Mortality Risk in Children With Epilepsy: The Dutch Study of Epilepsy in Childhood
Objective. Long-term follow-up studies of patients with epilepsy have revealed an increased mortality risk compared with the general population. Mortality of children who have epilepsy in modern times is as yet unknown. Therefore, the objective of this study was to determine mortality of children who have epilepsy in comparison with the general population.
Methods. Between August 1988 and August 1992, 472 children, aged 1 month to 16 years, who presented in 1 of the participating hospitals with 2 or more newly diagnosed unprovoked seizures or at least 1 status epilepticus were enrolled in the study. All children were followed for 5 years or until death. The number of deaths observed during follow-up was compared with the expected number of deaths in the same age group in the general population in the Netherlands.
Results. Nine children died during follow-up, amounting to a mortality rate of 3.8/1000 person-years, which is sevenfold higher than expected (95% confidence interval = 2.4–11.5). No deaths were observed among the 328 children who had epilepsy of nonsymptomatic cause. All deceased children had epilepsy that was caused by a static or progressive neurologic disorder (mortality risk = 22.9; 95% confidence interval = 7.9–37.9). None of them died from sudden unexpected and unexplained death of epilepsy.
Conclusions. In our cohort, we found no indication that children who have nonsymptomatic epilepsy have an increased mortality risk compared with the general population, whereas children who have symptomatic epilepsy have a 20-fold increased mortality risk. These data provide guidance for counseling parents of children who have epilepsy.
Long-term follow-up of patients who have epilepsy reveals a two- to threefold increased mortality risk compared with the general population.1–5 Excess mortality is observed predominantly in the first few years after diagnosis, in patients who have symptomatic epilepsy, and in younger age groups.1,3–5
Few studies have been performed in children only, giving mortality rates of 3.1 to 6.2/1000 person-years.6–8 One of these studies compared mortality risk between children who did and did not have epilepsy. In this retrospective study performed in Victoria (Australia) with the use of death certificates, the mortality risk for children who had epilepsy was 13.2 times higher than the mortality risk in children who did not have epilepsy.7 Death in most children who had epilepsy resulted from pneumonia (36%) or was related directly to the underlying condition (25%). A Swedish study, in which 194 children who had epilepsy were inquired after a follow-up period of 12 years, reported that mortality risk in children who have epilepsy without neurodeficit was not significantly higher than in the general population of the same age, whereas those who had neurodeficit had a significantly higher mortality risk.6 Other studies observed that patients (both children and adults) who had symptomatic epilepsy had a much higher mortality risk as compared with nonsymptomatic patients, but they also found that patients who had nonsymptomatic epilepsy still had a significantly increased mortality risk as compared with the general population.1–3
All of the above-mentioned studies were performed in children who had epilepsy and who were diagnosed in the 1960s to early 1980s. There are various reasons that mortality risk in both symptomatic and nonsymptomatic cases may have changed during the past decades. First, the prescription of many new antiepileptic drugs and increasing possibilities of epilepsy surgery during the past 10 years may have reduced mortality rates by better seizure control and improvement of general physical health. Second, new methods of neuroimaging, such as computed tomography (CT) and magnetic resonance imaging (MRI), have increased the possibilities to discriminate between symptomatic and nonsymptomatic epilepsy. Third, better education of parents and other caregivers in dealing with acute life-threatening situations and improvement of emergency care in hospitals may have led to a decrease in children's dying from epilepsy. To determine whether these factors have influenced mortality rates in children with epilepsy, we determined mortality in a cohort of 472 children who had epilepsy and who were newly diagnosed between 1988 and 1992.
The pediatric neurology departments of 2 university hospitals, 1 a children's hospital and 1 a general hospital, participated in the Dutch Study of Epilepsy in Childhood. Between August 1988 and August 1992, all children who were between the ages of 1 month and 16 years and who presented in 1 of the participating hospitals with 2 or more newly diagnosed unprovoked seizures or at least 1 status epilepticus were enrolled in the prognosis study, which is embedded in the Dutch Study of Epilepsy in Childhood.9The majority of the children were referred directly by their general practitioners, some were referred by pediatricians of the participating hospitals, and some were seen in the emergency departments. On the basis of the population size of the referral region of the participating hospitals and of the incidence rates of epilepsy in developed countries, the recruitment of this study has been estimated to be close to 80% of the expected number of children younger than 16 years with newly diagnosed epilepsy.10 A committee of 3 pediatric neurologists who were involved in the study judged whether the clinical description of the ictal event(s) fulfilled predefined simple descriptive diagnostic criteria.11 Children who had only febrile, neonatal, or acute symptomatic seizures were excluded from the study. Also excluded were children who had been referred from other hospitals and those who were using or had used antiepileptic drugs other than for the treatment of neonatal convulsions. Because of the above-mentioned criteria and the low threshold for referring children directly to pediatric neurologists in the Netherlands, we expect our population to be representative for the general epileptic population younger than 16 years.
The study was approved by the medical ethical committees of all of the participating hospitals. All parents or caregivers gave written informed consent.
From all children, a description of the seizures and the medical history were obtained. A standard electroencephalogram (EEG) was recorded for all children. This was followed by an EEG after partial sleep deprivation, unless the first EEG showed epileptiform activity. Brain imaging was performed with CT, unless the child had typical absences or the treating physician did not consider it necessary. MRI, not being a standard diagnostic procedure in the early years of this study, was done during follow-up in some of the children if indicated. The child's pediatric neurologist decided whether to treat the child and selected the antiepileptic drug. Valproic acid and carbamazepine were used as first-line drugs in all children. According to the protocol, at least 2 first-line and 1 second-line drug had to be tried consecutively up to maximum tolerated dosages if the epilepsy was judged not to be controlled adequately.
A committee of 3 pediatric neurologists who were involved in the study classified the seizures according to the revised classification of the International League Against Epilepsy (ILAE).12 Two years after intake or 2 years after starting medication, all epilepsies were classified according to the revised classification of epilepsies and epileptic syndromes of the ILAE.13 In this classification, idiopathic epilepsies are epileptic syndromes with particular clinical characteristics and with specific EEG findings. They are of unknown origin but have a presumed genetic cause. Symptomatic epilepsies are considered the consequence of a known or suspected disorder of the central nervous system. Cryptogenic epilepsies are epilepsies of unknown origin that do not conform to the criteria for the symptomatic or idiopathic categories. All children who had mental retardation and who had epilepsy of unknown cause were classified as having symptomatic epilepsy.
Of the 494 children who were enrolled initially, 22 were excluded from analysis. Three children were excluded because their seizures seemed to be nonepileptic, 10 children were excluded because they had both epileptic and pseudoseizures that could not be distinguished on the basis of the history that made it impossible to define the prognosis, and 7 children were excluded because they had their second seizure after closure of the intake period. Two children emigrated. The remaining 472 children were followed for 5 years or until death. Of the children who died during this period, the cause of death was assessed by medical records and, if available, autopsy reports.
The number of person-years of follow-up was calculated for all participants in the cohort. The observed number of deaths in each age group was compared with the number of deaths in the general Dutch child population in the same age groups and in the same calendar period. These numbers were obtained from the Central Bureau of Statistics in the Netherlands.14 The expected number of deaths among children who had epilepsy was calculated by multiplying the number of person-years by the corresponding mortality rate of the general Dutch child population. The standardized mortality ratio (SMR), which is the ratio of the observed to the expected number of deaths, was used to assess the increased or decreased risk of death. The 95% confidence interval (95% CI) of the SMR was calculated assuming a Poisson distribution for the observed number of deaths.
In the analyses, children who had idiopathic or cryptogenic epilepsy were grouped together as having nonsymptomatic epilepsy. Clinical characteristics and mortality risks were compared between nonsymptomatic and symptomatic cases with the help of the Mann-Whitney rank-sum test (age at onset), Pearson χ2(distribution epilepsy syndromes), and the SMR.
A total of 472 children (231 boys) were followed for 5 years or until death. Of this cohort, 328 (69.5%) children had nonsymptomatic epilepsy and 144 (30.5%) had symptomatic epilepsy (Table 1). During follow-up, the cause had been altered in 7 (1.5%) children. In 6 children, it changed from nonsymptomatic to symptomatic: CT and MRI findings of 2 children changed (a brain tumor was found in 1), 1 child seemed to have tuberous sclerosis, 1 had periventricular nodular heterotopias, 1 became autistic and mentally retarded, and 1 was found to have hypocalcemia caused by pseudohypoparathyroidism at the time of seizures. In 1 child, the cause changed from symptomatic to nonsymptomatic because he had a vanishing lesion that appeared on MRI. Brain imaging had been performed in 353 (74.8%) children. Of the children who did not undergo neuroimaging, 95 (79.8%) had an idiopathic epilepsy syndrome as defined by the ILAE, 11 (9.2%) from cryptogenic epilepsy, and 13 (10.9%) from a symptomatic epilepsy syndrome: 2 had a spastic tetraplegia, 7 had mental retardation, 3 had a chromosomopathy, and 1 had Niemann-Pick syndrome type C.
In our cohort, age at onset of epilepsy was significantly higher in children who had nonsymptomatic epilepsy than in children who had symptomatic epilepsy (P = .003; Table 1). Furthermore, syndrome distribution differed significantly between the 2 groups; children who had nonsymptomatic epilepsy had relatively more generalized epilepsy than children who had symptomatic epilepsy (P = .015; Table 1).
Nine children (1.9%; 5 boys) died during the follow-up period. This amounts to a mortality rate of 3.8/1000 person-years. Mean age at death was 7.3 years (range = 2.1–13.6 years). Death occurred within 2 years after the onset of epilepsy in 3 children (mean duration between intake and time of death = 2.8 years; range = 1.4–4.8 years). An autopsy was performed on only 1 child. All deceased children had had symptomatic epilepsy (see description in Table 2). Three children had had a progressive neurologic disorder with a bad prognosis already at the onset of epilepsy (cases 3, 4, and 9): 1 child had Niemann-Pick syndrome type C, 1 had a tumor ependymoma grade III, and 1 had infantile ceroid lipofuscinosis. Six children experienced static encephalopathies as a result of both prenatal and postnatal cerebral disease.
As far as we know, none of the deceased children died as a result of a seizure; details at the moment of death of 2 cases are missing, however. Six of the deceased children experienced frequent seizures around the time of death, 3 had not experienced any seizures during at least the last 6 months before death. All but 1 of the deceased children were using antiepileptic drugs at the time of death: 7 children had polytherapy, and 1 had monotherapy. The child with monotherapy (valproic acid) had been seizure-free for 18 months. The valproic acid medication of case 7 was discontinued 6 weeks before death because of cholestatic icterus and thrombopenia, with seizures being well-controlled at that time. Retrospectively, the icterus and thrombopenia seemed to be because of parenteral feeding and sepsis, respectively, and did not seem to be related to the use of valproic acid.
Not one of the children who died fulfilled the criteria of sudden unexpected and unexplained death (SUDEP)15; all children who died had poor health that makes it impossible to classify their death as SUDEP. Furthermore, death did not occur suddenly, and an obvious medical cause of death was found in all patients.
Mortality risk in children with epilepsy was sevenfold increased in relation to that of the general population (95% CI = 2.4–11.5;Table 3). The overall expected number of deaths was 0.75 for boys and 0.54 for girls, whereas, respectively, 5 and 4 deaths were observed. The mortality risk was similar for boys and girls: 6.6 (95% CI = 2.2–15.5) and 7.4 (95% CI = 2.0–19.0), respectively.
Table 4 describes mortality risk in children who had epilepsy for those with nonsymptomatic and symptomatic epilepsy apart. No deaths were observed in the group with nonsymptomatic epilepsy (SMR = 0; 95% CI = 0–2.2). On the basis of the mortality rates in the general population, 1 death was expected. All 9 deceased children had symptomatic epilepsy, whereas 0.39 were expected, amounting to a mortality risk of 22.9 (95% CI = 7.9–37.9).
The data from this study are important because the study is prospective; consists of a large cohort of children with newly diagnosed epilepsy, mostly primary referrals from a defined region in the Netherlands; and has a complete follow-up over 5 years. Almost 70% of the 472 children in our cohort experienced nonsymptomatic epilepsy, which is comparable to observations of other epidemiologic studies in both children and adults.16–21 Most important, no deaths were observed among children who had nonsymptomatic epilepsy, whereas the deaths among children who had symptomatic epilepsy could not be attributed to SUDEP.
In our cohort, mortality risk was sevenfold higher than expected from the general mortality rates of children in the Netherlands. This mortality rate (3.8/1000 person-years) is similar to those observed in previous studies that included only children and obviously has not changed with new treatment strategies or modern antiepileptic drugs.6–8 All children who died in our cohort experienced symptomatic epilepsy. Eight of 9 deceased children used antiepileptic drugs at the time of death, 6 of whom experienced frequent seizures at that time.
A hospital-based case-control study showed that risk of SUDEP increased with, among other factors, onset of epilepsy in childhood or young adolescence (among men) and generalized idiopathic epilepsy (especially in men).22 None of the children in our study group, however, died from SUDEP: all children who died had poor health, death did not occur suddenly, and an obvious medical cause of death was found in all patients. Very recently, others observed as well that SUDEP in children is very rare.23
The observed sevenfold increased mortality risk in our study group is much higher than has been found in other studies in which patients of all ages were included1–5 but lower than in the only previous study that determined mortality risk in a cohort of children who had epilepsy (SMR = 13.2).7 This difference may be because in that study, 94% of the children had symptomatic epilepsy, compared with 30.5% in ours. In that study, children who had symptomatic epilepsy had a 50-fold increased risk of death (95% CI = 31.7–77.9). As stated before, cause distribution in our cohort is comparable to observations of other epidemiologic studies, leading to the conclusion that our cohort reflects well the population that has epilepsy.16–21
Although we followed >300 children with nonsymptomatic epilepsy for 5 years, no deaths were observed. Mortality among children of this age in the general population also is low (∼0.5/1000 person-years). Because of the small number of deaths, the CI of the estimated mortality risk is broad, with an upper limit of 2.2. This upper limit of the mortality risk is in line with previous estimates of the mortality risk in patients (both children and adults) who had nonsymptomatic epilepsy.1–3 However, it must be emphasized that a twofold increase of mortality in children with a low absolute risk is still low (1/1000 person-years). Such a reasoning cannot be made for people who are middle aged and who experience absolute mortality rates that are 10-fold higher.
The interval between onset of epilepsy and time of death is distributed equally in our cohort: 2 to 3 children died in each year of follow-up. It is unknown how this distribution would be if we had continued following the children after 5 years.
Because the recruitment of patients in this study has been estimated to be close to 80% of the expected number of children who have epilepsy and who are younger than 16 years within the referral region of the participating hospitals10 and because the observed distribution of cause is similar as in other studies in children with epilepsy,18–21 we think that our results can be used to extrapolate the risk of death in children who have epilepsy.
This study determined mortality risk in a cohort of 472 children who had newly diagnosed epilepsy and who were followed prospectively and with complete follow-up for 5 years after diagnosis. Mortality risk is increased in children who have epilepsy, but deaths were observed only in the group with symptomatic cause. This substantiates the importance of accurate initial diagnosis. SUDEP was not found in our group and probably is rare in children. These findings provide guidance for counseling parents and caregivers who inquire about the risk of death in their child who has epilepsy.
The Dutch Study of Epilepsy in Childhood was financially supported by the National Epilepsy Fund, Houten, The Netherlands (A72 and A85).
Petra M. C. Callenbach is a research fellow supported by a grant from the Netherlands Organization for Scientific Research (NWO, 940-33-030) and the National Epilepsy Fund (98-14).
- Received May 16, 2000.
- Accepted September 13, 2000.
Reprint requests to (O.F.B.) Department of Neurology, University Hospital Groningen, Box 30001, 9700 RB Groningen, The Netherlands.
- CT =
- computed tomography •
- MRI =
- magnetic resonance imaging •
- EEG =
- electroencephalogram •
- ILAE =
- International League Against Epilepsy •
- SMR =
- standardized mortality ratio •
- CI =
- confidence interval •
- SUDEP =
- sudden unexpected and unexplained death of epilepsy
- Shackleton DP,
- Westendorp RGJ,
- Kasteleijn-Nolst Trenite DGA,
- Vandenbroucke JP
- van Donselaar CA,
- Geerts AT,
- Meulstee J,
- Habbema JDF,
- Staal A
- ↵Centraal Bureau voor de Statistiek. Statistisch Jaarboek. 1994:68
- Camfield CS,
- Camfield PR
- Copyright © 2001 American Academy of Pediatrics