 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
a Department of Community Health and Epidemiology
b Perinatal Epidemiology Research Unit, Departments of Obstetrics and Gynaecology and Pediatrics
c Department of Pediatrics, Dalhousie University, Halifax, Nova Scotia, Canada
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
PATIENTS AND METHODS. Children born between January 1986 and December 2000 in Nova Scotia, Canada were followed up to December 2001. Data on pregnancy and neonatal events and on diagnoses of childhood epilepsy were obtained through record linkage of 2 population-based databases: the Nova Scotia Atlee Perinatal Database and the Canadian Epilepsy Database and Registry. Factors analyzed included events during the prenatal, labor and delivery, and neonatal time periods. Cox proportional hazards regression models were used to estimate relative risks and 95% confidence intervals.
RESULTS. There were 648 new cases of epilepsy diagnosed among 124207 live births, for an overall rate of 63 per 100000 person-years. Incidence rates were highest among children <1 year of age. In adjusted analyses, factors significantly associated with an increased risk of epilepsy included eclampsia, neonatal seizures, central nervous system (CNS) anomalies, placental abruption, major non-CNS anomalies, neonatal metabolic disorders, neonatal CNS diseases, previous low birth weight infant, infection in pregnancy, small for gestational age, unmarried, and not breastfeeding infant at the time of discharge from hospital.
CONCLUSIONS. Our study supports the concept that prenatal factors contribute to the occurrence of subsequent childhood epilepsy.
Key Words: epilepsy • prenatal • pregnancy • neonatal • childhood • epidemiology • cohort
Abbreviations: RR—relative risk • CI—confidence interval • CNS—central nervous system • CEDaR—Canadian Epilepsy Database and Registry • EEG—electroencephalography
Epilepsy is one of the most common neurologic disorders of childhood,1 yet in 60% to 80% of all new cases, no definite cause can be determined.2 Previous studies have reported an association between approximately 40 pregnancy and neonatal factors and childhood epilepsy. Among the factors most frequently cited as being associated with childhood epilepsy are eclampsia/toxemia,3–5 bleeding in the prenatal period,3, 5–7 infection,4, 8 method of delivery,4, 7 asphyxia,3, 8 neonatal seizures,3, 9–11 congenital anomalies,4, 8, 9, 12 low Apgar score,11, 13 and preterm gestational age.3, 7 However, imprecise definitions of epilepsy, potential misclassification and selection bias, poorly defined pregnancy and neonatal variables, and limited statistical analyses make interpretation difficult. Furthermore, most studies are based on births that occurred >2 decades ago. Recent advances in obstetric management and neonatal intensive care technology may have had an effect of the subsequent development of epilepsy. We conducted a comprehensive investigation of the events during pregnancy, labor and delivery, and the neonatal period and their association with childhood epilepsy using recent information from population-based data sets in Nova Scotia, Canada.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Two sources were used to identify children with epilepsy: Nova Scotia cases within the Canadian Epilepsy Database and Registry (CEDaR) and Nova Scotia electroencephalography (EEG) records. Pediatric neurologists from the Division of Pediatric Neurology at Dalhousie University and the IWK Health Centre reviewed all EEG data for children in mainland Nova Scotia between 1986 and 2001 to identify cases not entered in the CEDaR database. All pediatric neurology services, including interpretation and storage of EEGs, for mainland Nova Scotia are located at the IWK Health Centre, Halifax. The diagnosis of epilepsy was made based on information written on the EEG requisition by the requesting physician, the clinical history obtained by the EEG technologist, and the EEG record without reference to other medical charts. This information was reviewed by 3 pediatric neurologists and a consensus was reached on doubtful cases. Approximately 50% of the epilepsy cases were identified from the CEDaR database and 50% from EEG records.
When possible, epilepsy was defined as the occurrence of at least 2 postneonatal (
28 days of age) nonfebrile seizures during the study period.14 CEDaR records age at onset of first seizure rather than age at diagnosis; therefore, criteria for defining an incident case were based on age at onset of first seizure. For the cases identified from the EEG files, the date of the EEG was used as a proxy for age of diagnosis.
It was determined that some cases of epilepsy among residents of Cape Breton Island were not included in the CEDaR database or in the EEG files at the IWK Health Centre. Therefore, this study was restricted to residents of mainland Nova Scotia (ie, residents of Cape Breton Island were excluded). The 1991 population of Nova Scotia was 915102 and approximately 82% live in mainland Nova Scotia.
The Atlee Perinatal database and CEDaR Database were linked using unique variables (such as health card number) common to both databases. An analysis file generated after the linkage included information pertaining to events during the prenatal period, labor and delivery events, and neonatal complications as well as subsequent diagnoses of childhood epilepsy up to 15 years old. Variables chosen for analyses were based on results from previous studies or on the plausibility of an event being an etiologic factor for epilepsy.
Factors analyzed in this study included events from the prenatal period (the time from the beginning of pregnancy, based on last menstrual period or other methods, to the onset of labor), events during labor and delivery, and events during the neonatal period (the time from birth to time of discharge from the hospital of birth).
Statistical Analysis
Initially, rates of epilepsy were determined by age of onset. At each age, the numerator was the number of new cases diagnosed and the denominator was the number of children at risk of developing epilepsy at that age. Because our cohort included births from 1986 through 2000, the denominator for the rates for the older age groups includes only those children born in the earlier years of the study (eg, the rate for onset at 13 years of age would include only those children born in 1988 or earlier).
Cox regression modeling was used to account for the varying lengths of follow-up during the study period. Relative risks (RRs) and 95% confidence intervals (CIs) were estimated from Cox models. Initially, each variable of interest was evaluated in a univariate analysis with epilepsy as the outcome and variables with P 
.1 were then tested for inclusion in a multivariate model. Variables were retained in the multivariate model based on the log-likelihood test (P < .05).
The proportional hazards assumption was tested for each covariate in the final model by examining log negative log plots. For the factors in which hazard functions crossed or showed pronounced divergence, the interaction between the covariate and time was tested. If the addition of the interaction term was significant (P < .05), the covariate was recoded to interact with the age of onset at which the log cumulative hazard function was observed to deviate.15 Data were analyzed with SAS statistical software.16
To assess the proportion of epilepsy that could be attributed to the identified risk factors, population attributable fractions were calculated. The formula used for calculating the population-attributable fractions was specific for situations when adjusted relative risks are used.17
Ethics approval for the study was obtained from the Research Ethics Board of the IWK Health Centre.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
.1). Among the labor and delivery factors (Table 2), only fetal distress, type of labor (no labor, spontaneous or induced), and trial of labor after previous cesarean section met the criteria for additional testing. All of the neonatal factors (Table 3) except birth trauma met the criteria for testing in a multivariate model.
|
|
|
|
Analyses examining the Cox model assumptions showed that the hazards for marital status (unmarried) intersected showing protection and risk over different follow-up periods. This was deemed a chance finding (not biologically plausible) and hence an interaction term was not included in the final model. The estimated relative risk for marital status, therefore, expressed the average risk over the follow-up period.
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Previous studies have shown that maternal epilepsy is a risk factor for epilepsy in the offspring.18 In our study, maternal history of epilepsy was associated with a nonstatistically significant relative risk of 1.7 (95% CI: 0.7–4.3). The wide CI around this relative risk suggests that we may have had insufficient power to detect a moderate increase in risk.
In previous studies of epilepsy and pregnancy factors, the greatest risks have been associated with bleeding in pregnancy.3, 5–7 In our study, there was significant risk associated with placental abruption but not with placenta previa or unspecified antepartum hemorrhage. Placental abruption is more likely to be associated with fetal hypoxia or anoxia than unspecified hemorrhage or placenta previa.
Other investigators have identified relationships between epilepsy and congenital anomalies, although the associations were often nonspecific.4, 8, 9, 12 In our study, children with major anomalies other than CNS anomalies were at risk of epilepsy but those with an anomaly of the CNS were at higher risk. As expected, abnormalities of the CNS (ie, CNS diseases) increased the risk for epilepsy, as did prenatal infections, and neonatal metabolic disorders.3, 8, 9, 19 Recently, these factors have been associated with neuroimaging findings of brain injury.20
A failure to attempt to breastfeed was associated with an increased risk of epilepsy in a previous study.10 Similarly, our results showed that infants who were not breastfed were at increased risk of epilepsy, although this relationship may not be causal because maternal or newborn complications, also associated with childhood epilepsy, may have prevented breastfeeding initiation.
Labor and delivery events were not associated with an increased risk for childhood epilepsy, a finding supported by several earlier studies.3, 5, 9, 10 Only 2 studies have noted any associations between labor and delivery variables and childhood epilepsy. In one, abnormal delivery was found to be significant when children with epilepsy diagnosed in the first year after birth were compared with children with febrile convulsions and with occasional epileptic seizures.4 However, this study grouped several labor and delivery exposures together (induced labor, labor 24 hours, fetal distress, breech, cesarean section, forceps, nuchal cord and "other"), assuming comparable consequences. In another study, cesarean section was associated with epilepsy in a multivariate analysis; however, the authors concluded that cesarean section itself should not be considered a risk factor but rather the result of existing prenatal disorders in the mother or fetus.7
Because the identified neonatal factors such as neonatal seizures may be initiated or exacerbated by prenatal events, the prenatal period is most likely the critical period of risk for epilepsy. Recently reported studies have identified the prenatal period as a critical time of risk for cerebral palsy for both term and preterm infants.21–23 Our epidemiologic study also supports the idea of a continuum of reproductive causality, first suggested in 1954 by Lilienfeld and Pasamanick.3 They proposed that the continuum consists of "brain damage incurred during the prenatal and paranatal periods as a result of abnormalities during these periods, leading to a gradient of injury extending from fetal and neonatal death through cerebral palsy, epilepsy, behavior disorder and mental retardation."3
Some limitations of the present study should be noted. For close to half of the cases in this study, the diagnosis of epilepsy was made only on the basis of the clinical information provided on the requisition for the EEG. Using this information alone likely resulted in our misclassifying some diagnoses, because 2 unprovoked seizures are required for the standard definition of epilepsy. It is reassuring that despite this, our rates were quite similar to the rates obtained from a previous study conducted in Nova Scotia that used more intensive case-ascertainment methodology.24, 25 In some situations, the EEG requisition may suggest epilepsy, but other diagnoses often may be made (eg, syncope). Because we relied heavily on EEG data for diagnoses of epilepsy, information on seizure etiology (idiopathic, cryptogenic, or symptomatic) was missing for the majority of children.
Our inclusion of all epilepsy cases, including cases known to be the result of trauma or infection after the neonatal period, would increase the heterogeneity and may have attenuated the relationship between some factors and epilepsy to a small extent.26 It has been estimated that the proportion of epilepsy attributed to trauma ranges from 4% to 10% and the proportion attributed to infection ranges from 4% to 6%.27 Therefore, we expect that at least 10% of the epilepsy cases would not have a pre- or perinatal influence. As well, we were unable to account for the contribution of genetic factors in this study. Twin studies have quantified the substantial contribution of inheritance factors in the etiology of epilepsy28 in which pre- or perinatal influences are less likely to influence the risk. It is unfortunate that we were not able to separate out these cases of epilepsy, with an identified etiology, to more clearly focus the pre- and perinatal factors on types of epilepsy that might actually be caused by these factors.
It has been estimated that identified risk factors can account for 25% to 45% of epilepsy.27 These estimates include factors such as infection and head trauma that are unrelated to prenatal or neonatal events. In our study, most of the prenatal and neonatal risk factors were associated with attributable fractions in the range of 2% to 8%. These relatively small attributable fractions reflect the moderate-sized relative risks for most factors and the fact that most of the prenatal and neonatal conditions occur infrequently.
It was not possible to assess loss to follow-up from sources such as outmigration from the province or unsuccessful record linkage as a result of inaccurate or incomplete identifying information. It was assumed that a birth that did not link to the CEDaR database was a child that did not have epilepsy. Therefore, a certain proportion of children were likely incorrectly classified as not having epilepsy. Any misclassification would have been nondifferential, which generally results in risk estimates biased toward the null.
The strengths of our study include the fact that it was based on a large, population-based, contemporary cohort. The Nova Scotia Atlee Perinatal Database has extensive clinical information on all births in the province. Pregnancy and neonatal data were collected before outcome was known, thereby minimizing selection and recall bias. Extensive information collected in the prenatal, labor, and delivery and neonatal periods allowed for a comprehensive evaluation of risk factors. In addition, the study spans an era when advanced obstetric and neonatal care has resulted in better survival of small, premature infants.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
We thank Barry Campbell for conducting the data linkage, John Fahey for technical assistance, and Deshayne Fell for assistance with the analysis. We are grateful to the Reproductive Care Program of Nova Scotia and the Canadian Epilepsy Database and Registry for data access.
|
FOOTNOTES |
|---|
Address correspondence to Linda Dodds, PhD, Perinatal Epidemiology Research Unit, Departments of Obstetrics and Gynaecology and Pediatrics, Dalhousie University, 5980 University Ave, Halifax, Nova Scotia, Canada B3H 4N1. E-mail: l.dodds{at}dal.ca
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  Y. Sun, M. Vestergaard, J. Christensen, A. J. Nahmias, and J. Olsen Prenatal Exposure to Maternal Infections and Epilepsy in Childhood: A Population-Based Cohort Study Pediatrics, May 1, 2008; 121(5): e1100 - e1107. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Sun, M. Vestergaard, C. B. Pedersen, J. Christensen, O. Basso, and J. Olsen Gestational Age, Birth Weight, Intrauterine Growth, and the Risk of Epilepsy Am. J. Epidemiol., February 1, 2008; 167(3): 262 - 270. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2006 American Academy of Pediatrics. All rights reserved. |
||
|
||
