Febrile convulsions occur following fever, without evidence of intracranial infection or defined cause, typically between the age of 3 months and 5 years. 1 Febrile convulsions affect 2% to 5% of all children 2,3 and are by far the most common type of seizure in childhood. Febrile convulsions tend to run in families, 4–6 suggesting that genes or environmental factors shared by family members play a causal role. Genetic loci that may account for the susceptibility to convulse with fever have been mapped, 7–9 and several models of inheritance have been suggested; most evidence favors a polygenetic or multifactorial model. 10 The relative importance of genes and environment has been estimated in twin studies in which the probandwise concordance rate was approximately 35% for monozygotic twins and 15% for dizygotic twins. 11,12 However, twins often experience a considerable retardation in intrauterine growth, and risk factors identified in twins may not be valid in the general population. 13
We used the computerized square dance design 14 to evaluate the effect of genes and early environment on the risk of febrile convulsions in a study of two cohorts. One cohort consisted of children who had an older sibling who had had febrile convulsions (FC+ cohort), and the other cohort consisted of children whose older sibling had never had febrile convulsion (FC− cohort). According to the causal field model, 15 the older siblings in the FC+ cohort were exposed to a sufficient set of causal factors to develop febrile convulsions whereas the older siblings in the FC− cohort were short of at least one component cause. We aimed to study the effect of changes that took place between pregnancies in these two cohorts such as change of parents, social conditions, and residence. If an important causal factor was subtracted from the sufficient causal field in the FC+ cohort or added to the causal field in the FC− cohort, we expected the risk of febrile convulsions to decrease in the FC+ cohort and increase in the FC− cohort. Furthermore, we aimed to evaluate the effect of birth weight, gestational age, and fetal growth on febrile convulsions in the two cohorts, which were partly matched for genetic susceptibility.
Our follow-up study was approved by the Danish data-protection board and based on information from two nationwide registers in Denmark. The National Hospital Register contains information on almost all discharges (99%) from Danish hospitals since 1977, and outpatients have been included in the register since 1995. 16 Diagnostic information is classified in the National Hospital Register according to a Danish version of the International Classification of Diseases (ICD); ICD8 was used from 1980 to 1993, and ICD10 from 1994 to 1998. The register includes up to 20 diagnoses for each hospital discharge. We included children with febrile convulsions (ICD8 code 780.21 or ICD10 code R56.0), who were between 3 and 60 months old at the time of discharge, and had no recorded history of non-febrile convulsions, cerebral palsy, severe head traumas, intracranial tumors, meningitis, or encephalitis. The Fertility Database links several population-based databases to obtain data on family structure, social conditions, and pregnancy outcome. 17
We linked information from the two registries by means of the personal identification number assigned to all Danish citizens at birth. We then identified two cohorts consisting of children born in Denmark between 1980 and 1994. The FC+ cohort consisted of the subsequent sibling of children who had been hospitalized with febrile convulsions (N = 10,224; 8,590 full-siblings and 1,634 half-siblings). The FC− cohort consisted of the subsequent sibling of randomly selected children who had never been hospitalized with febrile convulsions (N = 21,218; 18,205 full-siblings and 3,013 half-siblings). We excluded 115 (1%) siblings from the FC+ cohort and 214 (1%) siblings from the FC− cohort because they had no fathers registered. We included only one pair of siblings from each family. In both cohorts, the younger sibling was the unit of observation and will be referred to as the “outcome child,” and the older sibling will be referred to as the “index child.” We categorized social status as low, middle, or high according to the parent with the higher status based on information updated on 1 January of the year of birth of each child. Low social status comprised unskilled manual workers, persons on national supplementary disability pension, and persons with unspecified and unknown job level. Middle social status comprised office workers (including nurses, secondary school teachers, and journalists), students, skilled manual workers, and assisting spouses (that is, women working in the husband’s enterprise). High social class comprised high-ranking office workers (including professors and doctors), managers, and self-employed (including general practitioners and lawyers). Residence at birth was classified according to the municipality in which the family had their permanent address on 1 January of the year during which each child was born, and the category “Change in residence” consisted of families who changed from one municipality to another between the two births. Denmark has 275 municipalities, an area of 44,000 km, 2 and a population of 5.2 million inhabitants.
Birth weight ratio was defined as the ratio of observed-to-expected birth weight, and was calculated for full-siblings only. Calculation of the expected birth weight was based on gender and gestational age of the outcome child, and birth weight of the older sibling, as suggested by Skjaerven et al. 18 Birth weight ratios were categorized before analysis in five categories (Table 1).
The incidence of febrile convulsions in the outcome child was studied according to changes in the following risk factors during the interpregnancy interval: parents (mother or father), social status (decline or rise), and residence at birth for each of the two cohorts. We used a Cox proportional hazards model to assess hazard ratios (HR) and 95% confidence intervals (95% CI) for febrile convulsions. The period of follow-up for febrile convulsions began at the age of 3 months and ended at the day of admission for febrile convulsions, day of death, day of emigration, when the child reached 5 years of age, or 31 December 1998, whichever came first. Results were adjusted for maternal age, parity, social status of the family at birth of index child, year of birth of outcome child, and the interpregnancy interval. We also used a Cox proportional hazards model to estimate HR and 95% CI for febrile convulsions in relation to birth weight, gestational age at birth, and birth weight ratio. Results were adjusted for maternal age, parity at birth of the index child, social status of the family, year of birth of outcome child, and interpregnancy interval. All variables in regression models were categorized as shown in Table 1.
The risk of febrile convulsions for full-siblings and half-siblings was 12% and 8% in the FC+ cohort and 3% and 4% in the FC− cohort, respectively. In both cohorts, families with a change in one parent (that is, half-siblings) were more likely to be of low social status at the time of birth of the older sibling, have a higher social mobility (probably driven by the new partner), involve young mothers, and have a longer interpregnancy interval compared with stable couples (Table 1).
Table 2 shows that change of either parent decreased the risk of febrile convulsions in the FC+ cohort and increased the risk in the FC− cohort. For paternal and maternal half-siblings, the hazard ratio of febrile convulsions was 0.6 (95% confidence interval [CI] = 0.4–0.8) and 0.7 (95% CI = 0.6–0.9), respectively, compared with full-siblings in the FC+ cohort. The corresponding hazard ratios were 1.2 (95% CI = 0.9–1.6) and 1.3 (95% CI = 1.0–1.8) respectively, in the FC− cohort. The results remained virtually unchanged after restricting the cohorts to first- and second-born children (data not shown). In the FC− cohort, we found an increased risk of febrile convulsions if the outcome child was a boy and the index child was a girl. A change in residence or social status during the interpregnancy interval was not associated with the risk of febrile convulsions.
The number of hospitalizations for febrile convulsions experienced by the index child was strongly associated with the risk of febrile convulsions in the outcome child, and a dose-response relation was found. A two-fold increase in the risk of a febrile convulsion was thus found if the index child was hospitalized three times or more with febrile convulsions compared with children whose older sibling was hospitalized one time only (Table 3). We found no association between the age of the index child at the time of the first febrile convulsions and the risk of febrile convulsions in the outcome child (data not shown).
Table 4 shows perinatal determinants of febrile convulsions for full-siblings. The risk of febrile convulsions increased consistently with decreasing birth weight; the adjusted HR for the lowest (<2500 gm) compared with the highest (>3999 gm) birth weight category was 1.5 (95% CI = 1.0–2.1) in the FC+ cohort and 1.6 (95% CI = 1.0–2.5) in the FC− cohort. The corresponding estimates were 1.3 (95% CI = 0.9–1.9) and 1.1 (95% CI = 0.7–1.9) when the association between birth weight and febrile convulsions in the two cohorts was adjusted for gestational age at birth (data not shown). Children born preterm (<37 weeks) had an increased risk of febrile convulsions compared with children born at term; the adjusted HR was 1.4 (95% CI = 1.1–1.8) in the FC+ cohort and 1.9 (95% CI = 1.3–2.7) in the FC− cohort. In the FC+ cohort, the risk of febrile convulsions increased with decreasing birth weight ratio; compared with children within the expected birth weight range, children extremely small for gestational age had an adjusted HR of 1.9 (95% CI = 1.3–2.7).
Children who have had febrile convulsions have, by definition, been exposed to a sufficient set of causal factors. If genes play a causal role, we would expect the risk to decrease when a parent is changed in the FC+ cohort, and this expectation was corroborated in the present study. The FC− cohort was a low risk population and the genetic factors need not be present, but could be introduced by a new parent. We would therefore expect the risk to increase when a parent is changed, as was seen. Thus, our data suggest that maternal as well as paternal genes play a causal role in the etiology of febrile convulsions. In recent years, several genetic loci that may be associated with febrile convulsions have been reported, 7–9 but the mode of inheritance remains controversial. Febrile convulsion is probably a phenotype made up of etiologically distinct subgroups, which may explain the controversy. Indeed, complex segregation analysis of a large sample of febrile convulsion pedigrees in Rochester, New York, suggested a polygenetic mode of inheritance in families where the proband had a single febrile convulsion, but a single-major-locus model with a nearly dominant seizure susceptibility in families where the proband had more than three febrile convulsions. 19 Like Hauser and co-workers, 4 we found a strong association between the number of febrile convulsions experienced by the index child and febrile convulsions in the outcome child, which corroborates the hypothesis that families with multiple febrile convulsions may constitute a separate etiological entity.
A new mother presents not only a change in genes but also a change in the intrauterine environment. If time-stable maternal characteristics such as chronic diseases, smoking, or dietary habits play a causal role, we would expect a larger reduction of risk in the FC+ cohort following a change of mother rather than change of father. Our results indicated this, but the difference was small. Changes in residence during the interpregnancy interval have been found to reduce the recurrence risk of congenital malformations in some 20 but not all studies. 21,22 In the present study, changes in residence had only a minor effect on the risk of febrile convulsions, perhaps because change in residence is a crude measure of most environmental exposures. Changes in social status may be associated with changes in maternal lifestyle factors such as smoking, drinking, and dietary habits, and a decline in social status increases the risk of giving birth to a child with a birth weight below 2,500 gm. 23 However, change in social status had only a modest effect on the risk of febrile convulsions.
Birth weight is associated with infant mortality, childhood morbidity, and cognitive function. 24,25 However, the association between birth weight and febrile convulsions remains controversial. Some studies have found an association, 5,26 while others have not. 27,28 In the present study, the risk of febrile convulsions increased consistently with decreasing birth weight. Furthermore, our data showed a strong association between preterm birth and febrile convulsions. The association of birth weight and gestational age with febrile convulsions was equally strong in the two cohorts consisting of children with high and low background risk. To estimate the effect of fetal growth, we calculated the ratio between observed and expected birth weight. The expected birth weight was based on gestational age and gender of the outcome child, and birth weight of the older sibling. Since birth weights of siblings are highly correlated, 18 a birth weight ratio substantially lower than 1.0 suggests that the fetus was growth-retarded. We found a strong association between birth weight ratio and febrile convulsions in the FC+ cohort.
The data for this study were collected prospectively, and independent of parental recall. We had almost complete follow-up and an accurate recording of family structure and social indicators due to the personal number assigned to all Danish citizens at birth. However, some children may not be hospitalized with their first attack of febrile convulsions. We expect this number to be small because febrile convulsions are a frightening experience for most parents, all treatments in Danish hospitals are free of charge, and the recurrent risk of febrile convulsions for siblings in our study was in accordance with that found in other studies. 4,29 However, parental experience with febrile convulsions may reduce the perceived need for hospitalization. If so, our results may tend to underestimate the recurrence risk in the FC+ cohort.
In conclusion, our data suggest that the etiology of febrile convulsions depends on a genetic susceptibility that can be transmitted through both parents, and they corroborate the hypothesis that multiple febrile convulsions constitute a separate etiological entity. Furthermore, gestational age at birth and fetal growth may play a causal role in the etiology of febrile convulsions.
We thank Allen J. Wilcox, Michael Væth, and Niels Jørgen Secher for constructive comments and suggestions.
1. Proceedings - Consensus Development Conference on Febrile Seizures, National Institutes of Health, May 19–21, 1980. Epilepsia
2. Forsgren L, Sidenvall R, Blomquist HK, Heijbel J, Nystrom L. An incident case-referent study of febrile convulsions in children: genetical and social aspects. Neuropediatrics 1990; 21: 153–159.
3. Verity CM, Butler NR, Golding J. Febrile convulsions in a national cohort followed up from birth. I–Prevalence and recurrence in the first five years of life. BMJ (Clin Res Ed) 1985; 290: 1307–1310.
4. Hauser WA, Annegers JF, Anderson VE, Kurland LT. The risk of seizure disorders among relatives of children with febrile convulsions. Neurology 1985; 35: 1268–1273.
5. Nelson KB, Ellenberg JH. Prenatal and perinatal antecedents of febrile seizures. Ann Neurol 1990; 27: 127–131.
6. Berg AT, Shinnar S, Shapiro ED, Salomon ME, Crain EF, Hauser WA. Risk factors for a first febrile seizure: a matched case-control study. Epilepsia 1995; 36: 334–341.
7. Peiffer A, Thompson J, Charlier C, et al. A locus for febrile seizures (FEB3) maps to chromosome 2q23–24. Ann Neurol 1999; 46: 671–678.
8. Lopes-Cendes I, Scheffer IE, Berkovic SF, Rousseau M, Andermann E, Rouleau GA. A new locus for generalized epilepsy with febrile seizures plus maps to chromosome 2. Am J Hum Genet 2000; 66: 698–701.
9. Wallace RH, Wang DW, Singh R, et al. Febrile seizures and generalized epilepsy associated with a mutation in the Na+
-channel beta1 subunit gene SCN1B. Nat Genet 1998; 19: 366–370.
10. Berkovic SF, Scheffer IE. Febrile seizures: genetics and relationship to other epilepsy syndromes. Curr Opin Neurol 1998; 11: 129–134.
11. Corey LA, Berg K, Pellock JM, Solaas MH, Nance WE, DeLorenzo RJ. The occurrence of epilepsy and febrile seizures in Virginian and Norwegian twins. Neurology 1991; 41: 1433–1436.
12. Schiottz Christensen E. Genetic factors in febrile convulsions. An investigation of 64 same-sexed twin pairs. Acta Neurol Scand 1972; 48: 538–546.
13. Phillips DIW. Twin studies in medical research: can they tell us whether diseases are genetically determined? Lancet 1993; 341: 1008–1009.
14. Olsen J, Schmidt MM, Christensen K. Evaluation of nature-nurture impact on reproductive health using half-siblings. Epidemiology 1997; 8: 6–11.
15. Rothman KJ. Causes. Am J Epidemiol 1976; 104: 587–592.
16. Andersen TF, Madsen M, Jorgensen J, Mellemkjoer L, Olsen JH. The Danish National Hospital Register. Dan Med Bull 1999; 46: 263–268.
17. Knudsen LB. The Danish Fertility Database. Dan Med Bull 1998; 45: 221–225.
18. Skjaerven R, Gjessing HK, Bakketeig LS. New standards for birth weight by gestational age using family data. Am J Obstet Gynecol 2000; 183: 689–696.
19. Rich SS, Annegers JF, Hauser WA, Anderson VE. Complex segregation analysis of febrile convulsions. Am J Hum Genet 1987; 41: 249–257.
20. Lie RT, Wilcox AJ, Skjaerven R. A population-based study of the risk of recurrence of birth defects. N Engl J Med 1994; 331: 1–4.
21. Christensen K, Schmidt MM, Vaeth M, Olsen J. Absence of an environmental effect on the recurrence of facial- cleft defects. N Engl J Med 1995; 333: 161–164.
22. Basso O, Olsen J, Christensen K. Recurrence risk of congenital anomalies - the impact of paternal, social, and environmental factors. Am J Epidemiol 1999; 150: 598–604.
23. Basso O, Olsen J, Johansen AM, Christensen K. Change in social status and risk of low birth weight in Denmark: population based cohort study. BMJ 1997; 315: 1498–1502.
24. McCormick MC. The contribution of low birth weight to infant mortality and childhood morbidity. N Engl J Med 1985; 312: 82–90.
25. Sorensen HT, Sabroe S, Olsen J, Rothman KJ, Gillman MW, Fischer P. Birth weight and cognitive function in young adult life: historical cohort study. BMJ 1997; 315: 401–403.
26. Greenwood R, Golding J, Ross E, Verity C. Prenatal and perinatal antecedents of febrile convulsions and afebrile seizures: data from a national cohort study. Paediatr Perinat Epidemiol 1998; 12: 76–95.
27. Van den Berg BJ, Yerushalmy J. Studies on convulsive disorders in young children. I. Incidence of febrile and nonfebrile convulsions by age and other factors. Pediatr Res 1969; 3: 298–304.
28. Forsgren L, Sidenvall R, Blomquist HK, Heijbel J, Nystrom L. Pre- and perinatal factors in febrile convulsions. Acta Paediatr Scand 1991; 80: 218–225.
29. Doose H, Maurer A. Seizure risk in offspring of individuals with a history of febrile convulsions. Eur J Pediatr 1997; 156: 476–481.