 0.974.

system with the only one currently available,21,22 a difference in ascertainment method emerges. Inclusion of patients with neonatal seizures was based only on clinical criteria without synchronized video EEG recording. This technique has since become an essential tool in confirming the epileptic nature of neonatal paroxysmal events. This limitation might have resulted in the inclusion of neonates with nonepileptic paroxysmal events, who are known to have different prognoses, or exclusion of patients with very subtle or electrical only seizures. Differences in ascertainment methods could have resulted in either an overestimation or an underestimation of the incidence of the neonatal seizures, which has been reported to range between 0.5% and 22.2%.38 This wide variability is likely to be related to different inclusion criteria and can make the comparison between available studies quite challenging. PPV and NPV are indicators of the usefulness of a diagnostic tool, and both measures are related to the prevalence of the disease. In the absence of population-based data on the prevalence of unfavorable outcome of newborns with neonatal seizures in Italy, we chose to rely on clinical data from our center14 rather then adopt data collected in countries with potentially different standards of service provision. Although on the basis of the relaPEDIATRICS Volume 124, Number 4, October 2009

tively low prevalence of neonatal seizures we might have legitimately expected a low PPV in our study, our scoring system was very robust with values as high as 89.6%. Although the exact prognostic value of the presence of specific seizure types remains a challenge for the clinician, a number of studies converge in reporting that patients with subtle seizures have a worse outcome compared with those with clonic seizures.6,7,9 In the only currently available scoring system, neonatal seizures were scored as follows: 0 for the subtle seizures, 1 for the focal clonic or multifocal clonic seizures, and 2 for the tonic or myoclonic seizures.21,22 In our study, despite the improvement in ascertainment methods and a nonnegligible sample size, we could not draw any conclusion on the effect of specific seizure types on neurodevelopmental outcome, and this led us to the choice not to include seizure semeiology in the scoring system. Seizure duration is an additional important point to consider. Prolonged or recurrent seizures are thought to be associated with poor neurologic outcome2,11,13,14; however, an accurate quantification of seizure duration requires a time-consuming calculation and a systematic data collection of the duration of each individual seizure. We adopted a less ambitious but more realistic variable: the presence/absence of status epilepticus. Although we are aware that

a definition of neonatal status epilepticus is far from being universally accepted, the one that we used was in line with that adopted by our group in a previous study14 and by others and was used to allow some comparison with existing literature. An additional element that we considered in our decision was that prolonged seizures can be related to specific clinical conditions with a known favorable outcome, as for focal clonic seizure in neonatal hypocalcemia, whereas neonatal status epilepticus is almost always of symptomatic origin and is more likely to be related to diffuse and extensive structural brain injury, which is likely to have a significant influence on outcome. With regard to the instrumental investigations, the only currently available scoring system21,22 did not include neuroimaging data, which have since become commonly applied in NICUs. In particular, cerebral ultrasound is now readily available at the bedside of the newborn; it is cost-efficient and noninvasive, and its prognostic value for later neurologic sequelae has already been suggested by several studies.39,40 MRI-based techniques, including diffusion-weighted imaging and magnetic resonance spectroscopy, are the new frontier in neonatal brain imaging. Their use is particularly promising in periventricular white matter lesions,41 but they still present inherent difficulties in patient preparation, safety, timing, and sequence optimization. Furthermore, although some abnormalities are better detected by magnetic resonance techniques, cerebral ultrasound will be sensitive enough to identify most abnormalities that are known to be associated with adverse neurologic outcome.42,43 In light of these considerations and our goal for the score to be easily applicable in daily clinical routine, we chose to limit the imaging variable to the traditional ultrasound methods. The proposed score is able to identify only 3 of 4 infants with good outcome
e585

Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on September 16, 2009

with a sensitivity of 85% and a specificity of 80%. On the basis of these properties, we emphasize that the proposed scoring system should not be used in decisions on discontinuation of care or any form of euthanasia. An additional element of debate lies in definition of outcome and the choice of 24 months as follow-up interval. Outcome after neonatal seizures is characterized by a wide spectrum of clinical situations. Disabilities lie in a continuum and can involve multiple domains of development. For this reason, categorical classification can be challenging, hence our choice of using only favorable or adverse as levels of the variable “outcome.” Significantly larger sample size will allow future

studies to be focused on 1 outcome category only such as ongoing seizures/chronic epilepsy. The choice of the 24-month cutoff was motivated by two main considerations. First, observational studies are somewhat limited to short follow-up times, and this has provided limited information on the timing of emergence of neurologic deficits. A second factor in the choice of a cutoff of 2 years is its theoretical advantage in identification of the severity and the type of pathologies such as cerebral palsy and mental retardation more reliably.

CONCLUSIONS
We propose this scoring system for newborns with seizures because of REFERENCES

its immediate availability, its relative simplicity, and its potential for providing early prognostic information on the neurodevelopmental outcome. If confirmed with prospective studies, then the scoring system has the potential to become a useful tool for neonatologists in guiding treatment and follow-up of neonates with seizures. The proposed scoring model needs prospective validation in a new sample of newborns. A prospective study in a small network of NICUs with a different set of patients is due to start this year to increase our understanding of the potential role of the proposed scoring method outside our immediate working environment.

1. Holden KR, Mellits ED, Freeman JM. Neonatal seizures: I— correlation of prenatal and perinatal events with outcomes. Pediatrics. 1982;70(2):165–176 2. Mellits ED, Holden KR, Freeman JM. Neonatal seizures: II—a multivariate analysis of factors associated with outcome. Pediatrics. 1982;70(2):177–185 3. Scher MS. Neonatal seizures and brain damage. Pediatr Neurol. 2003;29(5):381–390 4. Tekgul H, Gauvreau K, Soul G, et al. The current etiologic profile and neurodevelopmental outcome of seizures in term newborn infants. Pediatrics. 2006;117(4):1270 –1280 5. Volpe JJ. Neonatal seizures. In: Neurology of the Newborn. 4th ed. Philadelphia, PA: WB Saunders; 2001:178 –214 6. Brunquell PJ, Glennon CM, DiMario FJ Jr, Lerer T, Eisenfeld L. Prediction of outcome based on clinical seizures type in newborn infants. J Pediatr. 2002;140(6):707–712 7. Mizrahi EM, Kellaway P. Characterization and classification of neonatal seizures. Neurology. 1987; 37(12):1837–1844 8. Sheth RD, Hobbs GR, Mullet M. Neonatal seizures: incidence, onset, and etiology by gestational age. J Perinatol . 1999;19(1):40 – 43 9. Ronen GM, Buckley D, Penney S, Streiner DL. Long-term prognosis in children with neonatal seizures. Neurology. 2007;69(19):1816 –1822 10. McBride MC, Laroia N, Guillet R. Electrographic seizures in neonates correlate with poor neurodevelopmental outcome? Neurology. 2000;55(4):506 –513 11. Bye AM, Cunningham CA, Chee KY, Flanagan D. Outcome of neonates with electrographically identified seizures, or at risk of seizures. Pediatr Neurol. 1997;16(3):225–231 12. Scher MS, Hamid MY, Steppe DA, Beggarly ME, Painter MJ. Ictal and interictal electrographic seizures duration in preterm and term neonates. Epilepsia. 1993;34(2):284 –288 13. Pisani F, Copioli C, Di Gioia C, Turco E, Sisti L. Neonatal seizures: relation of ictal video-EEG findings with neurodevelopmental outcome. J Child Neurol. 2008;23(4):394 –398 14. Pisani F, Cerminara C, Fusco C, Sisti L. Neonatal status epilepticus vs recurrent neonatal seizures: clinical findings and outcome. Neurology. 2007;69(23):2177–2185 15. Bergman I, Painter MJ, Hirsch RP, Crumrine PK, David R. Outcome in neonates with convulsions treated in an intensive care unit. Ann Neurol. 1983;14(6):642– 647 16. Sun Y, Vestergaard M, Pedersen CB, Christensen J, Olsen J. Apgar scores and long-term risk of epilepsy. Epidemiology. 2006;17(3):296 –301 17. Legido A, Clancy RR, Berman PH. Neurologic outcome after electroencephalographically proven neonatal seizures. Pediatrics. 1991;88(3):583–596 18. Rowe JC, Holmes GL, Hafford J, et al. Prognostic value of the electroencephalogram in term and

e586

PISANI et al

Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on September 16, 2009

ARTICLES

19. 20.

21. 22. 23. 24.

25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.

37. 38. 39.

40. 41.

42.

43.

preterm infants following neonatal seizures. Electroencephalogr Clin Neurophysiol. 1985;60(3): 183–196 Connell J, Oozeer R, de Vries L, Dubowitz LM, Dubowitz V. Continuous EEG monitoring of neonatal seizures: diagnostic and prognostic considerations. Arch Dis Child. 1989;64(4):452– 458 Ancel PY, Livinec F, Larroque B, et al. Cerebral palsy among very preterm children in relation to gestational age and neonatal ultrasound abnormalities: the EPIPAGE cohort study. Pediatrics. 2006;117(3):828 – 835 Ellison PH, Largent JA, Bahr JP. A scoring system to predict outcome following neonatal seizures. J Pediatr. 1981;99(3):455– 459 Ellison PH, Horn JL, Franklin S, Jones MG. The results of checking a scoring system for neonatal seizures. Neuropediatrics. 1986;17(3):152–157 European Community collaborative study of outcome of pregnancy between 22 and 28 weeks’ gestation. Working Group on the Very Low Birthweight Infant. Lancet. 1990;336(8718):782–784 Kramer MS, Demissie K, Yang H, Platt RW, Sauve R, Liston R. The contribution of mild and moderate ` preterm birth to infant mortality. Fetal and Infant Health Study Group of the Canadian Perinatal Surveillance System. JAMA. 2000;284(7):843– 849 Mercuri E, Rutherford M, Barnett A, et al. MRI lesions and infants with neonatal encephalopathy: is the Apgar score predictive? Neuropediatrics. 2002;33(3):150 –156 Wood NS, Marlow N, Costeloe K, Gibson AT, Wilkinson AR. Neurological and developmental disability after extremely preterm birth. N Engl J Med. 2000;343(6):378 –384 Sun Y, Vestergaard M, Pedersen CB, Christensen J, Basso O, Olsen J. Gestational age, birth weight, intrauterine growth, and the risk of epilepsy. Am J Epidemiol. 2008;167(3):262–270 Lombroso CT, Neonatal seizures: a clinician overview. Brain Dev. 1996;18(1):1–28 Holmes GL, Lombroso CT. Prognostic value of background patterns in the neonatal EEG. J Clin Neurophysiol. 1993;10(3):323–352 Ortibus EL, Sum JM, Hahn JS. Predictive value of EEG for outcome and epilepsy following neonatal seizures. Electroencephalogr Clin Neurophysiol. 1996;98(3):175–185 Amess PN, Baudin J, Townsend J, et al. Epilepsy in very preterm infants: neonatal cranial ultrasound reveals a high-risk subcategory. Dev Med Child Neurol. 1998;40(11):724 –730 de Vries LS, Eken P, Dubowitz LM. The spectrum of leukomalacia using ultrasound. Behav Brain Res. 1992;49(1):1– 6 Griffiths R. The Abilities of Young Children. London, England: Child Development Research Centre; 1970 Ivens J, Martin N. A common metric for the Griffiths Scales. Arch Dis Child. 2002;87(2):109 –110 SPSS Professional Statistics Edition [computer program]. Version 15.0.1. Chicago, IL: SPSS Inc; 2006 Stoelhorst GM, Rijken M, Martens SE, et al. Changes in neonatology: comparison of two cohorts of very preterm infants (gestational age 32 weeks)—the Project on Preterm and Small for Gestational Age Infants 1983 and the Leiden Follow-up Project on Prematurity 1996 –1997. Pediatrics. 2005;115(2):396 – 405 Portman RJ, Carter BS, Gaylord MS, Murphy MG, Thieme RE, Merenstein GB. Predicting neonatal morbidity after perinatal asphyxia: a scoring system. Am J Obstet Gynecol. 1990;162(1):174 –182 Scher MS. Neonatal seizures classification: a fetal perspective concerning childhood epilepsy. Epilepsy Res. 2006;70(suppl 1):S41–S57 Ment LR, Bada HS, Barnes P, et al. Practice parameter: neuroimaging of the neonate—report of the Quality Standards Subcommittee of the American Academy of Neurology and the Practice Committee of the Child Neurology Society. Neurology. 2002;58(12):1726 –1738 Leijser LM, de Vries LS, Cowan FM. Using cerebral ultrasound effectively in the newborn infant. Early Hum Dev. 2006;82(12):827– 835 Mirmiran M, Barnes PD, Keller K, et al. Neonatal brain magnetic resonance imaging before discharge is better than serial cranial ultrasound in predicting cerebral palsy in very low birth weight preterm infants. Pediatrics. 2004;114(4):992–998 Leijser LM, de Bruine FT, Steggerda SJ, van der Grond J, Walther FJ, van Wezel-Meijler G. Brain imaging findings in very preterm infants throughout the neonatal period: part I—incidence and evolution of lesions, comparison between ultrasound and MRI. Early Hum Dev. 2009;85(2):101–109 Leijser LM, Steggerda SJ, de Bruine FT, van der Grond J, Walther FJ, van Wezel-Meijler G. Brain imaging findings in very preterm infants throughout the neonatal period: part II—relation with perinatal clinical data. Early Hum Dev. 2009;85(2):111–115

PEDIATRICS Volume 124, Number 4, October 2009

Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on September 16, 2009

e587

A Scoring System for Early Prognostic Assessment After Neonatal Seizures Francesco Pisani, Lisa Sisti and Stefano Seri Pediatrics published online Sep 14, 2009; DOI: 10.1542/peds.2008-2087
Updated Information & Services Permissions & Licensing including high-resolution figures, can be found at: http://www.pediatrics.org Information about reproducing this article in parts (figures, tables) or in its entirety can be found online at: http://www.pediatrics.org/misc/Permissions.shtml Information about ordering reprints can be found online: http://www.pediatrics.org/misc/reprints.shtml

Reprints

Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on September 16, 2009