What this paper adds
- • Febrile seizures are not associated with problems of behaviour or executive functioning in preschool children.
- • Recurrent febrile seizures are associated with an increased risk of delayed vocabulary development.
Aim General developmental outcome is known to be good in school-aged children who experienced febrile seizures. We examined cognitive and behavioural outcomes in preschool children with febrile seizures, including language and executive functioning outcomes.
Method This work was performed in the Generation R Study, a population-based cohort study in Rotterdam from early fetal life onwards. Information about the occurrence of febrile seizures was collected by questionnaires at the ages of 1, 2, and 3 years. At the age of 3 years, behaviour and emotion were assessed using the Child Behavior Checklist. Information on expressive language development was obtained by the Language Development Survey at the age of 2 years 6 months. To assess executive functioning, parents completed the Behaviour Rating Inventory of Executive Function – Preschool Version when their children were 4 years old. Final analyses were based on 3157 children.
Results No associations were found between febrile seizures and the risk of behavioural problems or executive functioning. In contrast to single febrile seizures, recurrent febrile seizures were significantly associated with an increased risk of delayed vocabulary development (odds ratio 3.22, [95% confidence interval 1.30–7.94]).
Interpretation Febrile seizures are not associated with problem behaviour or executive functioning in preschool children, but the results suggest that children with recurrent febrile seizures might be at risk for delayed language development.
Child Behavior Checklist
Febrile seizures are common in young children, with a cumulative incidence of 2 to 5% in the first 5 years of life. The highest incidence has been reported in the first 2 years of life. The aetiology has remained unclear, although genetic and environmental mechanisms are known to be involved.1
The behavioural and cognitive outcomes of children who had febrile seizures have been the subject of many studies. Most population-based cohort-studies did not find clear differences in developmental outcomes between children with and without febrile seizures, and suggest that febrile seizures are an essentially benign disorder with a good prognosis.2–5 Most of these studies assessed general intelligence, academic progress, and behavioural outcome in school-aged children. However, other aspects of cognitive function or behaviour still might be adversely affected in children with febrile seizures. Also, at a young age, differences in cognitive function or development might be apparent that disappear at a later age. Besides this, certain subgroups of children with febrile seizures may have a less favourable prognosis. Several studies have reported that children with recurrent or prolonged febrile seizures perform worse on neuropsychological tests than typically developing children or those with single, simple febrile seizures6–8. Other studies have suggested an adverse outcome for children with a first febrile seizure before their first birthday.3,5,9
The current study examined whether a history of febrile seizures during the first 3 years of life is associated with verbal and non-verbal cognitive and/or behavioural functioning in preschool children in a population-based cohort among 3157 children. Measures included language development, executive functioning, behaviour, and emotion. We also studied these outcomes in children with recurrent febrile seizures and in children with a first febrile seizure in infancy. We hypothesized that outcome might be less favourable in these groups.
The present study was performed within the Generation R Study, a population-based cohort study from early fetal life onwards in Rotterdam, the Netherlands. The Generation R Study has been described in detail before.10 Participant recruitment started in April 2002 and baseline data collection was completed January 2006. Follow-up assessments with parents and children included questionnaires, physical examinations, and ultrasound examinations. This study was approved by the Medical Ethics Committee of the Erasmus Medical Center, Rotterdam, the Netherlands. Written informed consent was obtained from all parents.
When their children reached the ages of 1, 2, and 3 years, the parents received a questionnaire on the occurrence of febrile seizures. Parents were asked if their child had experienced a febrile seizure or epileptic fit. In addition, they were asked about several symptoms that could have been caused by a seizure. If the answer to one or more of these questions was ‘yes’, then parents were asked to fill in an extended questionnaire about this event. Any medical records resulting from additional visits to the physician were also obtained. Febrile seizures were defined as ‘a condition in childhood characterized by seizures occurring during an acute febrile episode but without evidence of intracranial infections or other defined causes’.11 According to this definition, seizures with fever in children with known neurological disabilities were not considered as febrile seizures. Based on the information available, the first and last author independently classified the events as a febrile seizure or another event. In some cases, insufficient information was available to classify the event or consensus could not be reached. These events were classified as ‘unknown’ and excluded from further analyses. Recurrent febrile seizures were defined as more than one febrile seizure.
When their children reached the age of 3 years, parents were asked about possible behavioural and emotional problems in their children in mailed questionnaires. For this, the Dutch translation of the Child Behavior Checklist for ages 1.5 to 5 years (CBCL/1½–5) was used.12 The CBCL/1½–5 uses 99 questions to obtain ratings of behavioural and emotional problems by parents of 1½- to 5-year-old children, and is validated on a general population sample. These 99 items can be scored on seven syndrome scales: emotionally reactive, anxious/depressed, somatic complaints, withdrawn, sleep problems, attention problems, and aggressive behaviour. The CBCL also uses two broadband scales, designated as internalizing and externalizing, which include the first four scales and the last two scales respectively. Finally, the total problems score is the sum of all items. More problems result in higher scores on the CBCL scales. Good reliability and validity have been reported for the English and Dutch CBCL/1½–5.12 Within Generation R the CBCL was filled in by the primary caregiver, for more than 90% of participants by the mother of the child. Because the resulting scores did not satisfy the assumption of normality, test results were dichotomized as has been described in previous studies.13 We defined a non-optimal score in our population as the highest 20% of problem item scores.
Language skills were assessed by mailed questionnaires. At the age of 1.5 years this was done using the Dutch version of the MacArthur Short Form Vocabulary Checklist, which is appropriate for measuring the word production and comprehension of children aged 16 to 30 months.14 Parents reported on the production of a set of 112 words. To identify expressive language delay at 1.5 years the expressive vocabulary raw scores were converted into age- and sex-specific centile scores based on the whole Generation R sample. Expressive language delay was defined as scores below the 10th centile, in line with a previous definition of expressive language delay based on the MacArthur Short Form Vocabulary Checklist.15,16
At the age of 2.5 years, parents completed a Dutch translation of the Language Development Survey. This is a 310-word vocabulary checklist, with words arranged within 14 semantic categories (e.g. animals, foods, etc.). Parents have to identify the words the child uses spontaneously. The total vocabulary sum scores were converted in age- and sex-specific centile scores based on the complete Generation R sample. The parents were also asked if the child had begun to combine phrases. Expressive language delay at 2.5 years was defined as word production scores below the 10th centile, or not being able to use word combinations in line with previous studies on language delay.16 The Language Development Survey has been generated using community-based samples in the USA and shown to be a valid and reliable instrument to assess expressive language development in children under age 3 in diverse populations.12,17,18
Executive functioning, as evidenced in daily life, was assessed by having the parents complete the Behavior Rating Inventory of Executive Function – Preschool Version in mailed questionnaires at the age of 4 years.19 This is ‘the only standardized executive functioning scale designed specifically for use with preschoolers’ (Sherman19, p1). It yields five non-overlapping theoretically and empirically derived clinical scales, each of which reflects a specific aspect of executive functioning: inhibit, shift, emotional control, working memory, and plan/organize. These clinical scales yield three composite indexes: the Inhibitory Self- Control Index, Flexibility Index, and Emergent Metacognition Index. The overall composite index is the Global Executive Composite (Sherman19, p2). Results are reported as T scores standardized for sex and age; higher scores indicate greater degrees of dysfunction. Scores greater than 65 on the Behavior Rating Inventory of Executive Function are considered to be potentially clinically significant.19 The normative sample, although based on the general population, contained relatively high proportions of upper- and middle-class families.
Questionnaires at enrolment in the Generation R study included items on maternal age and education, and on ethnicity and medical history of both parents. Midwife and hospital registers provided information on the date of birth, as well as birthweight and sex of the child. We assessed maternal psychopathology at 36 months postpartum with the Brief Symptom Inventory. We used the depression/anxiety score as an indicator of psychopathological problems.20
In total, 7295 participants were enrolled for the postnatal phase of the study (Fig. S1). Twins (n=179) were excluded. In total, 527 mothers with two children and 12 mothers with three children participated in the study. For each mother, one child was randomly selected and included in the analysis. Children without information on febrile seizures (n=3409) were excluded. The remaining 3157 participants were eligible for the present study. Of these, information on behavioural outcome (CBCL) was available for 3109 (98.5%) and on executive functioning (Behavior Rating Inventory of Executive Function – Preschool Version) for 2901 (91.9%) participants. Concerning language functioning, word production (MacArthur Short Form Vocabulary Checklist) at the age of 1.5 years was available for 3023 (95.8%) participants. The vocabulary score of the Language Development Survey was available for 2611 (82.7%) participants and the information on word combinations for 2206 (69.9%).
To examine whether non-response was selective, we compared baseline characteristics of participants with and without information on febrile seizures using χ2 statistics for categorical variables, independent t-tests for continuous variables with a normal distribution, and Mann–Whitney U tests for continuous variables that were not normally distributed. In the same way, characteristics of participants with and without febrile seizures were compared with similar models.
Logistic regression models were used to examine whether febrile seizures are associated with internalizing and externalizing behavioural problems as assessed with the CBCL at 3 years of age. We adjusted for covariates known to influence child behaviour from literature: maternal age, educational level, and psychopathology; sex and ethnicity of the child. We also adjusted for parity as this was associated with febrile seizures in univariate analyses and changed the effect sizes of the associations by at least 5%. Using similar models we examined whether febrile seizures were associated with expressive language delay at the age of 1.5 and 2.5 years. Ethnicity, socioeconomic status, maternal age, low birthweight, and sex have been identified as factors influencing early language development in previous studies and were considered as possible confounders. Low birthweight, maternal age, and ethnicity did not change the effect measures found by at least 5% and as such were not included in the final model. We again additionally adjusted for parity as this was associated with febrile seizures in univariate analyses and changed the effect sizes of the associations found. Finally we examined the association between febrile seizures and the standardized scores on the Behavior Rating Inventory of Executive Function – Preschool Version. Differences of mean T scores between participants with and without febrile seizures were calculated, and adjusted for ethnicity and educational level of the mother as these might have influenced executive functioning and the association found.19
Missing values of covariates were multiple imputed, by means of which five imputed datasets were created. Each dataset was analysed separately and the results pooled using the method of Rubin.21 Confidence intervals reflect the uncertainty about the imputed values. The number and percentages of missing values are provided in Table I.
|No febrile seizure||n a||Febrile seizure||n b||p value|
|Mean age (SD), y||31.6 (4.5)||3057||31.2 (4.8)||100||0.356|
|Ethnicity, European, n (%)||2331 (77.2)||3018||67 (67.0)||100||0.017|
|Educational level, n (%)||2982||99||0.333|
|Primary||128 (4.3)||7 (7.1)|
|Secondary||1059 (35.5)||31 (31.1)|
|High||1745 (60.2)||61 (61.6)|
|Primipara, n (%)||1862 (62.2)||2993||70 (72.0)||97||0.046|
|Smoking during pregnancy, n (%)||331 (12.6)||2619||12 (14.6)||82||0.593|
|GSI score, mother median (IQR)||0.05 (0.19)||2402||0.05 (0.14)||77||0.132|
|Febrile seizures, mother, n (%)||2725||92||0.180|
|Yes||78 (2.9)||4 (4.3)|
|No||2468 (90.6)||86 (93.5)|
|Don’t know||179 (6.6)||2 (2.2)|
|Febrile seizures, father, n (%)||2344||72||0.006|
|Yes||55 (2.3)||6 (8.3)|
|No||2141 (91.3)||61 (84.7)|
|Don’t know||148 (6.3)||5 (6.9)|
|Sex, male n (%)||1508 (49.3)||3057||53 (53.0)||100||0.470|
|Ethnicity, European, n (%)||2408 (78.8)||3023||74 (74.0)||100||0.226|
|Birthweight||3474 (547)||3055||3432 (586)||100||0.449|
|Birthweight <2500g, n (%)||122 (4.0)||3055||6 (6.0)||100||0.317|
|Gestational age at birth, median (IQR)||40.1 (1.7)||3056||40.0 (2.0)||100||0.710|
|Preterm birth (<37wks), n (%)||123 (4.0)||3056||5 (5.0)||100||0.627|
The Statistical Package of Social Sciences version 17.0 for Windows (SPSS, Chicago, IL, USA) was used to perform the statistical analyses.
Of the 3157 participants included in the analyses, 90 had experienced a febrile seizure until the age of 2.5 and 100 until the age of 3 years (Table SI). Thirty-three children had recurrent febrile seizures and 23 had a first febrile seizure before their first birthday. Table I shows that mothers of children with febrile seizures more frequently were of other than European origin and more often were primiparous (p<0.05). Parents of children with febrile seizures more often reported a history of febrile seizures themselves. No other differences were found.
Primary caregivers of children with febrile seizures tended to report more sleep problems in their children. Febrile seizures were not significantly associated with an increased risk of problem behaviour at the age of 3 years (Table II). Also, the subgroup of participants with a first febrile seizure before the age of 12 months did not have more problem behaviour than children without febrile seizures (results not shown). Tables III and SII show that we did not find a significant association between single febrile seizures and expressive language functioning at the age of 2.5 years. However, recurrent febrile seizures were associated with expressive language delay at the age of 2.5 (odds ratio 3.22, 95% confidence interval 1.30–7.94). In the same group of children there was no apparent expressive language delay at the age of 18 months (odds ratio 0.99, 95% confidence interval 0.45–2.20). We also did not find a significant association between febrile seizures in the first year of life and expressive language delay at 18 months (data not shown). Only six children already had recurrent febrile seizures before the age of 18 months; this group was too small to justify separate analyses.
|Risk of behavioural problems|
|No febrile seizures||Febrile seizures, all||Single febrile seizures||Recurrent febrile seizures|
|OR (95% CI) % (n)||OR (95% CI) % (n)||OR (95% CI) % (n)||OR (95% CI) % (n)|
|Internalizing problems||1.0 (reference) |
|0.70 (0.38–1.30) |
|0.82 (0.39–1.69) |
|0.52 (0.17–1.60) |
|Externalizing problems||1.0 (reference) |
|0.91 (0.52–1.59) |
|0.75 (0.36–1.54) |
|0.84 (0.32–2.22) |
|Sleep problems||1.0 (reference) |
|1.62 (0.99–2.66) |
|1.41 (0.75–2.65) |
|2.06 (0.92–4.59) |
|Total problems||1.0 (reference) |
|1.05 (0.61–1.82) |
|1.17 (0.61–2.25) |
|1.26 (0.52–3.06) |
|Odds ratio (95% CI)|
|Risk of delayed vocabulary development at 18mo||Risk of delayed vocabulary development at 30mo||Risk of delayed phrase development at 30mo|
|Febrile seizures, all (n=90)||0.98 (0.58–1.67)||1.24 (0.64–2.38)||1.14 (0.60–2.17)|
|Single febrile seizures (n=59)||0.98 (0.51–1.87)||0.55 (0.19–1.58)||0.85 (0.36–1.98)|
|Recurrent febrile seizures (n=31)||0.99 (0.45–2.20)||3.22 (1.30–7.94)||1.88 (0.69–5.12)|
We did not find differences in executive functioning between children with and without febrile seizures (Table SIII), even if only the subgroups of children with recurrent febrile seizures, or with a first febrile seizure before their first birthday were included.
We found no differences in behavioural problems or executive functioning between children with and without a history of febrile seizures. Children with recurrent febrile seizures seemed to be at increased risk of expressive language delay.
The issue of whether febrile seizures are associated with brain damage or dysfunction has been the subject of many epidemiological and experimental studies. Earlier epidemiological studies reported a significant risk of intellectual and behavioural problems among children with febrile seizures.8,22 Most of these studies were hospital based, and because of this selection towards more severe cases may have occurred. Also, some studies included children with known neurological disabilities. Two large national birth cohort studies assessed developmental outcomes in children with a history of febrile seizures. A large prospective population-based study in the USA reported febrile seizures not to be associated with a reduction in IQ or early academic performance at the age of 7.23 The Child Health and Education Study followed a large cohort of children born in the same week in 1970 in the UK. At the age of 5 and 10 years no differences in academic progress, intelligence, and behaviour were found between children with or without a history of febrile seizures.5 Chang et al.2,3 even reported higher scores on tests of intelligence and academic achievement in Chinese schoolchildren with a history of febrile seizures. In a second study on learning and memory aspects, again children with a history of febrile seizures performed better.
Considering that the current study also involved a population-based cohort, in contrast to a clinical population, the results of our study are very much in line with the USA and UK studies. Together, these large population-based studies, which focused on different age groups and different aspects of cognitive functioning, are strongly indicative of a good developmental outcome for most children with febrile seizures.
However, the British, the Chinese, and a Danish study reported a less favourable outcome in children with a first febrile seizure in the first year of life.9 Also, recurrent and prolonged febrile seizures have been associated with a less benign outcome.6–8 It is also well known that epilepsy can be associated with neuropsychological impairment. Several studies already identified cognitive impairments at epilepsy onset, or shortly after diagnosis, even in children with idiopathic epilepsy.24 Taken together, some evidence exists that at least in some of the children with seizures, brain injury or dysfunction can be involved.
We studied language development, executive functioning, and behaviour in preschool children with febrile convulsions in a population-based cohort. Most children begin to use words and rapidly extend their vocabulary between 1 and 3 years of age, the same age most febrile seizures manifest. We hypothesized the occurrence of (recurrent) febrile seizures might be associated with delayed language development. Also, executive functioning has been described to be disturbed in a rat model of febrile seizures and was added because of this. We also studied these outcomes separately in subsets of children with febrile seizures who might be more vulnerable for an adverse outcome: those with early seizure onset or with recurrent febrile seizures. The size of the group of children with prolonged febrile seizures was too small to justify separate analyses.
In accordance with most population-based studies, we did not find any significant differences in behavioural disturbances or executive functioning in children with febrile seizures. Moreover, no associations were found between recurrent febrile seizures or early seizure onset and problem behaviour or impaired executive functioning later in childhood. In contrast to single febrile seizures, we did find recurrent febrile seizures to be associated with an increased risk of expressive language delay at the age of 2.5 years. These children did not show expressive language delay yet at the age of 1.5 years. As far as we are aware, language functioning has not been examined before in association with febrile seizures. A few earlier studies reported poorer linguistic ability in school-aged children with a history of febrile seizures.22 Our findings should be replicated before definitive conclusions can be drawn.
Strengths of this population-based birth cohort-study include the large number of participants who were followed prospectively. Data on febrile seizures were collected every year to increase parental recall. Limitations are that information on paroxysmal events, language, and executive functioning and behaviour was missing for some of the participants. Subgroup analysis showed that non-response was selective. Mothers of children without information on febrile seizures were more often of a non-European origin and lower educated. In children without information on febrile seizures, behavioural problems, language delay, and problems with executive functioning were reported more often (Table SIV). It seems unlikely that the associations studied would be different in this group of participants, but the possibility of selection bias cannot be ruled out completely. Concerning our outcome measures, the non-response was highest for the Language Development Survey, so results about language development are based on fewer individuals. No validated screening instrument exists for febrile seizures. We based our screening questions on previously used screening questionnaires for seizure disorders. These questions were adjusted to young children and some clinically relevant questions were added. In accordance with the definition of febrile seizures, we excluded children with intracranial infections or other defined causes. However, it cannot be completely excluded that in some children a more subtle underlying pathology was involved that we were not aware of. In some children the febrile seizure may have been a first manifestation of an underlying genetically determined seizure disorder. We did not take antiepileptic drug treatment into account. However, within the Netherlands, antiepileptic drugs on a daily basis are only very rarely described for febrile seizures, even if they are recurrent. Therefore it is not likely that these children were treated with antiepileptic drugs.
In conclusion, in line with previous studies our results suggest that febrile seizures are not associated with poor behavioural or cognitive outcome in young children. Experiencing recurrent febrile seizures, however, might adversely affect expressive language development. Further studies for replication and long-term follow-up of these results are needed.
The Generation R Study is conducted by the Erasmus Medical Center Rotterdam in close collaboration with the Faculty of Social Sciences of the Erasmus University Rotterdam, the Municipal Health Service Rotterdam area, the Rotterdam Homecare Foundation, and the Stichting Trombosedienst & Artsenlaboratorium Rijnmond (STAR), Rotterdam. We acknowledge the contribution of general practitioners, hospitals, midwives, and pharmacies in Rotterdam. The first phase of the Generation R Study was made possible by financial support from the Erasmus Medical Center Rotterdam, the Erasmus University Rotterdam, and the Netherlands Organization for Health Research and Development (ZonMw). The present study was supported by an additional grant from the Dutch National Epilepsy Foundation.