Antiepileptic drug use, folic acid supplementation, and congenital abnormalities: a population-based case–control study
Dr D Kjær, The Danish Epidemiology Science Centre, Department of Epidemiology, Institute of Public Health, University of Aarhus, Vennelyst Boulevard 6, DK-8000 Aarhus C, Denmark. Email firstname.lastname@example.org.
Objective To investigate whether folic acid supplementation in early pregnancy modifies the association between the prevalence of congenital abnormalities in the offspring and maternal use of carbamazepine (CBZ), phenobarbital (PB), phenytoin (PHT), and primidone (PRI).
Design A population-based case–control study.
Setting The Hungarian Case–Control Surveillance of Congenital Abnormalities (HCCSCA) (1980–1996) and its information on children from the Hungarian Congenital Abnormality Registry and the Hungarian National Birth Registry.
Population Children with congenital abnormalities (cases; n= 20 792, of whom 148 had been exposed to antiepileptic drugs [AEDs]) and unaffected children (controls; n= 38 151, of whom 184 had been exposed to AEDs).
Methods Information on drug exposure and background variables for the mothers were collected from antenatal logbooks, discharge summaries, and structured questionnaires completed by the mothers at the time of HCCSCA registration.
Main outcome measures Congenital abnormalities detected at termination of pregnancy, at birth or until 3 months of age according to CBZ, PB, PHT, or PRI exposure at 5–12 weeks from first day of the last menstrual period (LMP), stratified by folic acid supplementation.
Results Compared with children unexposed to AEDs and folic acid, the odds ratio of congenital abnormalities was 1.47 (95% CI 1.13–1.90) in children exposed to AEDs without folic acid supplementation and 1.27 (95% CI 0.85–1.89) for children exposed to AEDs with folic acid supplementation.
Conclusion The results indicate that the risk of congenital abnormalities in children exposed in utero to CBZ, PB, PHT, and PRI is reduced but not eliminated by folic acid supplementation at 5–12 weeks from LMP. The statistical precision in our study is limited due to rarity of the exposures, and further studies are needed.
The prevalence of major congenital abnormalities in newborns exposed to antiepileptic drugs (AEDs) in utero ranges from 3.3 to 9.0%, two- to three-fold higher than that among nonexposed newborns.1–8 The biological mechanism underlying this association is unclear, but folic acid may be involved in the pathogenesis.1 Some AEDs, such as carbamazepine (CBZ), phenobarbital (PB), phenytoin (PHT), and primidone (PRI), alter the folic acid metabolism, and folic acid blood levels decrease with increasing plasma levels of AEDs.9 Although folic acid supplementation reduces the risk of neural tube defects,10–12 three recent studies did not find a protective effect of folic acid supplementation in women taking AEDs.13–15 Pregnant women taking AEDs are nevertheless encouraged to take high doses of folic acid.16
Our study explores whether folic acid supplementation taken early in pregnancy, that is at the time of organogenesis reduces the prevalence of congenital abnormalities among children exposed in utero to CBZ, PB, PHT, and PRI.
Our case–control study is based on data from the Hungarian Case–Control Surveillance of Congenital Abnormalities (HCCSCA), which included children born in Hungary between 1980 and 1996. The HCCSCA obtained information from the Hungarian Congenital Abnormality Registry and the Hungarian National Birth Registry.17 Children without Hungarian citizenship and children who had an unknown address were excluded.
We identified all children diagnosed with congenital abnormalities at termination of pregnancy or within 3 months of birth. Diagnostic data from terminations were included in the Hungarian Congenital Abnormality Registry from 1984 if an ultrasound examination detected a fetal defect. Children with the following conditions were excluded from the HCCSCA: congenital dislocation of the hip based on Ortolani’s click (n= 11 377), congenital inguinal hernia (n= 9641), haemangioma (n= 3755), minor anomalies such as umbilical hernia and hydrocele (n= 4440), Down syndrome (n= 834), and Mendelian and other chromosomal congenital abnormality syndromes when specified (n= 507). In our study, we excluded undescended testes (n= 2051) from the study base. Children with congenital abnormalities (n= 20 792) were divided into several categories of congenital abnormalities: neural tube defects (n= 1202), cardiovascular defects (n= 4479), cleft lip with or without cleft palate (n= 1374), posterior cleft palate (n= 582), hypospadias (n= 3038), talipes (n= 2424), poly/syndactyly (n= 1744), other isolated congenital abnormalities (n= 4600), and multiple congenital abnormalities (n= 1349).
During the study period, two population-based controls were selected from the Hungarian National Birth Registry for each case, except from 1986 to 1992, when three controls were selected. Cases and controls were matched on sex, residence, and calendar week of birth. The controls were excluded if they were already in the Hungarian Congenital Abnormality Registry, had remarks of congenital abnormalities in discharge summaries after birth, or had remarks on congenital abnormalities in the structured questionnaires (see below). The participation rate was 96.3% among cases and 83.1% among controls. Due to different follow-up strategies towards nonrespondent cases and controls, the study group ended up with 38 151 controls and 22 843 cases.18
Information on exposure and potential confounders
Data on drug use, including folic acid supplementation, were obtained from three different sources; two sources with prospectively collected data (the antenatal logbooks that include the pregnancy and hospital discharge summaries written by physicians) and one with retrospectively collected data (structured questionnaires completed by the mothers of the cases and controls at the time of HCCSCA registration).19,20 A child was considered ‘exposed’ to AEDs or folic acid if the intake was noted in any one of these sources. Information on AEDs was partly obtained from the prospective sources (29–46%), and information on folic acid was mainly based on the parental questionnaire.21 Gestational age at time of drug use was estimated by using the first day of the last menstrual period (LMP). Data on maternal age, parity, and sex of the child were obtained from medical records or the parental questionnaires. The children (cases and controls) were divided into four groups according to maternal intake of AEDs and folic acid during the second and/or third gestational month, that is at 5–12 weeks from LMP: (1) no AED use and no folic acid supplementation (reference), (2) no AED use and folic acid supplementation, (3) AED use and no folic acid supplementation, and (4) both AED use and folic acid supplementation.
Prior to the analyses, we decided not to report the association between drugs and specific abnormalities if we had fewer than four cases and controls when drug exposure was stratified by folic acid supplementation. These drugs were valproate (cases = 3, controls = 3), sultiame (cases = 2, controls = 0), mephenytoin (cases = 1, controls = 1), clonazepam (cases = 0, controls = 3), morsuximide (cases = 1, controls = 0), ethosuximide (cases = 2, controls = 2), trimethadione (cases = 1, controls = 0), and phenacemide (cases = 0, controls = 0). Lamotrigine was not used in Hungary during this time period.
Statistical analyses were performed using SAS statistical software version 8.02 (SAS Institute, Cary, NC, USA). Contingency tables were constructed for the main study variables. We compared the use of AEDs among cases and controls in logistic regression models by taking the matching factors into the model. In the logistic regression models, we estimated the odds ratios with 95% CI of specific birth defects among infants whose mothers received AEDs (CBZ, PB, PHT, and PRI) at 5–12 weeks from LMP with or without folic acid supplementation in the same period. The odds ratios were further adjusted for maternal age (<25, 25–29, and 30 years or more) and parity (primiparity and multiparity) since these variables differed among cases and controls. We used stratified analyses to evaluate whether the effect of AEDs on the prevalence of congenital abnormalities was modified by folic acid supplements at 5–12 weeks from LMP. We also estimated the effect modification by folic acid on drug use and congenital abnormalities. The Wald chi-square test was used for the comparison of odds ratios (with 95% CI) and generated the P values for this test. P values below 0.05 were considered to be significant.
Permission to collect data was given by the Ethics Committee in Hungary. All mothers who participated in the HCCSCA provided written informed consent.
Among the 20 792 children with congenital abnormalities and 38 151 children without congenital abnormalities participating in the HCCSCA, 135 cases and 178 controls had been exposed to CBZ, PB, PHT, and PRI in monotherapy at 5–12 weeks from LMP. Of these children, exposure had occurred to CBZ (17 cases, 9 controls), PB (72 cases, 96 controls), PHT (16 cases, 20 controls), and PRI (12 cases, 4 controls) without folic acid supplementation, and exposure had occurred to CBZ (5 cases, 8 controls), PB (28 cases, 36 controls), PHT (6 cases and 12 controls), and PRI (7 cases, 6 controls) with folic acid supplementation. Thirteen cases and six controls took two or more drugs.
A slightly higher prevalence of congenital abnormalities was associated with maternal age greater than 30 years at delivery (OR 1.06; 95% CI 1.02–1.11) and multiparity (OR 1.20; 95% CI 1.14–1.26) (Table 1), and we adjusted for these factors. Overall, the prevalence of congenital abnormalities was lower in children unexposed to AEDs when using folic acid 5–12 weeks from LMP (OR 0.84; 95% CI 0.81–0.87) and higher in children exposed to CBZ, PB, PHT, and PRI at 5–12 weeks from LMP without folic acid supplementation compared with nonexposed children (OR 1.47; 95% CI 1.13–1.90), even when neural tube defects were excluded (OR 1.43; 95% CI 1.09–1.86) (Table 2). The strongest associations with AEDs were seen for cleft lip with or without cleft palate. The association between AEDs and congenital abnormalities tended to be lower among offspring of mothers who took folic acid supplements in early pregnancy compared with offspring of mothers who did not (OR 1.27; 95% CI 0.85–1.89); but these differences in relative measures of association were not statistically significant, except for multiple congenital abnormalities, where the highest prevalence was seen for those who took folic acid (Table 2). The odds ratios for all congenital abnormalities following in utero exposure to individual AEDs were lower when folic acid supplements were used concomitantly (test for interaction; P values from 0.18 to 0.62), except for PB (Table 3). The prevalence of congenital abnormalities increased with increasing number of AEDs used but tended to be lower when folic acid supplements had been used. For AED in monotherapy without folic acid, the odds ratio was 1.38 (95% CI 1.05–1.80) and with folic acid, the odds ratio was 1.21 (95% CI 0.80–1.83) (test for interaction; P= 0.61) (data not shown). For AEDs in polytherapy without folic acid, the odds ratio was 5.23 (95% CI 1.42–19.33) and with folic acid, the odds ratio was 2.42 (95% CI 0.54–10.83). We also analysed odds ratios after exposure to each of the four AEDs and specific abnormalities, but most strata had limited information (data not shown).
Table 1. Characteristics of cases and controls, Hungary (1980–1996)
|Maternal age (years)||<25||10 017||48.2||17 994||47.2|
|Parity||0||12 663||60.9||22 750||59.6|
|Antiepileptic drugs* (5–12 weeks from LMP)||0||20 644||99.3||37 967||99.5|
|Folic acid (5–12 weeks from LMP)||No||15 555||74.8||27 224||71.4|
|Sex||Males||12 864||61.9||24 799||65.0|
Table 2. Odds ratios of congenital abnormalities according to antiepileptic drug and folic acid exposure at 5–12 weeks from LMP, Hungary
|Controls||27 098|| ||10 869|| ||126|| ||58|| |
|All congenital abnormalities||15 449||1.00||5195||0.84 (0.81–0.87)||106||1.47 (1.13–1.90)||42||1.27 (0.85–1.89)|
|All congenital abnormalities except NTD***||14 531||1.00||4923||0.85 (0.81–0.88)||97||1.43 (1.09–1.86)||39||1.26 (0.84–1.89)|
|Neural tube defects||918||1.00||272||0.74 (0.64–0.85)||9||2.11 (1.07–4.17)||3||1.54 (0.48–4.93)|
|Cleft lip ± cleft palate||1008||1.00||347||0.86 (0.76–0.98)||14||2.97 (1.70–5.18)||5||2.38 (0.95–5.94)|
|Posterior cleft palate||430||1.00||146||0.86 (0.71–1.03)||5||2.56 (1.04–6.30)||1||1.13 (0.16–8.21)|
|Hypospadias||2283||1.00||731||0.80 (0.73–0.87)||19||1.76 (1.08–2.85)||5||1.02 (0.41–2.54)|
|Cardiovascular defects||3374||1.00||1078||0.80 (0.74–0.86)||17||1.08 (0.65–1.80)||10||1.41 (0.72–2.70)|
|Poly/syndactyly||1274||1.00||456||0.89 (0.80–0.99)||11||1.83 (0.99–3.40)||3||1.11 (0.35–3.54)|
|Other congenital abnormalities||3346||1.00||1228||0.92 (0.86–0.98)||19||1.21 (0.74–1.96)||7||0.99 (0.44–2.16)|
|Multiple congenital abnormalities||1027||1.00||311||0.76 (0.67–0.86)||5||1.04 (0.42–2.54)||6||2.79 (1.20–6.49)|
Table 3. Antiepileptic drug treatment at 5–12 weeks from LMP and congenital abnormalities stratified by folic acid supplementation at 5–12 weeks from LMP, Hungary (1980–1996)
|Carbamazepine (P= 0.18)***|
|No folic acid||27 224||9||15 555||17||3.34 (1.49–7.49)|
|Folic acid||10 927||8||5237||5||1.31 (0.43–4.02)|
|Phenobarbital (P= 0.46)|
|No folic acid||27 224||96||15 555||72||1.30 (0.96–1.77)|
|Folic acid||10 927||36||5237||28||1.62 (0.99–2.66)|
|Phenytoin (P= 0.62)|
|No folic acid||27 224||20||15 555||16||1.40 (0.72–2.69)|
|Folic acid||10 927||12||5237||6||1.03 (0.39–2.75)|
|Primidone (P= 0.35)|
|No folic acid||27 224||4||15 555||12||5.24 (1.69–16.27)|
|Folic acid||10 927||6||5 237||7||2.51 (0.84–7.47)|
Our data support previous studies showing an association between CBZ, PB, PHT, and PRI and congenital abnormalities1–8 and suggest that a teratogenic effect on single abnormalities may be somewhat reduced by folic acid supplementation taken during organogenesis. However, this possible effect modification was small to moderate and did not reach statistical significance. It takes a very large study or meta-analysis to obtain sufficient power to detect a true effect modification of this size. The lack of a statistically significant effect modification is also compatible with no effect modification as would be expected if the causal pathways between AEDs and congenital abnormalities were unrelated to depletion of folic acid.22
Few studies have assessed the effect of folic acid supplementation among women treated with AEDs.13–15 Hernandez-Diaz et al. studied folic acid antagonists, that is AEDs (CBZ, PB, PHT, and PRI) and dihydrofolate reductase inhibitors (aminopterin, methotrexate, sulfasalazine, pyrimethamine, triamterene, and trimethoprim). They found an increased prevalence of neural tube defects, cardiovascular defects, oral clefts, and urinary tract abnormalities, and folic acid supplementation did not modify the association between AEDs and congenital abnormalities.13,14 Meijer et al.15 investigated the association between folic acid antagonists exposure in the first 10 weeks of pregnancy and the prevalence of congenital abnormalities. No increased prevalence of congenital abnormalities was found for the whole group of folic acid antagonists, but exposure to AEDs (CBZ, PB, PHT, PRI, valproate, and lamotrigine) increased the prevalence of congenital abnormalities significantly, especially heart anomalies, neural tube defects, and limb reduction defects. Folic acid supplementation did not modify the association between AEDs and congenital abnormalities.15
Randomised placebo-controlled trials,10 controlled trials,11 and noncontrolled intervention studies12 of high- and low-risk women have shown that folic acid taken before or early in pregnancy reduces the prevalence of primary and recurrent neural tube defects and other congenital abnormalities. Folic acid is essential for the biosynthesis of many compounds including amino acids.23 Recent studies have shown that maternal antibodies, that bind to folic acid receptors and block the cellular uptake of folic acid, may reduce the beneficial effect of periconceptional use of folic acid.22 The reduction in congenital abnormalities is most pronounced for neural tube defects, and one study has questioned the beneficial effect of folic acid in preventing congenital abnormalities of the nonneural tube type.24 Our finding of an increased prevalence of multiple congenital abnormalities associated with folic acid supplementation may support this conclusion.
The biological mechanism underlying the association between AEDs and congenital abnormalities is unknown. Certain AEDs, including CBZ, PB, PHT, and PRI, influence folic acid absorption.14 CBZ and PHT induce the formation of epoxide intermediates,9,25 which may interfere with DNA synthesis and organogenesis.25 However, whether this is affected by folic acid supplementation is unknown.
The recommended amount of folic acid for women with epilepsy is under debate.16,26 The randomised and nonrandomised trials of folic acid use in women in the general population published during the 1990s used 4 mg/day,10 0.8 mg/day,11 or 0.4 mg/day.12 In the UK, women with epilepsy are recommended to take 5 mg of folic acid per day if using AEDs.27 In Canada, America, Norway, and Denmark, they are advised to take at least 0.4 mg of folic acid per day early in pregnancy and before conception, but 4–5 mg/day if they are using older AEDs.16 In the Norwegian part of the EURAP study, Nakken et al. found that among 263 pregnant women, half the women (48%) had been using 4 mg of folic acid per day in early pregnancy, including during treatment with the latest AEDs.26
In our study, the information on dose of folic acid was available only for a subgroup of 600 women.28 The data indicated that large doses of folic acid were taken; 22.5% of the women took 3 mg, 68.8% took 6 mg, and 8.9% took 9 mg folic acid per day. Our study did not have sufficient data on folic acid dose to improve the recommendation. More information will emerge from studies using prospectively collected data, for example EURAP, the North American Antiepileptic Drug Pregnancy Registry, and the UK Epilepsy Pregnancy Registry.29–31
It is possible that the underlying disease rather than the intake of AEDs increases the prevalence of congenital abnormalities. Several studies have addressed but have not corroborated this hypothesis.1,2,4,32 In a meta-analysis, ten cohort and case–controls studies of abnormalities in children of women with epilepsy, exposed (n= 1443) or unexposed (n= 400) to AEDs, were compared with the outcome of children of women without epilepsy (n= 2492).1 The adjusted odds ratio of major abnormalities in children of women with untreated epilepsy was not increased compared with children of women without epilepsy (OR 0.99; 95% CI 0.49–2.01).1 This suggests that it is the AEDs rather than the disease that causes the congenital abnormalities, although untreated cases of epilepsy are expected to be less severely affected than those who are treated.
AEDs are used for a number of indications other than epilepsy, including chronic pain, alcohol withdrawal symptoms, and bipolar disorders. Unfortunately, we were unable to differentiate between women with epilepsy and women using AEDs for other indications; this may reduce the confounding by indication from epilepsy.
Our results may be confounded by factors beyond our control since the women were not assigned to folic acid treatment by randomisation.17 If women suffering from more severe disease take folic acid supplementation more often and if the disease causes birth defects, we may underestimate the protective effect of folic acid supplementation due to confounding by indication. Women with a family history of congenital abnormalities may be more likely to receive folic acid supplementation than those without this history, and this could attenuate a true modifying effect of folic acid supplementation. Pregnancy planners may be more likely to receive folic acid supplementation and have a lower risk of congenital abnormalities due to a healthy lifestyle and optimal drug treatment than those with an unplanned pregnancy.33 This would intensify a possible modifying effect of folic acid. If taking folic acid correlates with other behaviours that may reduce birth defects, we are overestimating the protective effect of folic acid supplementation.
Case–control studies are vulnerable to recall and selection bias.20 It is to be expected that mothers of cases would recall drug use better than controls in their search for a cause, and this would bias results towards higher odds ratios. However, if cases more often report drug intake incorrectly close to the time of interviewing, rather than at organogenesis, this might undermine an association for drugs taken at 5–12 weeks from LMP.20 However, differential recall of long-term drug use, such as AEDs, is less common than for short-term drug use,20 and the drugs under study are probably used as a long-term treatment, with a minimum of recall bias.
Another potential source of bias is the difference between participation rates among the mothers of cases and controls (96.3 and 83.1%, respectively). Part of the difference was due to different follow-up strategies of nonresponding mothers of cases and controls. The Ethics Committee in Hungary allowed a regional nurse to help all nonresponding mothers of cases (n= 8962) to fill in the questionnaires, while such permission was given for only 200 nonresponding mothers of controls.18
A validation study of these 200 selected nonresponders showed, however, no large differences in the use of common drugs between respondent and nonrespondent mothers of controls.21 This validation study furthermore showed that one-quarter of the drugs used was not recorded in the prospectively collected data (antenatal logbooks and medically recorded discharge summaries). Thus, it was necessary to add retrospective information on drug use involving a risk of recall bias.21
Since prenatal diagnostic methods are more often performed in high-risk pregnancies (i.e. women with epilepsy) than low-risk pregnancies, and since detection of major congenital abnormalities may lead to termination of the pregnancy, the association between AEDs and abnormalities may be underestimated in studies including liveborn babies only.34 To keep such bias to a minimum, we included terminations that were induced after prenatal detection of a fetal defect.
In conclusion, our results indicate that the risk of congenital abnormalities in children exposed in utero to CBZ, PB, PHT, and PRI is reduced but not eliminated by folic acid supplementation. The statistical precision in our study is limited, so further studies are needed to confirm these results.
The study was supported by the Lennart Grams Memorial Foundation, the Danish Epilepsy Society; the Grete Jonsens Foundation, the Danish Epilepsy Association; the Danish Pharmaceutical Association (Apotekerfonden af 1991), and the Obel Family Foundation (Den Obelske Familiefond).