SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References

Objective To investigate the impact of epilepsy and antiepileptic drugs on length of gestation and anthropometric measures of the newborn.

Design Cohort study based on questionnaires mailed to all pregnant women who attended for prenatal care at our department from August 1989 to January 1997.

Setting Department of Obstetrics and Gynaecology at Aarhus University Hospital, Denmark.

Participants One hundred and ninety-three singleton pregnancies in women with epilepsy were compared with 24,094 singleton pregnancies in women without epilepsy.

Main outcome measures Preterm delivery, small for gestational age, mean gestational age, gestational age-adjusted birthweight, head circumference, and body length.

Results Children of women with epilepsy who smoked had lower gestational age and were at increased risk of preterm delivery (OR 3.4; 95% CI 1.8–6.5), compared with children born by nonepileptic women who smoked. Birthweight adjusted for gestational age was reduced by 102 g (95% CI 40–164) in women with epilepsy, and the risk of delivering a child who was small for gestational age was increased (adjusted OR 1.9, 95% CI 1.3–2.7), compared with women without epilepsy. Newborn babies of women with epilepsy treated by drugs had a reduced adjusted birthweight (208 g, 95% CI 116–300), head circumference (0.4 cm, 95% CI 0.0.0.7), and body length (0.5 cm, 95% CI 0.1–1.0), compared with the newborn infants of women without epilepsy.

Conclusions Women with epilepsy who smoked were at increased risk of preterm delivery compared with healthy smokers. Children of women with drug treated epilepsy had lower birthweight, length, and head circumference than children of women without epilepsy.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References

Women with epilepsy are considered at high risk in pregnancy1. Attention has mainly been given to the teratogenicity of anticonvulsants and congenital malformations, and a number of studies have reported an increased risk2,3. Much less is known about the association between epilepsy and preterm delivery and intrauterine growth restriction, which are frequently occurring predictors of childhood mortality and morbidity4. Furthermore, the consequences of low birthweight seem to go far beyond childhood5.

Children of women with epilepsy may be at increased risk of preterm delivery and intrauterine growth restriction for a number of reasons. Factors related to maternal disease include genetic aspects, seizures during prenancy, and exposure to antiepileptic drugs. Other possible causes include environmental factors that may be associated with epilepsy (e.g. maternal smoking and alcohol consumption). However, the results of previous studies are inconsistent, and adjustment of anthropometric measures for gestational age and adjustment for confounding variables were rarely carried out.

We set out to investigate the impact of maternal epilepsy and anticonvulsant drugs on length of gestation and anthropometric measures at birth in a population of about 25,000 pregnant women. Emphasis was put on adjustment for potential confounders and investigation of modification of the effect by maternal lifestyle.

METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References

All Danish-speaking women who attended for antenatal care at Aarhus University Hospital from July 1989 to January 1997 (n= 32,365) were asked to complete a questionnaire regarding medical and obstetric history (97%) completed the questionnaire. Mean gestational age on completing the questionnaire was 15 weeks (SD 3.5 weeks). Of the women who returned the questionnaire, 28,068 were delivered in our maternity unit. All women who reported chronic disease other than epilepsy were excluded from the study. Further restriction was made to singleton pregnancies with known sex and birthweight of the child, leaving 24,287 pregnancies in 19,460 women for analysis.

Maternal epilepsy was reported in the questionnaire. Women were categorised as users or nonusers of anticonvulsant drugs, by self-reported daily intake of any anticonvulsant drug during the first trimester. Therapy was further categorised into monotherapy and polytherapy, and the monotherapies were described according to the drug used. To validate diagnoses, hospital records from the University Hospital Department of Neurology were reviewed by one of the authors (M.D.), a senior specialist in the treatment of epilepsy, who was blinded to the obstetric outcome. The type of epilepsy was classified according to the guidelines of the International League Against Epilepsy6.

Gestational age was estimated by ultrasound measurement of the fetal biparietal diameter before 20 weeks of gestation (74% of all pregnancies). If scanning was not performed, gestational age was calculated from the last menstrual period. Women with and without epilepsy had similar frequencies of ultrasound-determined gestational age. Gestational age (days), birthweight (g), and head circumference and body length (cm) were analysed as continuous measures. Dichotomous outcome measures were preterm delivery (< 37 completed weeks), low birthweight (< 2500 g), and small for gestational age (birthweight < 10th centile for children of same sex, born at same gestational week by women without epilepsy in the present cohort). Information about major congenital malformations and induction of labour in term and preterm deliveries was also collected.

Potential confounders included parity, maternal age at delivery, pre-pregnant weight, height, body mass index, smoking habits (number of cigarettes per day), alcohol intake (number of drinks per week), marital status, educational level, working status, and sex of the child. All variables were categorised as in Table 1.

Table 1.  Maternal characteristics in pregnancies in 145 women with epilepsy (n= 193) and pregnancies in 19,460 women without chronic disease (n= 24,094). Values are given as n and n (%)*. LBW = low birthweight.
 Epilepsy no treatment (n= 106)Epilepsy with treatment (n= 87)No chronic disease (n= 24,094)
  1. *Percentages are calculated from nonmissing data.

  2. Percentages of pregnancies with parity < 1.

  3. Smokers defined as smoking ≥1 cigarette/day.

Parity   
  058(55)48(55)12,152(50)
  >148(45)39(45)11,942(50)
  Previous preterm or LBW6(13)3(8)1161(10)
Maternal age (years)   
  <205(5)1(1)408(2)
  20–2433(31)20(23)3721(15)
  25–2935(33)30(34)9742(40)
  30–3421(20)28(32)7358(31)
  <3512(11)8(9)2865(12)
Height (cm)   
  <1609(9)13(15)2016(9)
  160– 16420(19)18(21)5079(22)
  165–16933(31)21(24)7023(30)
  170–17434(32)26(30)6278(27)
  >1759(9)8(9)2972(13)
Missing11725
Pre-pregnant weight (kg)   
  <503(3)4(5)1281(6)
50–5945(43)29(33)9010(39)
60–6936(35)31(36)8654(37)
70–7916(14)12(14)2950(13)
<806(6)11(13)1480(6)
  Missing00719
Body mass index (kg/m2)   
  <2032(30)24(28)6019(26)
  20–2455(52)39(45)13,523(59)
25–2916(15)11(13)2543(11)
  >302(2)12(14)848(4)
Missing111161
Alcohol (drinks/week)   
  <168(69)64(74)15,654(69)
  1–423(24)22(26)6530(29)
5–96(6)0(0)416(2)
  >101(1)0(0)85(.4)
  Missing811409
Smoking habits   
  Nonsmokers57(54)60(69)16,398(69)
  Smokers49(46)27(31)7467(31)
  Missing00229
Education (years)   
  <912(16)12(18)2140(12)
  9–1122(29)24(36)5180(30)
  >1241(55)30(46)10,234(58)
  Missing31216540
Working status   
  Working41(55)35(55)11,847(67)
  Unemployed13(17)11(17)2418(14)
  Welfare payments12(16)13(20)1196(7)
  Student/others9(12)5(8)2114(12)
  Missing31236519
Marital status   
  Married/cohabiting92(88)83(96)22,675(94)
  Living alone12(12)3(4)1389(6)
  Missing2330

The study was approved by the local ethics committee.

Statistical methods

Bivariate associations were tested with Student's t test for continuous outcomes, and the χ2 test for independence within contingency tables for dichotomous outcomes. Adjustment for confounding factors was performed by multivariate linear and logistic regression analyses. All potential confounding variables were coded categories minus 1. They remained in the final model when the estimated risk changed more than 10%7. Interactions were investigated by stratified analyses and multivariate analyses with interaction terms. Additional analyses were carried out after excluding, in turn, women with previous preterm or low birthweight delivery, children with major congenital malformations, and stillbirths. Some women had more than one pregnancy during the study period, and the inclusion of more than one pregnancy per woman in the analyses could influence the risk estimates. This potential bias from dependence between pregnancy outcomes in the same woman was adjusted for by the use of generalised estimating equations8. The correction remained in the final model only if it changed the risk estimates significantly. Statistical significance was defined as a two-sided P value of < 0.05. Adjusted odds ratios (OR) and risk differences are presented with 95% confidence intervals (CI). SPSS software was used for all statistical analyses except the generalised estimating equations where SAS was used.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References

A total of 193 singleton pregnancies were identified in 145 women with epilepsy. In 87 pregnancies, the woman received anticonvulsant drugs; in five of these, the woman started treatment during the first trimester. In another 15 pregnancies the woman had stopped treatment just before pregnancy. Maternal characteristics, lifestyle, and social factors are shown in Table 1, by epilepsy and treatment. The distribution of drug therapy during the first trimester is listed in Table 2. Characteristics of the children are presented in Table 3.

Table 2.  Anticonvulsant therapy during the first trimester in pregnancies of 64 women with epilepsy (n= 87). Values are given as n (%).
 No. of patients
Mono/polytherapy 
Monotherapy73(84)
Polytherapy14(16)
Monotherapy drugs 
Carbamazepine37(51)
Clobazam1(1)
Clonazepam15(21)
Oxcarbazepine7(10)
Phenytoin1(1)
Valproic acid12(16)
Table 3.  Characteristics in children of women with epilepsy (n= 193) and of women without chronic disease (n= 24,094). Values are given as n and n (%).
 Epilepsy without treatment (n= 106)Epilepsy with treatment (n= 87)No chronic disease(n= 24,094)
Sex   
 Male53 (50)46 (53)12,380 (51)
 Female53 (50)41 (47)11,714 (49)
Major congenital malformations   
 No106 (100)83 (95)23,547 (99)
 Yes0 (0)4 (5)280 (1)
 Missing00267
Liveborn   
 Yes104 (98)87 (100)23,979 (100)
 No2 (2)0 (0)115 (0)
Induced labour   
 No87 (82)66 (76)21,033 (87)
 Induced, at term17 (16)20 (23)2775 (12)
 Induced, preterm2 (2)1 (1)286 (1)
Preterm delivery   
 No100 (94)80 (92)23,170 (96)
 Yes6 (6)7 (8)924 (4)
Low birthweight   
 No101 (95)78 (90)23,316 (97)
 Yes5 (5)9 (10)778 (3)
Small for gestational age   
 No89 (84)70 (80)21,773 (90)
 Yes17 (16)17 (20)2321 (10)

Gestational age

Nonsmokers with epilepsy gave birth at the same gestational age as nonsmokers without epilepsy, whereas gestational age was reduced in women with epilepsy who smoked. The reduction in mean gestational age for women treated with anticonvulsant drugs who smoked was 7.7 days (95% confidence interval (CI) 2.8 to 12.6), compared with healthy women who smoked. The risk of preterm delivery was increased only for women with epilepsy who smoked (OR 3.4, 95% CI 1.8–6.5), compared with healthy women who smoked (Tables 4 and 5). Restriction to pregnancies with ultrasound-determined gestational age did not change the risk estimates. The risk tended to increase with the number of cigarettes smoked per day, but small numbers restricted further stratification. The mean number of cigarettes smoked per day and the frequency of women who smoked more than 14 cigarettes per day did not differ between the groups. No other interactions or confounders were identified.

Table 4.  Mean and adjusted difference in gestational age and birthweight, head circumference, and body length in women with epilepsy compared with women without chronic disease. Data on gestational age are stratified by smoking habits. Values are given as mean (SD) or adjusted difference [95% CI].
 All with epilepsy (n= 193)Epilepsy without treatment (n= 106)Epilepsy with treatment(n= 87)No chronic disease (n= 24,094)
  1. *Gestational age adjusted for maternal age, height.

  2. Birthweight adjusted for gestational age, smoking habits, maternal age, pre-pregnant weight.

  3. Dagger;Head circumference adjusted for gestational age, smoking habits, pre-pregnant weight, marital status.

  4. §Body length adjusted for gestational age, smoking habits, pre-pregnant weight, height.

Gestational age (days), nonsmokers*282 (11) 0.6 [−1.5; 2.8]283 (11) 1.1 [−1.9; 4.1]282 (11) 0.1 [−2.8; 3.1]282 (12) reference
Gestational age (days), smokers*274 (15) −5.5 [−8.5; −2.5]276 (13) −4.2 [−7.9; −0.5]272 (18) −7.7 [−12.6; −2.8]280 (13) reference
Birthweight (g)3366 (589) −102 [−164; −40]3437 (605) −15 [−98; 68]3279 (560) −208 [−300; −116]3532 (549) reference
Head circumference (cm)Dagger;34.9 (1.8) 0.0 [−0.3; 0.2]35.2 (1.7) 0.2 [−0.1; 0.5]34.5 (1.8)−0.4 [−0.7; 0.0]35.1 (1.8) reference
Body length (cm)§51.3 (2.9) −0.3 [−0.6; 0.0]51.5 (3.1) −0.1 [−0.5; 0.3]51.2 (2.5) −0.5 [−1.0; −0.1]52.0 (2.5) reference
Table 5.  Adjusted odds ratios for the association between epilepsy with and without treatment and preterm delivery (< 37 completed weeks), low birthweight (LBW) (birthweight < 2500 g), and small for gestational age (SGA) (birthweight < 10th centile for sex and gestational age). Risk estimates for preterm delivery are stratified by smoking habits. The comparison group consisted of women without chronic disease and with same smoking status. Values are given as adjusted difference (95% CI).
 All with epilepsy (n= 193)Epilepsy no treatment (n= 106)Epilepsy with treatment (n= 87)
  1. *LBW risk estimate adjusted for smoking habits.

  2. SGA risk estimate adjusted for smoking habits, maternal age, pre-pregnant weight, and working status.

Preterm delivery   
Nonsmokers0.5 (0.1; 2.0)0.5 (0.1; 3.7)0.5 (0.1; 3.5)
Smokers3.4 (1.8; 6.5)2.3 (0.9; 5, 8)5.7 (2.3; 14.2)
LBW*2.2 (1.3; 3.9)1.3 (0.5; 3.3)3.5 (1.7; 7.0)
SGA1.9 (1.3; 2.7)1.6 (0.9; 2.6)2.3 (1.3; 4.0)

Birthweight

Mean birthweight adjusted for gestational age was reduced by 102 g (95% CI 40–164) in children of women with epilepsy compared with women without epilepsy (Table 4). The birthweight reduction was related to anticonvulsant drugs, because children of women with epilepsy and anticonvulsant therapy, had a gestational age-adjusted birthweight reduction of 208 g (95% CI 116–300). The reduction was most pronounced in children of women who received carbamazepine and oxcarbazepine monotherapy (Table 6). Cessation of anticonvulsant therapy before pregnancy, which occurred weight. Mean birthweight was affected by gestational age, smoking habits, maternal age, and pre-pregnant weight.

Table 6.  Adjusted difference in birthweight in children of women with various anticonvulsant drugs (AED) treatments, compared with children of women with no chronic disease (n= 24,094).
Therapy groupnMean (SD)Adjusted difference (95% CI)*
  1. *Birthweight adjusted for gestational age, smoking habits, maternal age, pre-pregnant weight.

No chronic disease24,0943532 (549)0 (reference)
No AED treatment1063437 (605)−15 (−98; 68)
Carbamazepine monotherapy373303 (507)−257 (−397; −117)
Clonazeparn monotherapy153363 (394)−77 (−297; 143)
Oxcarbazepine monotherapy73063 (700)−363 (−685; −41)
Valproic acid monotherapy123223 (618)−161 (−418; 96)
All monotherapies733274 (527)−215 (−316; −115)
All polytherapies143305 (735)−131 (−359; 97)
Any AED therapy873279 (560)−208 (−300; −116)

The risks of low birthweight and small for gestational age were increased in women with epilepsy (Table 5). The risk of small for gestational age was affected by smoking habits, maternal age, pre-pregnant weight, and working status. The difference in birthweight tended to differ between children of women with epilepsy who smoked and those who did not, but the interaction between epilepsy and smoking was not statistically significant after adjustment for gestational age.

After exclusion in turn of women with previous preterm or low birthweight delivery, delivery of a malformed child and of stillbirths, the results remained unchanged. Inclusion of a correlation structure between pregnancy outcomes for the same woman by a generalised estimating equation algorithm did not change the risk estimates, except for birthweight in women who received anticonvulsant drugs, where the adjusted difference from children of women without chronic disease increased to 301 g (95% CI 184–418).

Other outcomes

Adjusted head circumference and body length were reduced only in children of women treated with anticonvulsant drugs (Table 4). Major congenital malformations were more frequent in children of women who received anticonvulsant drugs (Table 3). The major congenital malformations and drug treatment are listed in Table 7.

Table 7.  List of major congenital malformations in the epilepsy group (n= 4).
MalformationTreatment
Trisomy 21, heart failureCarbamazepine monotherapy (dose unknown)
Talipes equinovarusValproic acid monotherapy (600 mg/day)
MyelomeningoceleValproic acid monotherapy (1200 mg/day)
Hypoplasia of fingers and toesCarbamazepine monotherapy (800 mg/day)

Validation of diagnosis

Hospital records were studied for 104 women (72%) with 138 pregnancies. A verified diagnosis of epilepsy was present in 113 pregnancies; type of epilepsy was unknown in 11 pregnancies, and seizures were considered nonepileptic in three pregnancies (Table 8). Outcomes in the three pregnancies with nonepileptic seizures were children with normal weight, born at term, and not malformed. After restriction to pregnancies in women with a verified diagnosis of epilepsy (n= 124), risk estimates increased for all outcomes in the study. The increased risk of short gestation and reduced birthweight could not be related to any specific type of epilepsy or seizures.

Table 8.  Type of epilepsy in 138 pregnancies in 104 women with neurological records. Values are given as n.
Type of epilepsyWomenPregnancies
Generalised form5373
Focal form3040
Epilepsy, unclassified911
Unknown911
Nonepileptic seizures33

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References

There have been conflicting results in previous studies regarding epilepsy and preterm delivery. Some studies have reported an increased risk9,10 and others have not11–14. Most studies of epilepsy and pregnancy outcome have failed to report on duration of gestation, and methods of estimating gestational age have been described in few studies. Estimation of gestational age from the last menstrual period may be invalidated by early pregnancy bleeding mistaken for the last period, long follicular phases, or irregular periods15. Women with epilepsy may be more prone to such problems, leading to over-estimation of the frequency of preterm delivery because they have a higher incidence of oral contraceptive failure and vaginal bleeding in pregnancy16. However, dating by ultrasound will under-estimate gestational age if fetal head growth is restricted in early pregnancy17, and the relative risk of preterm delivery will be over-estimated. In our study these potential biases were reduced by estimating gestational age by ultrasound scanning before 20 weeks of gestation.

Sabers and Dam18 suggested that an increased risk of preterm delivery might be explained by a higher induction rate. Indeed, we found that induced labour occurred more frequently in women with epilepsy, but the induction rate was higher only for term pregnancies and thus failed to explain the higher frequency of preterm deliveries.

Stratification by smoking habits showed that women with epilepsy who smoked experienced a more than threefold increase in the risk of preterm delivery, compared with healthy smokers. As this risk was even higher among treated women the interaction may be due to a synergistic anti-oestrogen effect from smoking and anticonvulsant therapy; nicotine inhibits the formation of oestrogens, and some anticonvulsant drugs induce their breakdown19,20. No previous study investigated this interaction, and it was not an a priori hypothesis of this study. Thus, it needs to be tested in future cohort studies.

Our finding of a reduction in mean birthweight is in agreement with most previous studies9,12,21,22. However, only few previous studies adjusted birthweight for gestational age21,23. In general, a reduced birthweight may be caused by short gestation, a small growth potential, or intrauterine growth restriction. To serve as an acceptable proxy for growth restriction, birthweight must at least be adjusted for gestational age. We showed that adjustment for gestational age resulted in a decreased birthweight difference.

Adjustment for confounding was carried out in only a few studies22,23. In our study the association between epilepsy and birthweight for gestational age changed slightly after adjustment for smoking habits, maternal age, pre-pregnant weight, and working status. Congenital malformations may bias results related to gestational age and birthweight. Only one previous study excluded congenital malformations before analysis12. Exclusion of malformed children from our analyses did not change the results. We tested a variety of other potential confounders, but residual confounding (e.g. from nutritional factors) cannot be ruled out.

The relationship between anticonvulsant therapy and reduced birthweight may also be due to a drug-induced reduction in the availability of folate. Low serum folate levels have been associated with low birthweight in general populations24, and serum folate may be reduced in women with epilepsy due to anticonvulsant drugs25,26. On the other hand, Hiilesmaa et al.25 found serum folate unrelated to treatment with carbamazepine. Treatment may also indicate the severity of disease, and some authors have suggested that the severity of epilepsy is the main risk factor for stopped treatment a few months before pregnancy gave birth to children with birthweights comparable with those who continued treatment throughout pregnancy. This finding supports a disease-linked, rather than treatment-linked, reason for reduced birthweight. However, the tendency to an association with specific monotherapies suggests that drug-specific factors may be responsible. Unfortunately, information on serum levels of antiepileptic drugs or folate, drug doses, and the number of seizures was not available for our study.

In most previous studies women with epilepsy were identified from hospital records12,14,27, while other studies were based on registers9. Both study designs may underestimate the true prevalence of epilepsy. The estimated prevalence of epilepsy in our study was 0.7%, higher than 0.5–0.6% generally reported in pregnant women9,11. This may be due to inclusion of a high proportion (54%) of untreated women. Studies carried out in neurological and other special clinics may include only severely affected, but also carefully monitored women. This would potentially over-estimate the general risk and make results applicable only to severely affected, carefully monitored women with epilepsy. However, women with epilepsy who fail to attend special clinics may constitute a high risk group. We found that restriction to women with neurological records resulted in slightly higher risk estimates.

Selection bias could explain our results if women with epilepsy who failed to participate in our study were more likely to give birth at term to normal weight babies. However, as the outcome of pregnancy was unknown at the time of study entry, this bias seems unlikely.

Future research should be applied to the interaction between epilepsy and environmental factors, such as vitamin supplementation and maternal smoking. Preconceptional inclusion of women and careful documentation of medication, seizures, and type of epilepsy, which has been carried out in very few studies23, may also contribute to new knowledge on the course of pregnancy in women with epilepsy.

Acknowledgements

The authors would like to thank Professor N. J. Secher and Dr M. Hedegaard at the Perinatal Epidemiological Research Unit who participated with enthusiasm during establishment of the cohort. We would also like to thank Ms J. Frandsen, secretary at Department of Neurology for her assistance, and Dr K. Wisborg, Dr B. B. Nielsen and Dr M. Vestergaard for valuable comments on a previous version of the manuscript. Funding for this study was provided by The Danish Medical Research Council (Grants no. 9701203 and 12–1663).

References

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References