|Home | About | Journals | Submit | Contact Us | Français|
The authors evaluated the association between gestational age, birth weight, intrauterine growth and epilepsy in a population-based cohort of 1.4 million singletons born in Denmark (1979-2002). A total of 14,334 individuals were registered with epilepsy in the Danish National Hospital Register as inpatients (1979-2002) and outpatients (1995-2002). Information on gestational age and birth weight was obtained from Danish Medical Birth Registry. Children small at birth were identified through two methods: 1) sex-, birth order-, and gestational-age-specific z-score, and 2) deviation from the expected birth weight estimated based on the birth weight of an older sibling. The incidence rates of epilepsy increased consistently with decreasing gestational age and birth weight. The incidence rate ratios (IRR) for epilepsy in the first year of life were more than five-fold in children born at 22-32 weeks compared with children born at 39-41 weeks, and in children with a birth weight <2,000 grams compared with children of 3,000-3,999 grams. The IRRs decreased with age, but remained elevated into early adulthood. Children identified as growth-restricted according to either of the two methods had increased IRRs for epilepsy, even among children with a ‘normal’ birth weight of 3,500-3,999 grams. Low gestational age at birth and low birth weight are associated with an increased risk of epilepsy throughout childhood and persisting into puberty. Intrauterine growth restriction is associated with an increased risk of epilepsy, even among children with a birth weight in a normal range.
Epilepsy is a disorder of the brain characterized by an enduring predisposition to generate epileptic seizures (1). It affects one percent of the population before the age of 20 years (2). With the development of neuroimaging technology, more epilepsy cases are found with abnormal brain development indicating the role of intrauterine environment on the brain development of fetus. Abnormal development of the brain during prenatal life plays a causal role for some neurological disorders with a clinical onset later in life (3)(4). Preterm delivery and low birth weight have been associated with an increased risk of cerebral palsy (5), mental retardation (6, 7), and behavioral disorders (8). Low birth weight and shorter gestational age are also risk factors for seizure disorders, such as neonatal seizures and febrile seizures (9, 10). Studies on the association between gestational age, birth weight and epilepsy have, however, yielded conflicting findings. One earlier study found a higher frequency of low birth weight in children with epilepsy born to white mothers (11). Rocca and colleagues reported that individuals with complex partial seizures and absence seizures were more often small for gestational age or had a low birth weight (12, 13). Some studies (14-16), but not all (17, 18), reported an association between gestational age or birth weight and generalized tonic-clonic seizures or first unprovoked afebrile seizure. A recent population-based cohort study showed that small for gestational age children had an increased risk of epilepsy compared with children considered to be normally grown (19).
We explored the risk of epilepsy as a function of gestational age, birth weight, and fetal growth in a large population-based cohort followed for up to 24 years of age.
From the Danish Civil Registration System (20) we identified all singletons born in Denmark between January 1, 1979, and December 31, 2002 and alive at the 29th day of life (N=1,470,182). The registry includes the unique identification number assigned to all residents of Denmark and has continuously updated information on vital status. The identification number enables accurate linkage between all national registers.. Children were followed from the 29th day of life (to exclude neonatal seizure) until the onset of epilepsy, death, emigration from Denmark, or December 31, 2002, whichever came first. The study was approved by the Danish Data Protection Agency.
Diagnosis of epilepsy was derived from the Danish National Hospital Register (21), which contains information on discharge diagnoses of all inpatients from Danish hospitals since 1977, while discharge diagnoses for outpatients are recorded since 1995. Diagnostic information is based on the Danish version of the International Classification of Diseases, 8th revision (ICD-8) from 1977 to 1993, and 10th revision (ICD-10) from 1994 onwards. Cohort members were classified as having epilepsy if they had been hospitalized or had been in outpatient care with a diagnosis of epilepsy (ICD-8: 345; ICD-10: G40-G41). The time of onset of epilepsy was defined as the first day of contact with the hospital when they were hospitalized or were in outpatient care with the first diagnosis of epilepsy in the hospital registry.
Information on gestational age, birth weight, Apgar score at five minutes, and congenital malformations was obtained from the Danish Medical Birth Registry, which includes records for all births of Danish residents since 1973 (22). Gestational age in the Medical Birth Registry was mainly based on the date of last menstrual period (LMP) but, in the last 15 years, ultrasound measurements have been increasingly used. Information on cerebral palsy was obtained from the Danish National Hospital Register (21) using ICD-8 codes: 343.99, 344.99 and later ICD-10: G80.
We excluded 50,688 (3.5 percent) births due to missing information on birth weight or gestational age and 56 births registered with a gestational age of less than 22 weeks. Using a method described by Tentoni et al (23), we excluded children recorded as having both a short gestational age (22-33 weeks) and an improbably large birth weight. In brief, we considered the data as inconsistent if the observed birth weight at a given gestational age was more than three times the standard deviation of the major Gaussian component of the distribution of birth weight. Among the 20,656 births registered as having a gestational age between 22 and 33 weeks, we excluded 567 (2.7 percent) babies based on the above criterion.
We calculated the sex-, birth order-, and gestational-age-specific z-score for all babies, based on the data in our study. Birth order was divided in three categories (1, 2, 3+). The lowest 15 percentile of the z-score was used to define “small for gestation” among all babies. For the subset of second-born children with a gestational age greater than 27 completed weeks, we also used the birth weight ratio to identify potentially growth-restricted babies. The birth weight ratio is the observed birth weight divided by the expected birth weight multiplied by 100. The expected birth weight of the second born child was estimated from the birth weight of the first born child based on a method described in detail elsewhere (Our unpublished manuscript). Briefly, the method modified a model proposed by Skjaerven et al (24) by additionally taking into account gestational age and sex of the first-born child. Children in the lowest 15 percentile of the birth weight ratio were considered potentially growth-restricted.
We estimated incidence rate ratio (IRR) of epilepsy by using log-linear Poisson regression (25, 26). All IRRs were adjusted for calendar year, age, and the interaction between age and sex. Age, calendar year, cerebral palsy, and a history of epilepsy of parents or siblings were treated as time-dependent variables (27), whereas all other variables were treated as time-independent variables. Age was categorized in three-month intervals in the first year of life, in one-year intervals between the 1st and the 19th birthday, and ages 20-21 and 22-24 represented the last two categories. P-values were based on likelihood ratio tests and 95 percent confidence intervals (CI) were calculated using Wald’s test (27).
Among 1,418,871 children followed for up to 24 years of age (15.5 million person-years at risk), a total of 14,334 individuals were hospitalized with epilepsy, corresponding to a crude incidence rate of 92.6 per 100,000 person-years.
The incidence rate of epilepsy increased consistently with decreasing gestational age and birth weight, but the association became weaker as age at diagnosis of epilepsy increased (figure 1 and figure 2). For example, compared with children with a gestational age at birth of 39-41 weeks, children with a gestational age of 22-32 weeks had a five-fold (IRR=5.41, 95 percent CI: 4.44, 6.59) higher incidence rate of epilepsy in the first year of life, but a two-fold higher between the ages of 15 and 24 years (IRR= 2.05, 95 percent CI: 1.32, 3.19). Compared with children born at 39-41 gestational week, children born postterm (42 gestational weeks or later) had a slightly increased risk of epilepsy up to eight years of age, with IRRs of 1.07 (95 percent CI: 0.88, 1.30), 1.09 (95 percent CI: 0.87, 1.35), 1.16 (95 percent CI: 0.94, 1.43), 1.21 (95 percent CI: 0.97, 1.51), 1.25 (95 percent CI: 1.00, 1.56), 1.08 (95 percent CI: 0.91, 1.28), and 1.19 (95 percent CI: 1.00, 1.41) for each year of age from the first to the sixth, and seven to eight years of age. However, several of the confidence intervals included one (not shown in figure 1). Compared with children whose birth weight was 3,000-3,999 grams, children whose birth weight was <2,000 grams had an IRR of epilepsy of 5.09 (95 percent CI: 4.34, 5.96) in the first year of life, and 1.73 (95 percent CI: 1.24, 2.41) between the ages of 15 and 24 years. Children whose birth weight was 4,000+ grams had an incidence similar to those with a birth weight of 3,000-3,999 grams (data not shown in figure 2). The age-specific IRRs of epilepsy according to gestational age and birth weight were similar after excluding infants with cerebral palsy, congenital malformations, Apgar score <7 at five minutes, or a history of epilepsy of parents or siblings (2,438 cases and 881,686 person-years excluded) (figure 3 and figure 4).
Due to the change in IRRs over age of onset of epilepsy, we restricted the following analyses to children up to five years of age. The incidence of epilepsy increased with decreasing z-score (table 1). Children born at term with a z-score in the three lowest categories (<5 percent, 5-9 percent, 10-14 percent) of the distribution had a higher IRR of epilepsy compared with those with a z-score above the 15th percentile. Among children born preterm, only those with a birth weight below the lowest 5 percent of the z-score distribution had a significantly higher IRR of epilepsy compared with children with a z-score above the 15th percentile. Among second-born children, the birth weight ratio identified more children at high-risk for epilepsy than the z-score, especially among children born at term (table 1).
Compared to children who achieved their expected birth weight (i.e. the observed birth weight was between 90 percent and 109 percent of their expected birth weight), the IRRs of epilepsy tended to increase with increasing deviations from the expected birth weight, although the findings were statistically significant only among children born at term (table 2). Even children in the ‘normal range’ of birth weights (3,500 - 3,999 grams) had a higher IRR of epilepsy if they had not reached their expected birth weight (IRR=1.48, 95 percent CI: 1.03, 2.14).
Among children who achieved their expected birth weight, the incidence rate of epilepsy increased with decreasing gestational age and birth weight (table 3).
The associations between gestational age, birth weight, and fetal growth indicators, and the risk of epilepsy did not change when we excluded the 3,888 individuals who were only treated as outpatients (data not shown). (this paragraph should be changes when we compare the overall results with those of 1995-2004).
The incidence of hospitalizations with epilepsy increased consistently with decreasing gestational age and birth weight. The association was modified by age at diagnosis of epilepsy; the incidence rate ratio of epilepsy decreased with age at diagnosis, but remained elevated into early adulthood. Children with likely impairment of fetal growth had an increased incidence rate ratio of epilepsy, even if they had a birth weight within the “normal” range.
In general, children with a low birth weight include children who are constitutionally small, have a short gestational age at birth, or are growth-restricted (28). In an attempt to disentangle these factors, we calculated the expected birth weight based on the birth weight of their older siblings to provide a better estimate of the biologic growth potential (29) and reduce the risk of misclassifying children who are constitutionally small as growth-restricted (30). We estimated the risk of epilepsy as a function of the ratio between the observed birth weight and this expected birth weight. Among children who achieved their expected birth weight, the risk of epilepsy increased consistently with decreasing gestational age indicating that preterm birth is an independent risk factor for childhood epilepsy.
Total brain tissue volume increases linearly in the third trimester of fetal life, with a four-fold increase in cortical grey matter between 29 and 41 weeks and a five-fold increase in myelinated white matter between 35 and 41 weeks (31). Premature birth itself may lead to subtle neuropathologies, including cerebral white matter gliosis, hippocampal sclerosis and subarachnoid haemorrhage, as shown in non-human primates (31, 32). Furthermore, children born preterm are more often exposed to infections, pre-eclampsia, eclampsia, and smoking which may increase the risk of epilepsy (19). The association between gestational age and the risk of epilepsy may, therefore, reflect the effect of both immaturity and that of a suboptimal intrauterine environment.
Cerebral palsy and congenital malformations especially in central nervous system are associated with an increased risk of epilepsy (33). We and others have shown that children with a low Apgar scores have an increased risk of epilepsy and that the risk of epilepsy increases with decreasing Apgar scores (34, 35). In this analysis we found that the risk of epilepsy related to low birth weight or shorter gestational age was not mediated by cerebral palsy, congenital malformation or low Agpar scores. An early study showed that low birth weight, preterm birth, and smallness for dates at term were not significantly related to the risk of seizure disorders in children free of cerebral palsy (5). However, the highest risk of afebrile seizures in children who did not have cerebral palsy was in those who were smallness for dates (5).
In our analysis, the association between low birth weight and shorter gestational age and the risk of epilepsy was particularly strong within the first five years of life, perhaps because the immature brain is more susceptible to seizures when exposed to risk factors operating in prenatal life than the immature brain (36). Children with a low birth weight or born preterm also have an increased risk of febrile seizure (10). We can not exclude that some of those febrile seizure have been miscoded as epilepsy, although this bias is likely to be small.
In this study we focused on children born with a short gestational age or children who did not fulfill their growth potential, but postterm delivery is also a risk factor for perinatal complications. A recent study showed that children born postterm had an increased risk of epilepsy in the first year of life (37), but our results failed to show similar results.
Our study is based on a large population-based cohort that was followed for up to 24 years with virtually no loss of follow-up (20). Thus, bias due to selection of study participants cannot explain our findings. The quality of the gestational age assessment is, however, not optimal; estimates from LMP may be biased by early pregnancy bleeding, irregular periods, use of contraceptive methods and recall problems (38) and estimates from ultrasound may be biased by exposures that impair early fetal growth (39). A validation study in Denmark showed that, between 1982 and 1987, 64% of gestational age estimates from the medical records was based on the LMP, 35% on early ultrasound, and 1% on clinical estimates (40) but, in recent years, ultrasound measurements have been increasingly used (41). On the other hand, information on gestational age was recorded before the diagnosis of epilepsy and misclassification is thus most likely to be non-differential, which often attenuates effect measures (42). It is, however, possible that children born preterm or small may have been more likely to receive a diagnosis of epilepsy compared to babies born at term or with normal growth, especially if these factors increased their probability of being hospitalized or diagnosed with seizures.
This study only included singletons and thus the results cannot be applied to children born of multiple deliveries, for whom the significance of preterm and low birth weight are likely to be different.
Diagnoses of epilepsy were obtained from the Danish Hospital Registry, which holds information on discharge diagnoses from Danish hospitals of all inpatients since 1977 and all outpatients since 1995. The positive predictive value of the diagnosis of epilepsy in the Danish Hospital Register has been assessed according to the criteria (which requires ≥ 2 unprovoked seizure episodes) of International League Against Epilepsy (43) and found to be 81 percent (95 percent CI: 75, 87) (44). Unfortunately, we did not have information on completeness of the epilepsy diagnosis in the Danish National Hospital Register. We believe that severe cases of epilepsy are more likely to be hospitalized. Including outpatients in the register in 1995 was followed by a 17% increase in incidence rates of epilepsy (45). When we restricted the analyses to data from 1995-2002, we found the same trend of association between gestational age and birth weight and the risk of epilepsy. An incomplete registration of epilepsy would cause underestimation of the cumulative incidence of epilepsy, but rate ratios would be affected only if it registration depended on exposure status, which is possible. However, this mechanism would result in higher IRR estimates of epilepsy during early childhood, but it is unlikely to explain the long-term impact of gestational age and birth weight on risk for epilepsy.
Our study was limited by lack of detailed clinical data on types of epilepsy. Our validation study showed that the data on epilepsy classification was imprecise (44) and more than half of the epilepsy cases in the first year of life received a code of “unspecified” or “other.”(37) Thus, more studies are needed to evaluate whether the association with birth weight, gestational age, and intrauterine growth is restricted to certain types of epilepsy.
Our study showed that gestational age at birth and intrauterine growth restriction are associated with the subsequent risk of epilepsy. Environmental factors operating in fetal life or short gestation in itself may play a causal role in the development of epilepsy, especially among young children.