Monotherapy treatment of epilepsy in pregnancy: congenital malformation outcomes in the child

Abstract

Background

Prenatal exposure to certain anti‐seizure medications (ASMs) is associated with an increased risk of major congenital malformations (MCM). The majority of women with epilepsy continue taking ASMs throughout pregnancy and, therefore, information on the potential risks associated with ASM treatment is required.

Objectives

To assess the effects of prenatal exposure to ASMs on the prevalence of MCM in the child.

Search methods

For the latest update of this review, we searched the following databases on 17 February 2022: Cochrane Register of Studies (CRS Web), MEDLINE (Ovid, 1946 to February 16, 2022), SCOPUS (1823 onwards), and ClinicalTrials.gov, WHO International Clinical Trials Registry Platform (ICTRP). No language restrictions were imposed.

Selection criteria

We included prospective cohort controlled studies, cohort studies set within pregnancy registries, randomised controlled trials and epidemiological studies using routine health record data. Participants were women with epilepsy taking ASMs; the two control groups were women without epilepsy and untreated women with epilepsy.

Data collection and analysis

Five authors independently selected studies for inclusion. Eight authors completed data extraction and/or risk of bias assessments. The primary outcome was the presence of an MCM. Secondary outcomes included specific types of MCM. Where meta‐analysis was not possible, we reviewed included studies narratively.

Main results

From 12,296 abstracts, we reviewed 283 full‐text publications which identified 49 studies with 128 publications between them. Data from ASM‐exposed pregnancies were more numerous for prospective cohort studies (n = 17,963), than data currently available for epidemiological health record studies (n = 7913). The MCM risk for children of women without epilepsy was 2.1% (95% CI 1.5 to 3.0) in cohort studies and 3.3% (95% CI 1.5 to 7.1) in health record studies. The known risk associated with sodium valproate exposure was clear across comparisons with a pooled prevalence of 9.8% (95% CI 8.1 to 11.9) from cohort data and 9.7% (95% CI 7.1 to 13.4) from routine health record studies. This was elevated across almost all comparisons to other monotherapy ASMs, with the absolute risk differences ranging from 5% to 9%. Multiple studies found that the MCM risk is dose‐dependent. Children exposed to carbamazepine had an increased MCM prevalence in both cohort studies (4.7%, 95% CI 3.7 to 5.9) and routine health record studies (4.0%, 95% CI 2.9 to 5.4) which was significantly higher than that for the children born to women without epilepsy for both cohort (RR 2.30, 95% CI 1.47 to 3.59) and routine health record studies (RR 1.14, 95% CI 0.80 to 1.64); with similar significant results in comparison to the children of women with untreated epilepsy for both cohort studies (RR 1.44, 95% CI 1.05 to 1.96) and routine health record studies (RR 1.42, 95% CI 1.10 to 1.83). For phenobarbital exposure, the prevalence was 6.3% (95% CI 4.8 to 8.3) and 8.8% (95% CI 0.0 to 9277.0) from cohort and routine health record data, respectively. This increased risk was significant in comparison to the children of women without epilepsy (RR 3.22, 95% CI 1.84 to 5.65) and those born to women with untreated epilepsy (RR 1.64, 95% CI 0.94 to 2.83) in cohort studies; data from routine health record studies was limited. For phenytoin exposure, the prevalence of MCM was elevated for cohort study data (5.4%, 95% CI 3.6 to 8.1) and routine health record data (6.8%, 95% CI 0.1 to 701.2). The prevalence of MCM was higher for phenytoin‐exposed children in comparison to children of women without epilepsy (RR 3.81, 95% CI 1.91 to 7.57) and the children of women with untreated epilepsy (RR 2.01. 95% CI 1.29 to 3.12); there were no data from routine health record studies. Pooled data from cohort studies indicated a significantly increased MCM risk for children exposed to lamotrigine in comparison to children born to women without epilepsy (RR 1.99, 95% CI 1.16 to 3.39); with a risk difference (RD) indicating a 1% increased risk of MCM (RD 0.01. 95% CI 0.00 to 0.03). This was not replicated in the comparison to the children of women with untreated epilepsy (RR 1.04, 95% CI 0.66 to 1.63), which contained the largest group of lamotrigine‐exposed children (> 2700). Further, a non‐significant difference was also found both in comparison to the children of women without epilepsy (RR 1.19, 95% CI 0.86 to 1.64) and children born to women with untreated epilepsy (RR 1.00, 95% CI 0.79 to 1.28) from routine data studies. For levetiracetam exposure, pooled data provided similar risk ratios to women without epilepsy in cohort (RR 2.20, 95% CI 0.98 to 4.93) and routine health record studies (RR 0.67, 95% CI 0.17 to 2.66). This was supported by the pooled results from both cohort (RR 0.71, 95% CI 0.39 to 1.28) and routine health record studies (RR 0.82, 95% CI 0.39 to 1.71) when comparisons were made to the offspring of women with untreated epilepsy. For topiramate, the prevalence of MCM was 3.9% (95% CI 2.3 to 6.5) from cohort study data and 4.1% (0.0 to 27,050.1) from routine health record studies. Risk ratios were significantly higher for children exposed to topiramate in comparison to the children of women without epilepsy in cohort studies (RR 4.07, 95% CI 1.64 to 10.14) but not in a smaller comparison to the children of women with untreated epilepsy (RR 1.37, 95% CI 0.57 to 3.27); few data are currently available from routine health record studies. Exposure in utero to topiramate was also associated with significantly higher RRs in comparison to other ASMs for oro‐facial clefts. Data for all other ASMs were extremely limited. Given the observational designs, all studies were at high risk of certain biases, but the biases observed across primary data collection studies and secondary use of routine health records were different and were, in part, complementary. Biases were balanced across the ASMs investigated, and it is unlikely that the differential results observed across the ASMs are solely explained by these biases.

Authors' conclusions

Exposure in the womb to certain ASMs was associated with an increased risk of certain MCMs which, for many, is dose‐dependent.

Summary of findings

Background

This review is an update of the Cochrane Review first published in 2004 (Adab 2004), and last updated in 2016 (Weston 2016).

Description of the condition

Epilepsy is a common neurological disorder with a lifetime prevalence of 7.60 per 1000 persons (Fiest 2017). A significant number of women with epilepsy will be in their childbearing years (NICE 2022) and, of these, approximately 0.5% to 0.6% of all annual pregnancies are reportedly exposed to an anti‐seizure medication (ASM) in utero (Man 2012, NICE 2022). ASM treatment of epilepsy in the childbearing years requires careful optimisation to improve maternal outcomes whilst minimising, where possible, foetal risks. Research demonstrates an association between children born to women with epilepsy treated with ASMs and an increased risk of major congenital malformations, including cardiac, neural tube and craniofacial defects (EURAP 2018; Jentink 2010a; Meador 2008).

Description of the intervention

ASMs are the most common treatment for epilepsy, and most women with epilepsy require treatment continuation during pregnancy.

How the intervention might work

ASMs readily cross the placenta from the mother into the foetusus (Brent 2004; Tetro 2017). Prospective observational studies (e.g. Milan Study 1999), registry‐based studies (e.g. Tomson 2011), case‐control studies (Jentink 2010a), and epidemiological studies using datasets of routine health records (e.g. Denmark Health Record Registers) provide evidence of an association between ASM treatment and an increased prevalence of major congential malformations. The level of risk varies for different types of ASM, with first trimester valproate (VPA) exposure associated with the largest increase in prevalence (EURAP 2018; Meador 2006; Milan Study 1999; North American Epilepsy and Pregnancy Register; UK and Ireland Epilepsy and Pregnancy Register). The mechanisms through which prenatal exposure to ASMs are associated with an increased prevalence of major malformations likely differs by treatment type and may be multifactorial.

This review investigates the outcomes for monotherapy treatment with different ASMs to identify currently available evidence on which to base treatment decisions.

Why it is important to do this review

The decision to continue ASM treatment during pregnancy requires taking a risk‐benefit decision. On the one hand, there is the potential risk posed to the foetus when the medication is a teratogen yet on the other hand, there is the health and well‐being of the mother, who requires treatment throughout her pregnancy to minimise the risk of seizures (Tomson 2015); the choice of ASM depends on the type of epilepsy and the seizures (Marson 2007). A lack of knowledge regarding foetal safety limits treatment options for women with epilepsy in their childbearing years, as women and their doctors may avoid ASMs with limited data. Conversely, a lack of evidence may lead to an ASM with a higher foetal risk profile being used extensively, prior to a full understanding of its risks.

While a number of studies indicate a teratogenic risk from certain ASMs, there are conflicting results regarding the degree of risk and the types of malformations associated with specific ASMs. Data are slow to accumulate and an earlier version of this review (Weston 2016) found extremely limited data on ASMs with a decade or more of clinical use. Such a lack of evidence makes it difficult to counsel women about treatment choices before or during pregnancy. There is, therefore, a clear need for a systematic review and meta‐analysis of existing data to inform these decisions. Randomised controlled trials (RCTs) would provide the most reliable evidence about the effects of ASMs in pregnancy, but are essentially precluded by ethical considerations and logistical challenges pertaining to study design, recruitment and interpretation.

In view of this, we performed a systematic review of all available evidence including registry‐based, prospective cohort studies, RCTs and epidemiological studies using routine health record databases. At the protocol stage, we decided not to include malformation case‐control studies (e.g. Jentink 2010a; Jentink 2010b) due to the substantial differences in the approach in these studies and how these methods compare to prospective observational cohort studies. This decision is discussed further in Overall completeness and applicability of evidence. This review is an update of two previous reviews (Adab 2004; Weston 2016). Evidence from this review, along with the related review by the same Cochrane team (Bromley 2014), will aid the decisions that clinicians and women with epilepsy have to make about the treatment of epilepsy during the potential childbearing years.

Objectives

To assess the effects of prenatal exposure to commonly prescribed ASMs on the prevalence of major congenital malformations in the child.

This review examines the association between specific ASM exposures and the prevalence of major congenital malformations compared to the general population or unexposed pregnancies in women with epilepsy. It also compares the prevalence of specific major congenital malformations types across the ASM treatment groups.

Methods

Criteria for considering studies for this review

Types of studies

We considered the following types of studies.

Randomised controlled trials (RCTs). These studies included women with epilepsy who were randomised to a particular ASM prior to conception. The intervention group(s) comprised women with epilepsy taking ASM monotherapy.Prospective observational cohort studies. These included consecutive participants whose clinical information was collected prior to the birth of the child. The intervention group(s) comprised women with epilepsy treated with ASM monotherapy.Registry studies. These involve the collection of data from a wide region, country or number of countries, and recruitment is often based on self‐referral or clinician‐referral, leading to non‐sequential case ascertainment. We considered both disease‐based registries (e.g. pregnancy and epilepsy registries) and industry‐sponsored product registry datasets. Pregnant women with epilepsy prescribed ASM monotherapy were recruited prospectively prior to childbirth.Population‐based routine health record datasets. These studies utilise data collected for routine health monitoring, administrative or reimbursement reasons for entire national populations or specific populations (e.g. medical insurance databases). Individual recruitment of participants is not required. The intervention group(s) comprised women with epilepsy taking ASM monotherapy.

Types of participants

Pregnant women with epilepsy taking a single ASM of interest were eligible for the intervention group.

Participants eligible for the comparator groups were:

pregnant women with epilepsy taking an ASM;pregnant women with epilepsy taking no ASM; orpregnant women who do not have epilepsy.

We excluded studies reporting ASM use solely in pregnant women with other conditions (e.g. mood disorders, pain). We included studies involving women taking ASMs for epilepsy and other conditions if the non‐epilepsy conditions accounted for 30% or less of the total treatment group. This percentage criterion was increased from the previous review to accommodate data from population healthcare datasets, which often include a wider group of participant indications.

Types of interventions

Intervention group

Women with epilepsy who received any of the following ASMs as monotherapy: acetazolmide, brivaracetam, bromide, carbamazepine, cenobamate, clomethiazole, clonazepam, clorazepate, diazepam, dimethyloxazolidinedione, eslicarbazepine, ethosuximide, estazolam, felbamate, flunarizine, gabapentin, lacosamide, lamotrigine, levetiracetam, lorazepam, magnesium sulphate, medazepam, methylphenobarbital, mephenytoin, meprobamate, methazolamide, methsuximide, methyloxazepam, midazolam, nimetazepam, nitrazepam, oxcarbazepine, perampanel, phenobarbitone, phenytoin, primidone, pregabalin, remacemide, retigabine, rufinamide, sodium valproate, stiripentol, sulthiame, tiagabine, topiramate, trimethadione, trifluoromethoxy benzothiazole, valnoctamide, vigabatrin, or zonisamide.

Comparator groups

We used two separate types of comparator groups in this review, as currently there is no clear evidence regarding the reliability of combining data from these two different groups. The two comparator groups are:

controls: women with a diagnosis of epilepsy who were not taking ASMs and women without epilepsy.comparator treatment: women with epilepsy treated with ASM monotherapy, evaluated in subgroup analyses to enable treatment comparisons.

Types of outcome measures

Primary outcomes

Major congenital malformations

The proportion of children who present with any type of major congenital malformation (as defined by study authors). Major malformations are structural abnormalities of the body or organs present from birth and which require intervention (e.g., corrective surgery) or have a significant level of impact on the child's daily functioning (EUROCAT).

Secondary outcomes

Specific major congenital malformations

The proportion of children who present with the following specific major congenital malformations by area of the body.

Neural tube malformations.Cardiac malformations.Oro‐facial cleft/craniofacial malformation.Skeletal or limb malformations.

We chose the above disorders because they are important major malformations associated with exposure to ASMs in utero, because these are the most prevalent congenital malformations in the general population (ref: https://eu‐rd‐platform.jrc.ec.europa.eu/eurocat/eurocat‐data/prevalence_en), and because of the availability of data within the included studies. When extracting data from included studies, we compiled a list of all the specified malformations. Author JCS, a clinical geneticist, then reviewed the list and classified the items into one of the four specific malformation categories.

Search methods for identification of studies

Electronic searches

Searches for the original review were run in January 2012. Subsequent searches were run in March 2013, May 2014, and September 2015. For the latest update, we searched the following databases on 17 February 2022:

Cochrane Register of Studies (CRS Web), using the search strategy set out in Appendix 1;MEDLINE (Ovid, 1946 to February 16, 2022) using the search strategy set out in Appendix 2;SCOPUS (1823 onwards) using the search strategy set out in Appendix 3;ClinicalTrials.gov using the search strategy set out in Appendix 4;WHO International Clinical Trials Registry Platform (ICTRP) using the search strategy set out in Appendix 5.

CRS Web includes randomised or quasi‐randomised, controlled trials from PubMed, Embase, ClinicalTrials.gov, the World Health Organization International Clinical Trials Registry Platform (ICTRP), the Cochrane Central Register of Controlled Trials (CENTRAL), and the Specialized Registers of Cochrane Review Groups including Epilepsy. In MEDLINE (Ovid), the coverage end date always lags a few days behind the search date. Previously we also searched Embase, Pharmline and Reprotox.

We did not impose any language restrictions in the search and, when necessary, we obtained translations of articles written in languages other than English.

Searching other resources

We reviewed conference abstracts from neurology meetings published from 2010 to 2022, including abstracts from the International League Against Epilepsy meetings (American Epilepsy Society, International Epilepsy Congress, European Congress on Epileptology, Asian and Oceanian Epilepsy Congress and Latin American Congress on Epilepsy) and Teratology meetings (Teratology Society and European Teratology Society). Where possible, we linked abstracts to published datasets or categorised them as awaiting classification.

We cross‐matched reference lists of original research and review articles to the studies generated from the electronic searches. We handsearched reference lists of recent review articles and contacted lead and corresponding authors in the area for any relevant unpublished material.

Data collection and analysis

Selection of studies

Five authors (RB, JW, JG, KE, RMcG) reviewed the titles and abstracts of articles highlighted by the searches and removed studies that obviously did not meet the inclusion criteria. Four authors (RB, JW, KE, RMcG) used full‐text reports to determine study eligibility. We discussed disagreements and sought the opinion of a third author (JG, CJ, RB), when necessary. Multiple reports from single studies are common in this field. To ensure that each cohort was represented only once in our analysis, therefore to avoid double‐counting the population across papers of included studies, we linked studies by recruitment date and sought confirmation from authors whether reports referred to single study populations. Where this was unclear, we contacted study authors for clarification.

Data extraction and management

Eight authors (RB, JW, NA, JG, AM, KE, RMcG, SK, CJ) undertook data extraction of the included studies. We used pre‐standardised electronic data extraction forms that members of the review team piloted and then amended, where necessary. We then cross‐checked data extraction. All entries into RevMan were also double‐checked.

Assessment of risk of bias in included studies

Due to the observational design of the majority of the studies, we utilised the Risk Of Bias In Non‐Randomized Studies ‐ of Interventions (ROBINS‐I) tool which the Cochrane Non‐Randomised Studies Methods Group has developed (Sterne 2016). The ROBINS‐I tool for assessing risk of bias examines bias in the domains of confounding, selection, treatment classification, missing data, measurement and reported results. ROBINS‐I uses signalling questions on a four‐point scale to determine level of bias in specific elements of biases for each of these domains. Overall domain bias ratings are then classed as low, moderate, serious, critical or no information.

ROBINS‐I was developed for treatment studies and not pharmacovigilance studies, where the person taking the medication (the mother) is not the same person in which the outcome can occur (the child). Therefore, ROBINS‐I needed to be adapted for use in this review. The adaption was led by author RB with input from other authors. Important confounder and mediator variables were selected based on published evidence of an association both in the general population and specifically in investigations regarding in utero ASM exposure and congential malformation outcomes. See Appendix 6 for further information. Eight authors completed risk of bias ratings (RB, JW, NA, JG, AM, KE, SK, MBD). Each included study was reviewed by two independent raters and the opinion of a third author (RB) was sought where there were disagreements in the domain level ratings. For RCTs, we intended to use the original Cochrane tool for assessing risk of bias (RoB1) (Higgins 2011).

We intended, where applicable, to create Summary of findings tables for outcomes and to grade each outcome accordingly using the GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach (Guyatt 2008). However, we found GRADE to not be optimised for these types of data and using it would have led to differential ratings across comparisons, depending on whether there was a difference in MCM rate or not; thus, producing ratings of lower evidence confidence for comparisons with no difference between the ASMs. Further work is required on GRADE and ROBINS‐I to optimise them for pregnancy pharmacovigilance investigations.

Measures of treatment effect

We considered that different study design types or comparator groups may lead to different outcome results and, therefore, we did not combine all data into a single meta‐analysis containing mixed study types, groups of different ASMs and comparator groups. Meta‐analyses were instead stratified by study type, by comparator group (e.g. women with epilepsy untreated and women without epilepsy and with no treatment), and by ASM versus ASM comparison. We computed pooled prevalences of malformations within AED (antiepileptic drug) groups (using fixed‐effect models, unless otherwise stated) and reported them at the beginning of each drug section. The primary and secondary outcomes are presented as risk ratios (RRs). We also computed risk differences (RDs) using Review Manager (RevMan) to take into account studies with no reported events. We calculated these effect estimates in accordance with the Cochrane Handbook for Systematic Reviews of Interventions and reported them in the results section (Higgins 2011). Where treatment effects were reported from individual studies, we used the summary effect measure that had been utilised by the study authors to report results from the study. In some cases, OR instead of RR was reported by individual study authors.

The RR is a measure of relative effect expressed as the ratio of the risk of an event in the two groups. If the 95% confidence interval includes the value of 1.00, this implies there is no difference between the groups (i.e. a non‐significant result). If the value of 1.00 lies outside the 95% confidence interval, this implies there is a difference between the groups (i.e. a significant result). The RD is a measure of absolute effect expressed as the difference of the risk of an event in the two groups. If the 95% confidence interval contains the value of 0.00, this implies there is no difference between the groups (i.e. both groups have the same risk). If the value of 0.00 lies outside the 95% confidence interval, this implies there is a difference between the groups (i.e. a significant result). The significance of the RR and RD may be different, as the RD takes into account comparisons where there were no events in either arm, whilst the other does not. Although the RR estimates are large in many comparisons, the corresponding risk difference estimates are fairly small; but even a small increase in risk for a specific major malformation is clinically meaningful. In these cases, it would be up to the patient/clinician to interpret these risk estimates in the context of the adverse outcome and in relation to the potential benefits of treatment (e.g. seizure control). We did not account for multiple testing and the totality of the evidence for a particular exposure should be considered rather than the outcomes of a single comparison. Finally, we did not carry out any formal analysis of a dose‐response relationship. We have taken any dose‐response results reported directly from the study papers.

Unit of analysis issues

Data published in studies are often duplicated as they are updated, particularly in the case of the prospective pregnancy registries, which update their publications as the numbers of enrolled pregnancies increase. In such cases, we considered the latest time point as the 'primary' study for inclusion. In some cohorts, this meant that we used different publications for different ASMs. Further, there are studies that report combined data from a number of different registers (e.g. EURAP 2018; Samren 1997) which also report independently and routine health record studies with cohort overlap (e.g. UK Clinical Research Practice Database; UK Health Record THIN Register). Where the combined data reported provided greater numbers for a particular ASM comparison, it was included in the meta‐analysis but, where individual initiatives had greater numbers for a specific comparison (e.g. ASM vs control group), we included the individual study data and provided a narrative report of the collaborative initiatives. We carefully examined data to ensure that we did not include them more than once in the analysis and that we did not omit any non‐duplicated data.

Dealing with missing data

We contacted study authors to obtain missing statistics from included studies to input into the meta‐analysis. We also investigated study reasons for missing data to determine if they were missing at random or not.

Assessment of heterogeneity

We assessed clinical heterogeneity by examining the differences in study characteristics in order to inform decisions regarding the combination of study data in meta‐analysis. A priori hypotheses of sources of clinical heterogeneity included: type of population (regional, national or international, single or multicentre), loss to follow‐up, maternal factors including age, duration of ASM treatment, family history of congenital malformation, lifestyle factors, monotherapy, socioeconomic status, type of epilepsy, use of other medications and years of education. Child factors included: age of assessment, sex, seizure exposure, length of follow‐up and outcome measurement.

Where applicable, we also assessed statistical heterogeneity by examining the I² statistic and a Chi² test, using the guidelines outlined in Higgins 2011 for interpreting the results. According to these guidelines, an I² statistic of 0% to 40% may not be important, 30% to 60% may indicate moderate heterogeneity, 50% to 90% may indicate substantial heterogeneity and 75% to 100% may indicate considerable heterogeneity. Therefore, for this review, we considered an I² statistic of more than 50% to indicate significant heterogeneity. The I² statistic was not applicable in comparisons where there was only a single study or when only one study contributed data to the analysis. When interpreting the Chi² test, a P value of less than 0.01 was considered to indicate significant heterogeneity. When we found statistical heterogeneity, we presented both fixed‐effect and random‐effects analyses to enable exploration of differences.

Assessment of reporting biases

We included studies using the Outcome Reporting Bias in Trials (ORBIT) classification system if we suspected selective outcome reporting bias. We requested all protocols from included study authors to enable comparison of outcomes of interest; however, we received very few responses, complicating our performance of this comparison.

Our comprehensive search of multiple sources and data types, together with our requests for unpublished data or clarification from authors, minimised the risk of publication bias.

Data synthesis

We employed both fixed‐effect and random‐effects meta‐analyses to synthesise the data. We presented the primary outcome (major congenital malformations) and the secondary outcome of specific malformations as a risk ratio (RR). Within certain comparisons, we have also presented the risk differences (RD) for both primary outcome (overall malformation rate). In the event that we deemed meta‐analysing inappropriate (e.g. presence of clinical heterogeneity), we applied a narrative form to the review, discussing all comparisons according to the findings presented within the studies.

Comparisons carried out included:

specific ASM monotherapy group versus controls on major congenital malformations;specific ASM monotherapy group versus controls on specific major congential malformation types;specific ASM monotherapy group versus specific ASM monotherapy group on major congential malformations;specific ASM monotherapy group versus specific ASM monotherapy group on specific major congential malformations.

We stratified each comparison by control group, comparator group and study design to ensure appropriate combination of study data. For example, cases reported in a national pregnancy and epilepsy register may also be represented in epidemiological datasets of routine health data which covers the same region or a case in an administrative insurance database may also have been reported to a national epilepsy and pregnancy register and therefore data were not combined across these different data sources.

Subgroup analysis and investigation of heterogeneity

Subgroup analysis was stratified by ASM and type of control or comparator group. When heterogeneity was present across outcomes, we carried out a random‐effects analysis. We examined differences between analyses and reported the appropriate analysis.

Sensitivity analysis

We adopted a cautious approach to combining data extracted from different types of study, and also where different comparator groups were included as outlined in Measures of treatment effect. Additionally, we only included studies where over 70% of the cohort were women taking ASMs for the treatment of epilepsy. This was due to the heterogeneity around doses prescribed, across women taking ASMs for different conditions. This decision is supported by the findings of Hernandez Diaz and colleagues (US Medicaid Registers) who found that differences in the dose of topiramate prescribed for women with epilepsy compared to women prescribed it for other conditions altered the risk of oro‐facial anomalies.

Summary of findings and assessment of the certainty of the evidence

In this review, we considered ASM use in during pregnancy in women with epilepsy and the major malformation rate in their exposed children (Figure 1). Comparisons were made across the different ASM treatments and to unexposed children. The outcomes are summarised in Table 3 along with Table 1, Table 2 for lamotrigine and levetiracetam and in Table 4, Table 5, Table 6, and Table 7 for carbamazepine, oxcarbazepine, topiramate and valproate, respectively. The data for other ASMs were too limited at this time for useful tables to be compiled. Relative risks and risk differences are displayed in Table 8 and Table 9.

1 PRISMA flow diagram*50 studies were included in the original review but, due to changes to the inclusion criteria, 16 studies were excluded.** for some studies only certain data were able to be included in the meta‐analysis. (Figure not reproduced here; see the original article.)

The Robins‐I was adapted for use here to understand the risk of biases but is not yet optimised for pregnancy pharmacovigilance work and, therefore, caution is required in the interpretation of its ratings. It did, however, show that different methodological approaches have different patterns of biases and are therefore in part complimentary (Figure 2). Cohort studies with primary data collection, for example, tend to have lower risks of misclassification of treatment and standardised review of the congenital malformation outcome in the children (leading to low risk of bias ratings), yet they are at higher risk of bias for cohort selection. The use of routine health record data at a national population level does not have these selection risks, however. Stratification of the results by study type provides an internal validation for the results (Figure 3) and the evidence presented in this review should be considered more certain when the results of different comparisons are consistent across study types.

2 Risk of bias for included studies by individual domain (Figure not reproduced here; see the original article.)

3 Prevalence and 95% CI of major congenital malformations for each anti‐seizure medication by data source (Figure not reproduced here; see the original article.)

Malformations are rare outcomes and therefore larger groups are needed to reliably detect a higher risk of malformation in one group over another. Therefore, the certainty of the evidence is greater for medications such as VPA, carbamazepine (CBZ) and lamotrigine (LTG) where the numbers of children are higher within and across the comparisons. The available data were more moderate for levetiracetam (LEV), phenytoin (PHT) and phenobarbital (PB) in certain comparisons. Care should be taken in the interpretation of comparisons where there were fewer than 1000 pregnancies.

Results

Description of studies

Results of the search

In this updated review, electronic searches identified 1067 additional publications; this was in addition to the 11,695 records previously detected in searches for an earlier version of this review (Weston 2016). We found two additional records through handsearching. Following the removal of duplicates, 12,296 abstracts were screened for inclusion in the review across the original and this update. We excluded 12,013 abstracts due to irrelevance, leaving 283 full texts (156 new for this update) to be assessed for eligibility. As the inclusion criteria had been extended to include studies using routine health records in this update, we re‐evaluated search results from the last version for such studies and identified eight additional studies (14 papers). In total, we excluded 155 full‐text papers where they did not meet the inclusion criteria. See Characteristics of excluded studies and Figure 1 for the study flow diagram. We ultimately included 49 studies (128 publications) in this review. Of these, 113 records and 45 studies contributed data to the meta‐analyses, two studies had certain data included in the meta‐analysis whilst other data were narratively reviewed.

Included studies

A total of 128 included full‐text publications reported on the 49 independent studies included in this review, of which all but one were non‐randomised studies. The high number of publications per study were from longitudinal research initiatives such as epilepsy and pregnancy registers which update their results periodically. These full texts were related to an included study, as they presented information on the same cohort of children but either at a different time point or on a related, but not included, outcome (i.e. obstetric or neurodevelopmental outcome). Reported outcomes for each ASM were taken from the most relevant publication within a series; therefore, malformation information for specific ASMs may come from different publications within a series.

Excluded studies

We excluded 42 studies (55 papers) from the review (Excluded studies). Several of these papers were not written in the English language and, therefore, were sent for translation and data extraction in order to determine the study design and methodology used. The most frequent reasons for exclusion, however, were absence of reported ASM monotherapy‐specific malformation outcomes, retrospective study design, and case‐control study design. Studies were also excluded where the maternal indication was not epilepsy in 70% or more of participants, or if a subgroup analysis was not provided for women with epilepsy indication. These decisions were made to limit the likely heterogeneity regarding doses of ASMs used across indications, as dose is a significant driver of higher malformation risk (Brent 2004).

Risk of bias in included studies

Robins‐I ratings are displayed in Figure 2.

Bias in confounding

For bias in confounding, no studies were rated as low as no studies were comparable to a randomised controlled trial. Ten studies were rated as moderate (Australian Epilepsy and Pregnancy Register; EURAP 2018; Kaaja 2003; Meador 2006; Milan Study 1999; MONEAD 2020; Montreal Series; North American Epilepsy and Pregnancy Register; Tanganelli 1992; UK and Ireland Epilepsy and Pregnancy Register) which is the highest rating for non‐randomised studies in this domain. Twenty studies were rated as serious due to a lack of control for key confounders (Cassina 2013; Denmark Health Record Registers; Finland Health Record Registers; Hosny 2021; Italian Lombardy Region Health Register; Kaneko 1999; Kaur 2020; Kelly 1984; Kerala Epilepsy and Pregnancy Registry; Koch 1992; Lindhout 1992; Mawer 2010; Motherisk Registry; Omtzigt 1992; Samren 1997; Steegers‐Theunissen 1994; Sweden Health Record Registers; UK Health Record THIN Register; US Medicaid Registers; Waters 1994), and nine studies were rated as critical (Al Bunyan 1999; AlSheikh 2020; Bag 1989; Barqawi 2005; Delmiš 1991, D'Souza 1991; Eroglu 2008; Fairgrieve 2000; Fröscher 1991; Garza‐Morales 1996; Israeli Teratogen Service; Jimenez 2020; Martinez Ferri 2018; Meischenguiser 2004; Melikova 2020; Miskov 2016; Norwegian Health Record Registers; Pardi 1982; UK Clinical Research Practice Database).

Bias in selection

For bias in selection, three studies were rated as low (Denmark Health Record Registers, Finland Health Record Registers, Norwegian Health Record Registers) as they represented national datasets and one study was rated as moderate (Sweden Health Record Registers). All cohort or pregnancy register studies were at risk of selection biases and therefore 37 studies were rated as serious (AlSheikh 2020; Australian Epilepsy and Pregnancy Register; Barqawi 2005; Cassina 2013; Eroglu 2008; EURAP 2018; Fairgrieve 2000; Fröscher 1991; Garza‐Morales 1996; Hosny 2021; Israeli Teratogen Service; Italian Lombardy Region Health Register; Jimenez 2020; Kaaja 2003; Kaneko 1999; Kaur 2020; Kelly 1984; Kerala Epilepsy and Pregnancy Registry; Koch 1992; Martinez Ferri 2018; Mawer 2010; Meador 2006; Melikova 2020; Milan Study 1999; Miskov 2016; MONEAD 2020; Motherisk Registry; North American Epilepsy and Pregnancy Register; Omtzigt 1992; Samren 1997; Steegers‐Theunissen 1994; Tanganelli 1992; UK and Ireland Epilepsy and Pregnancy Register; UK Clinical Research Practice Database; UK Health Record THIN Register; US Medicaid Registers; Waters 1994) and three studies were rated as critical due to the risk of selection biases (Fröscher 1991; Mawer 2010; Meischenguiser 2004). There was not sufficient information to rate five studies (Al Bunyan 1999; Bag 1989; D'Souza 1991; Delmiš 1991; Montreal Series).

Bias in classification

For bias in classification, 14 studies were rated as low (EURAP 2018; Kaneko 1999; Kerala Epilepsy and Pregnancy Registry; Lindhout 1992; Mawer 2010; Meador 2006; MONEAD 2020; Motherisk Registry; North American Epilepsy and Pregnancy Register; Omtzigt 1992; Samren 1997; Steegers‐Theunissen 1994; UK and Ireland Epilepsy and Pregnancy Register; US Medicaid Registers), 17 studies were rated as moderate (Australian Epilepsy and Pregnancy Register; Bag 1989; D'Souza 1991; Delmiš 1991; Denmark Health Record Registers; Eroglu 2008; Fröscher 1991; Garza‐Morales 1996; Israeli Teratogen Service; Jimenez 2020; Kaaja 2003; Kelly 1984; Koch 1992; Martinez Ferri 2018; Meischenguiser 2004; Milan Study 1999; Pardi 1982), 16 studies were rated as serious (AlSheikh 2020; Barqawi 2005; Cassina 2013; Fairgrieve 2000; Finland Health Record Registers; Hosny 2021; Italian Lombardy Region Health Register; Kaur 2020; Melikova 2020; Montreal Series; Norwegian Health Record Registers; Sweden Health Record Registers; Tanganelli 1992; UK Health Record THIN Register;UK Clinical Research Practice Database; Waters 1994), one study was rated as critical (Al Bunyan 1999) and the other had limited information (Miskov 2016).

Bias in missing data

For bias in missing data, 17 studies were rated as low (Barqawi 2005; D'Souza 1991; Denmark Health Record Registers; Eroglu 2008; EURAP 2018; Finland Health Record Registers; Garza‐Morales 1996; Italian Lombardy Region Health Register; Kaaja 2003; Meador 2006; Meischenguiser 2004; MONEAD 2020; Motherisk Registry; Norwegian Health Record Registers; Omtzigt 1992; Pardi 1982; Sweden Health Record Registers), nine studies were rated as moderate (Australian Epilepsy and Pregnancy Register; Fairgrieve 2000; Jimenez 2020; Kaaja 2003; Kaur 2020; Kelly 1984; Lindhout 1992; Tanganelli 1992; UK Health Record THIN Register; UK Clinical Research Practice Database), ten studies were rated as serious (Cassina 2013; Fröscher 1991; Hosny 2021; Kerala Epilepsy and Pregnancy Registry; Melikova 2020; Milan Study 1999; Montreal Series; UK and Ireland Epilepsy and Pregnancy Register; US Medicaid Registers; Waters 1994), and no studies were rated as critical. There was not sufficient information to rate levels of missing data in 13 studies, however (Al Bunyan 1999, AlSheikh 2020; Bag 1989; Delmiš 1991; Israeli Teratogen Service; Kaneko 1999; Koch 1992; Martinez Ferri 2018; Mawer 2010; Miskov 2016; North American Epilepsy and Pregnancy Register; Samren 1997; Steegers‐Theunissen 1994).

Bias in measurement

For bias in measurement, 11 studies were rated as low (D'Souza 1991; EURAP 2018; Fröscher 1991; Israeli Teratogen Service; Kerala Epilepsy and Pregnancy Registry; Lindhout 1992; MONEAD 2020; Motherisk Registry; North American Epilepsy and Pregnancy Register; Omtzigt 1992; Steegers‐Theunissen 1994) due to undertaking standardised reviews of the outcomes blinded to ASM exposure history. Two studies were rated as moderate (Mawer 2010; Miskov 2016) and 27 studies were rated as serious (AlSheikh 2020; Bag 1989; Cassina 2013; Denmark Health Record Registers; Eroglu 2008; Fairgrieve 2000; Finland Health Record Registers; Garza‐Morales 1996; Hosny 2021; Italian Lombardy Region Health Register; Kaaja 2003; Kaneko 1999; Kaur 2020; Kelly 1984; Koch 1992; Meischenguiser 2004; Melikova 2020; Milan Study 1999; Montreal Series; Norwegian Health Record Registers; Samren 1997; Sweden Health Record Registers; UK and Ireland Epilepsy and Pregnancy Register; UK Health Record THIN Register; US Medicaid Registers; UK Health Record THIN Register; Waters 1994) due to their use of routine clinical data which did not have standardised assessment and were not blinded to ASM exposure history. No studies were rated as critical, but there was insufficient information to rate the likelihood of measurement biases in nine studies (Australian Epilepsy and Pregnancy Register; Barqawi 2005; Delmiš 1991; Fröscher 1991Jimenez 2020; Martinez Ferri 2018; Meador 2006; Pardi 1982; Tanganelli 1992).

Bias in reporting

This domain was difficult to assess as, for most of the studies, no protocol was available (particularly for older studies) or contact with the authors could not be established (Al Bunyan 1999; AlSheikh 2020; Bag 1989; Barqawi 2005; D'Souza 1991; Delmiš 1991; Eroglu 2008; Finland Health Record Registers; Fröscher 1991; Garza‐Morales 1996; Hosny 2021; Italian Lombardy Region Health Register; Jimenez 2020; Kaaja 2003; Kaneko 1999; Kaur 2020; Kelly 1984; Koch 1992; Lindhout 1992; Martinez Ferri 2018; Meischenguiser 2004; Melikova 2020; Milan Study 1999; Miskov 2016; Montreal Series; Motherisk Registry; Norwegian Health Record Registers; Omtzigt 1992; Pardi 1982; Samren 1997; Steegers‐Theunissen 1994; Tanganelli 1992; UK Clinical Research Practice Database; UK Health Record THIN Register; UK Clinical Research Practice Database). Fourteen studies were rated as having low risk for reporting bias, where the protocol could be reviewed in relation to the outcomes and comparisons investigated (Australian Epilepsy and Pregnancy Register; Cassina 2013; Denmark Health Record Registers; EURAP 2018; Fairgrieve 2000; Israeli Teratogen Service; Kerala Epilepsy and Pregnancy Registry; Mawer 2010; Meador 2006; MONEAD 2020; North American Epilepsy and Pregnancy Register; Sweden Health Record Registers; UK and Ireland Epilepsy and Pregnancy Register; US Medicaid Registers).

Effects of interventions

See: Table 1; Table 2

Each included comparison is reviewed below with both the meta‐analysis results being reported alongside any studies which required narrative review only. In comparisons where there were less than 50 children in both groups, the meta‐analysis is not reported, but the data is summarised narratively. Summary tables displaying the pooled prevalences, RR and RDs for each comparison are available in Table 3 along with Table 1 for lamotrigine; Table 2 for levetiracetam, Table 4 for carbamazepine, Table 5 for oxcarbazepine, Table 6 for topiramate, and Table 7 for valproate. A complete summary of all included ASM pooled prevalences, RR and RDs can be found in Table 3, Table 8, and Table 9, respectively with a visual presentation of the major malformation rates displayed in Figure 3.

Women without epilepsy

The prevalence of major malformations (any type) in the cohort studies for children of women without epilepsy (N = 3537), based on data from 12 studies, was 2.1% (95% CI 1.5 to 3.0). The prevalence of major malformations in routine health record studies for children of women without epilepsy (N = 373,028), based on data from three studies, was 3.3% (95% CI 1.5 to 7.1).

Women with epilepsy (no medication)

The prevalence of major malformations (any type) in the cohort studies for children of women with epilepsy (no medication) (N = 1708), based on data from 21 studies, was 3.0% (95% CI 2.1 to 4.2). The prevalence of major malformations in routine health record studies for children of women with epilepsy (no medication) (N = 11,286), based on data from three studies, was 3.2% (95% CI 1.7 to 6.1).

Carbamazepine

The prevalence of major malformations (any type) in the cohort studies for children exposed to carbamazepine (CBZ) (N = 5415), based on data from 37 studies, was 4.7% (95% CI 3.7 to 5.9). The prevalence of major malformations in routine health record studies for children exposed to CBZ (N = 2806), based on data from five studies, was 4.0% (95% CI 2.9 to 5.4).

1 CBZ versus controls

1.1 All major malformations

1.1.1 CBZ versus no medication (in women without epilepsy): cohort studies

Pooled results from 13 cohort studies suggested an increased risk with CBZ (RR 2.30, 95% CI 1.47 to 3.59; I² = 0%), with children exposed to CBZ (N = 1448) experiencing more major malformations than control children (N = 3599) (Analysis 1.1). The RD also suggested a higher absolute risk (RD 0.02, 95% CI 0.01 to 0.03; I² = 0%) (Analysis 1.1).

The multicentre study, Samren 1997, reported 22 (8%) cases of major malformations from 280 infants exposed to CBZ. However, the numbers from centres with a control group were smaller, with four cases of malformation out of just 14 exposed infants. This suggested an increased risk relative to the control children born to women without epilepsy (RR 4.9, 95% CI 1.3 to 18.0).

1.1.2 CBZ versus no medication (in women with epilepsy): cohort studies

Pooled findings from 20 cohort studies suggested an increased risk with CBZ (RR 1.44, 95% CI 1.05 to 1.96; I² = 0%), with children exposed to CBZ (N = 3598) experiencing more major malformations than control children (N = 1691) (Analysis 1.1). The RD also suggested an increased risk with CBZ (RD 0.01, 95% CI 0.00 to 0.02; I² = 1%) (Analysis 1.1).

1.1.3 CBZ versus no medication (in women without epilepsy): routine health record studies

Results from two routine health record studies suggested no evidence of a difference in risk (RR 1.14 95% CI 0.80 to 1.64; I² = 0%), with children exposed to CBZ (N = 983) experiencing a similar major malformation rate to control children (N = 372,111) (Analysis 1.1). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.01 to 0.01; I² = 0%) (Analysis 1.1).

1.1.4 CBZ versus no medication (in women with epilepsy): routine health record studies

Pooled results from four routine health record studies suggested an increased risk with CBZ (RR 1.42 95% CI 1.10 to 1.83; I² = 0%), with children exposed to CBZ (N = 2116) experiencing more major malformations than control children (N = 12,218) (Analysis 1.1). The RD suggested an increased level of risk for CBZ (RD 0.01, 95% CI 0.00 to 0.02; I² = 0%) (Analysis 1.1).

1.2 Neural tube malformations

1.2.1 CBZ versus no medication (in women without epilepsy): cohort studies

Pooled results from seven cohort studies suggested no evidence of a difference in risk (RR 3.09, 95% CI 0.38 to 25.40; I² = 0%), with no difference in the number of neural tube malformations in children exposed to CBZ (N = 269) and compared to control children (N = 1801) (Analysis 1.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.01 to 0.02; I² = 0%) (Analysis 1.2).

1.2.2 CBZ versus no medication (in women with epilepsy): cohort studies

Pooled results from nine cohort studies suggested a comparable level of risk (RR 2.54, 95% CI 0.63 to 10.20; I² = 0%), with no difference in the number of neural tube malformations in children exposed to CBZ (N = 1194) and in control children (N = 679) (Analysis 1.2). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI ‐0.01 to 0.02; I² = 0%) (Analysis 1.2).

1.2.3 CBZ versus no medication (in women without epilepsy): routine health record studies

There were no studies that provided data for this comparison.

1.2.4 CBZ versus no medication (in women with epilepsy): routine health record studies

There were no studies that provided data for this comparison.

1.3 Cardiac malformations

1.3.1 CBZ versus no medication (in women without epilepsy): cohort studies

Pooled results from seven cohort studies suggested no evidence of a difference in risk (RR 1.46, 95% CI 0.43 to 4.99; I² = 0%), with no difference in the number of cardiac malformations in children exposed to CBZ (N = 269) and in control children (N = 1801) (Analysis 1.3). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.02 to 0.01; I² = 0%) (Analysis 1.3).

1.3.2 CBZ versus no medication (in women with epilepsy): cohort studies

Pooled results from 11 cohort studies suggested no evidence of a difference in risk (RR 0.87, 95% CI 0.41 to 1.84; I² = 0%), with no difference in the number of cardiac malformations in children exposed to CBZ (N = 1212) and control children (N = 691) (Analysis 1.3). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.02 to 0.01; I² = 0%) (Analysis 1.3).

1.3.3 CBZ versus no medication (in women without epilepsy): routine health record studies

There were no studies that provided data for this comparison.

1.3.4 CBZ versus no medication (in women with epilepsy): routine health record studies

There were no studies that provided data for this comparison.

1.4 Oro‐facial cleft/craniofacial malformations

1.4.1 CBZ versus no medication (in women without epilepsy): cohort studies

Pooled results from seven cohort studies suggested an increased risk with CBZ (RR 9.04, 95% CI 2.16 to 37.87; I² = 10%), with children exposed to CBZ (N = 269) experiencing more oro‐facial cleft/craniofacial malformations than control children (N = 1801) (Analysis 1.4). The RD suggested no difference in the level of risk (RD 0.01, 95% CI ‐0.01 to 0.03; I² = 0%) (Analysis 1.4).

1.4.2 CBZ versus no medication (in women with epilepsy): cohort studies

Pooled results from nine cohort studies suggested no evidence of a difference in risk (RR 0.99, 95% CI 0.27 to 3.62; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to CBZ (N = 709) and control children (N = 347) (Analysis 1.4). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.02 to 0.02; I² = 0%) (Analysis 1.4).

1.4.3 CBZ versus no medication (in women without epilepsy): routine health record studies

There were no studies that provided data for this comparison.

1.4.4 CBZ versus no medication (in women with epilepsy): routine health record studies

There were no studies that provided data for this comparison.

1.5 Skeletal/limb malformations

1.5.1 CBZ versus no medication (in women with epilepsy): cohort studies

Pooled results from seven cohort studies suggested no evidence of a difference in risk (RR 5.13, 95% CI 0.52 to 50.67, I² = 0%), with no difference in skeletal/limb malformations in children exposed to CBZ (N = 269) and control children (N = 1801) (Analysis 1.5). The RD also suggested a comparable level of risk (RD 0.00, 95% CI ‐0.01 to 0.02; I² = 0%) (Analysis 1.5).

1.5.2 CBZ versus no medication (in women with epilepsy): cohort studies

Pooled results from nine cohort studies suggested no evidence of a difference in risk (RR 0.96, 95% CI 0.35 to 2.82; I² = 0%), with no difference in the number of skeletal and limb malformations in children exposed to CBZ (N = 1194) and control children (N = 679) (Analysis 1.5). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.01 to 0.01; I² = 0%) (Analysis 1.5).

1.5.3 CBZ versus no medication (in women without epilepsy): routine health record studies

There were no studies that provided data for this comparison.

1.5.4 CBZ versus no medication (in women with epilepsy): routine health record studies

There were no studies that provided data for this comparison.

Carbamazepine dose

The EURAP 2018 collaboration has reported on the largest uniformly assessed group of children exposed to CBZ (N = 1957). They reported a higher malformation rate with higher doses of CBZ. Doses =/< 700 mg/d were found to have a malformation risk of 4.5% (95% CI 3.5% to 5.8%), whilst dose > 700 mg/d were associated with a prevalence of 7.2%, (95% CI 5.4 to 9.4); a difference which suggested a dose association (OR 1.56, 95% CI 1.03 to 2.37, P = 0.0352). When compared to children exposed to =/< 325 mg/d of LTG, the prevalence was higher for doses =/< 700 mg/d (OR 1.71 95% CI 1.12 to 2.61, P = 0.0143), and doses over 700 mg/d were also higher (OR 2.68, 95% CI 1.71 to 4.19, P = 0.0002). In contrast, however, the North American Epilepsy and Pregnancy Register (N = 1033) failed to document an association between the risk of major malformation and the dose of CBZ; however, this group was smaller. The Australian Epilepsy and Pregnancy Register, the UK and Ireland Epilepsy and Pregnancy Register, and a number of smaller studies also did not identify a dose effect (Kaaja 2003; Kaneko 1999; Milan Study 1999; Motherisk Registry; Samren 1997).

Data regarding the impact of dose are limited from routine healthcare record‐based studies. Data analyses from Finland Health Record Registers did not establish a dose relationship, however, the number of carbamazepine monotherapy cases was small (N = 32). Results from the Norwegian Health Record Registers and Sweden Health Record Registers did not capture ASM doses, and researchers using the UK Health Record THIN Register or the UK Clinical Research Practice Database were not able to access dose information. Dose data have not currently been provided by the Denmark Health Record Registers for CBZ dose.

Clonazepam

2 CZP versus controls

2.1 All major malformations

The prevalence of major malformations (any type) in cohort studies for children exposed to clonazepam (CZP) (N = 95), based on data from four studies, was 2.1% (95% CI 0.2 to 17.3). The prevalence of major malformations in routine health record studies for children exposed to CZP (N = 161), based on data from one study, was 2.5% (95% CI 0.0 to 131.8).

2.1.1 CZP versus no medication (in women without epilepsy): cohort studies

Pooled results from two cohort studies suggested no evidence of a difference in risk (RR 2.76, 95% CI 0.55 to 13.94; I² = 0%), with children exposed to CZP (N = 65) experiencing comparable rates of major malformations to control children (N = 504) (Analysis 2.1). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI ‐0.03 to 0.07; I² = 0%) (Analysis 2.1).

2.1.2 CZP versus no medication (in women with epilepsy): cohort studies

Pooled findings from three cohort studies suggested no evidence of a difference in risk (RR 1.08, 95% CI 0.21 to 5.42; I² = 0%), with children exposed to CZP (N = 31) experiencing comparable rates of major malformations to control children (N = 524) (Analysis 2.1). The RD also suggested no difference in the level of risk (RD ‐0.03, 95% CI ‐0.11 to 0.04; I² = 0%) (Analysis 2.1).

2.1.3 CZP versus no medication (in women without epilepsy): routine health record studies

One study suggested no evidence of a difference in risk (RR 0.70, 95% CI 0.18 to 2.77; I² = NA (not available)) with children exposed to CZP (N = 113) experiencing comparable rates of major malformations to control children (N = 369,267). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI ‐0.03 to 0.02; I² = NA) (Analysis 2.1).

2.1.4 CZP versus no medication (in women with epilepsy): routine health record studies

One study suggested no evidence of a difference in risk (RR 0.69, 95% CI 0.17 to 2.79; I² = NA) with children exposed to CZP (N = 113) experiencing comparable rates of major malformations to control children (N = 1900). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI ‐0.03 to 0.02; I² = NA) (Analysis 2.1).

Specific malformation types were not reviewed due to the small amount of data.

CZP Dose

There is too little experience with CZP in pregnancy to be able to report on the potential of an association between the dose of CZP and MCM risk.

Gabapentin

The prevalence of major malformations (any type) in cohort studies for children exposed to gabapentin (GBP) (N = 192) based on data from four studies was 2.0% (95% CI 0.1 to 32.2). The prevalence of major malformations in routine health record studies for children exposed to GBP (N = 18), was based on data from one study and therefore could not be calculated.

3 GBP versus controls

3.1 All major malformations

3.1.1 GBP versus no medication (in women without epilepsy): cohort studies

Pooled results from two cohort studies suggested no evidence of a difference in risk of major malformations for the children exposed to gabapentin (N = 147) in comparison to children born to women without epilepsy (N = 570) (RR 1.78, 95% CI 0.50 to ‐6.29, P = 0.37, I² = 89%), but there was heterogeneity in the results (Analysis 3.1). A random‐effects RR was calculated which also suggested a comparable level of risk (RR 8.04, 95% CI 0.03 to 1898.73, P = 0.45, I² = 89%). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.02 to 0.03; I² = 75%). Due to heterogeneity, a random‐effects RD was calculated which also found a comparable level of risk (RD 0.19, 95% CI ‐0.37 to 0.74, P = 0.51, I² = 75%) (Analysis 3.1).

3.1.2 GBP versus no medication (in women with epilepsy): cohort studies

Pooled results from two cohort studies suggested no evidence of a difference in risk of major malformation for the children exposed to gabapentin (n=47) in comparison to control children (n= 721) (RR 1.77, 95% CI 0.46 to 6.90, P = 0.41, I² = 0% (Analysis 3.1).

3.1.3 GBP versus no medication (in women without epilepsy): routine health record studies

There were no studies that provided data for this comparison in a format that could be combined in a meta‐analysis. However, Patorno and colleagues (US Medicaid Registers) conducted a sensitivity analysis that was restricted to epilepsy indications and included 347 pregnancies exposed to gabapentin in comparison to an unexposed reference group of 11,861 pregnancies. There was no reported difference in the malformation outcome either in the epilepsy subgroup (RR 1.40, 95% CI 0.73 to 2.71, P = 0.31) or in the main analysis which included 3745 gabapentin‐exposed children (RR 1.07, 95% CI 0.94 to 1.21, P = 0.33).

3.1.4 GBP versus no medication (in women with epilepsy): routine health record studies

There were no studies that provided data for this comparison.

3.2 Neural tube malformations

3.2.1 GBP versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR for the included study due to there being no reported neural tube malformations in children exposed to GBP (N = 2) or control children (N = 128) (Analysis 3.2).

3.2.2 GBP versus no medication (in women with epilepsy): cohort studies

There were no studies that provided data for this comparison.

3.2.3 GBP versus no medication (in women without epilepsy): routine health record data studies

There were no studies that provided data for this comparison.

3.2.4 GBP versus no medication (in women with epilepsy): routine health record data studies

There were no studies that provided data for this comparison.

3.3 Cardiac malformations

3.3.1 GBP versus no medication (in women without epilepsy): cohort studies

Data from one study suggested a difference in risk (RR 129.00, 95% CI 6.49 to 2562.48, I² = NA) with children exposed to GBP (N = 2) being at higher risk than control children (N = 128) (Analysis 3.3). However, the RD suggested no difference in the level of risk (RD 0.50, 95% CI ‐0.07 to 1.07; I² = NA)

3.3.2 GBP versus no medication (in women with epilepsy): cohort studies

Included studies did not reach the threshold for reporting in the meta‐analysis (Analysis 3.3). However, available data showed that there was one case of cardiac malformation in children exposed to GBP (N = 2) in comparison to zero cases in the control children (N = 4), based on data from one study (Miskov 2016).

3.3.3 GBP versus no medication (in women without epilepsy): routine health record data studies

Patorno and colleagues, using data including the US Medicaid Registers, found a comparable level of risk for cardiac anomalies in children exposed to gabapentin (N = 347) versus children born to women without epilepsy (N = 11,861) (RR 1.40, 95% CI 0.73 to 2.71, P = 0.31).

3.3.4 GBP versus no medication (in women with epilepsy): routine health record data studies

There were no studies that provided data for this comparison.

3.4 Oro‐Facial Cleft/Craniofacial malformations

3.4.1 GBP versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR from one study due to there being no oro‐facial cleft / craniofacial malformations in children exposed to GBP (n=2) in comparison to no cases in 128 control children (Analysis 3.4).

3.4.2 GBP versus no medication (in women with epilepsy): cohort studies

There were no studies that provided data for this comparison.

3.4.3 GBP versus no medication (in women without epilepsy): routine health record data studies

There were no studies that provided data for this comparison.

3.4.4 GBP versus no medication (in women with epilepsy): routine health record data studies

There were no studies that provided data for this comparison.

3.5 Skeletal/Limb malformations

3.5.1 GBP versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR from one study due to there being no reported skeletal/limb malformations in children exposed to GBP (n=2) or 128 control children, based on data from one study (Analysis 3.5).

3.5.2 GBP versus no medication (in women with epilepsy): cohort studies

There were no studies that provided data for this comparison.

3.5.3 GBP versus no medication (in women without epilepsy): routine health record data studies

There were no studies that provided data for this comparison.

3.5.4 GBP versus no medication (in women with epilepsy): routine health record data studies

There were no studies that provided data for this comparison.

Gabapentin dose

The investigation of GBP dose and its potential association with an increased rate of malformations is limited due to the relatively small number of pregnancies where data are currently available. The US Medicaid Registers is the most reliable data source currently available. The study authors did not find that malformation risk increased with dose according to tertiles of the first and the highest prescribed daily dose filled. Doses of 600 mg/d through to 900 mg/d (RR 1.00, 95% CI 0.80 to 1.24, P = 0.98) or doses above 900 mg/d (RR 1.17, 95% CI 0.95 to 1.44, P = 0.15) were not associated with a risk above the baseline risk. The largest cohort study of GBP‐exposed pregnancies was from the North American Epilepsy and Pregnancy Register (N = 145) and no association between increasing dose and increased malformation risk was identified in this study. The participant numbers in other included studies of GBP were too small to investigate any effect of dose size and MCM risk.

Levetiracetam

The prevalence of major malformations (any type) in cohort studies for children exposed to levetiracetam (LEV) (N = 1242), based on data from 11 studies, was 2.6% (95% CI 1.6 to 4.4). The prevalence of major malformations in routine health record studies for children exposed to LEV (N = 248), based on data from two studies, was 2.8% (95% CI 0.0 to 321.9).

4 LEV versus controls

4.1 All major malformations

4.1.1 LEV versus no medication (in women without epilepsy): cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 2.20, 95% CI 0.98 to 4.93; I² = 0%), with children exposed to LEV (N = 574) experiencing comparable rates of major malformations to control children (N = 1022) (Analysis 4.1). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.00 to 0.03; I² = 0%).

4.1.2 LEV versus no medication (in women with epilepsy): cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.71, 95% CI 0.39 to 1.28; I² = 0%), with children exposed to LEV (N = 724) experiencing comparable rates of major malformations to control children (N = 1101) (Analysis 4.1). The RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.03 to 0.00; I² = 0%).

4.1.3 LEV versus no medication (in women without epilepsy): routine health record data studies

One study suggested no evidence of a difference in risk (RR 0.67, 95% CI 0.17 to 2.66; I² = NA) for children exposed to LEV (N = 118) experiencing comparable rates of major malformations to control children (N = 369,267). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI ‐0.03 to 0.02; I² = 0%).

4.1.4 LEV versus no medication (in women with epilepsy): routine health record data studies

Pooled results from two routine health record studies suggested no evidence of a difference in risk (RR 0.82, 95% CI 0.39 to 1.71; I² = 0%), with children exposed to LEV (N = 248) experiencing comparable rates of major malformations to control children (N = 10,377) (Analysis 4.1). The RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.03 to 0.01; I² = 0%).

4.2 Neural tube malformations

4.2.1 LEV versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR from two studies due to there being no reported cases of neural tube malformation in children exposed to LEV (N = 105) or control children (N = 383) (Analysis 4.2).

4.2.2 LEV versus no medication (in women with epilepsy): cohort studies

We were unable to estimate a RR from two studies due to there being no reported cases of neural tube malformations in children exposed to LEV (N = 173) or control children (N = 361) (Analysis 4.2).

4.2.3 LEV versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

4.2.4 LEV versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

4.3 Cardiac malformations

4.3.1 LEV versus no medication (in women without epilepsy): cohort studies

Pooled results from two cohort studies suggested no evidence of a difference in risk (RR 3.92, 95% CI 0.57 to 27.07; I² = 0), with children exposed to LEV (N = 105) experiencing comparable rates of major malformations to control children (N = 383) (Analysis 4.3). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.02 to 0.06; I² = 0%).

4.3.2 LEV versus no medication (in women with epilepsy): cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 0.90, 95% CI 0.31 to 2.60; I² = 0), with children exposed to LEV (N = 281) experiencing comparable rates of major malformations to control children (N = 384) (Analysis 4.3). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.03 to 0.03; I² = 0%).

4.3.3 LEV versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

4.3.4 LEV versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

4.4 Oro‐facial cleft/craniofacial malformations

4.4.1 LEV versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR from two studies due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to LEV (N = 105) or control children (N = 383) (Analysis 4.4).

4.4.2 LEV versus no medication (in women with epilepsy): cohort studies

Pooled results from three studies suggested no evidence of a difference in risk (RR 0.14, 95% CI 0.01 to 3.18; I² = N/A), with children exposed to LEV (N=186) experiencing comparable rates of oro‐facial cleft/craniofacial malformations as control children (N=44) (Analysis 4.4).

4.4.3 LEV versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

4.4.4 LEV versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

4.5 Skeletal/limb malformations

4.5.1 LEV versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR from two studies due to there being no skeletal / limb malformations in children exposed to LEV (N = 105) or control children (N = 383) (Analysis 4.5).

4.5.2 LEV versus no medication (in women with epilepsy): cohort studies

Pooled results from three studies suggested no evidence of a difference in risk (RR 3.21, 95% CI 0.46 to 22.50; I² = NA), with children exposed to LEV (N = 272) experiencing comparable rates of skeletal/limb malformations to control children (N = 376) (Analysis 4.5). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.02 to 0.03; I² = 0%).

4.5.3 LEV versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

4.5.4 LEV versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data for this outcome.

Levetiracetam dose

EURAP 2018 had the largest cohort of LEV‐exposed children to conduct dose investigations in 599 exposed children. Whilst they did not make comparisons between different levels of LEV dose directly, they did report that there was evidence of lower risk of any LEV dose (250‐4000 mg/d) in comparison to doses of VPA </= 650 mg/d and dose of CBZ > 700 mg/d, whilst there was no evidence of difference in comparison to doses of LTG either at </= 325 mg/d or > 325 mg/d, or in comparison to OXC at doses ranging from 75‐4500 mg/d. Additionally, the North American Epilepsy and Pregnancy Register reporting on LEV‐exposed children (N = 450), the UK and Ireland Epilepsy and Pregnancy Register (N = 304), the Australian Epilepsy and Pregnancy Register (N = 139), the Kerala Epilepsy and Pregnancy Registry (N = 106) and the MONEAD 2020 study (N = 99) also failed to find an association between increasing doses of LEV and congenital anomaly risk; however, group sizes may still be too limited at higher dose levels to detect increased levels of MCM risk.

Lamotrigine

The prevalence of major malformations (any type) in cohort studies for children exposed to lamotrigine (LTG) (N = 4704), based on data from 15 studies, was 2.7% (95% CI 1.9 to 3.8). The prevalence of major malformations in routine health record studies for children exposed to LTG (N = 2502), based on data from four studies, was 3.5% (95% CI 2.5 to 4.9).

5 LTG versus controls

5.1 All major malformations

5.1.1 LTG versus no medication (in women without epilepsy): cohort studies

Pooled results from seven studies suggested an increased risk with LTG (RR 1.97, 95% CI 1.16 to 3.39; I² = 0%), with children exposed to LTG (N = 1899) experiencing more major malformations to control children (N = 2693) (Analysis 5.1). The RD also suggested a higher risk (RD 0.01, 95% CI 0.00 to 0.03; I² = 0%).

5.1.2 LTG versus no medication (in women with epilepsy): cohort studies

Pooled results from eight studies suggested no evidence of a difference in risk (RR 1.04, 95% CI 0.66 to 1.63; I² = 0%), with children exposed to LTG (N = 2767) experiencing comparable rates of major malformations to control children (N = 1151) (Analysis 5.1). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.01 to 0.01; I² = 0%).

5.1.3 LTG versus no medication (in women without epilepsy): routine health record data studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 1.19, 95% CI 0.86 to 1.64; I² = 18%), with children exposed to LTG (N = 1177) experiencing comparable rates of major malformations to control children (N = 372,111) (Analysis 5.1). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.01 to 0.02; I² = 22%).

5.1.4 LTG versus no medication (in women with epilepsy): routine health record data studies

Pooled results from three studies suggested no evidence of a difference in risk (RR 1.00, 95% CI 0.79 to 1.28; I² = 0%), with children exposed to LTG (N = 2166) experiencing comparable rates of major malformations to control children (N = 11,279) (Analysis 5.1). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.01 to 0.01; I² = 0%).

5.2 Neural tube malformations

5.2.1 LTG versus no medication (in women without epilepsy)

Pooled results from five studies suggested an increased risk with LTG (RR 7.55, 95% CI 1.05 to 54.09; I² = 0%), with children exposed to LTG (N = 313) experiencing more major malformations to control children (N = 1654) (Analysis 5.2). However, the RD suggested no difference in the level of risk (RD 0.00, 95% CI −0.01 to 0.02; I² = 0%).

5.2.2 LTG versus no medication (in women with epilepsy)

We were unable to estimate a RR from five studies, as there were no reported neural tube malformations in children exposed to LTG (N = 521) or control children (N = 563) (Analysis 5.2).

5.2.3 LTG versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

5.2.4 LTG versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

5.3 Cardiac malformations

5.3.1 LTG versus no medication (in women without epilepsy): cohort studies

Pooled results from six studies suggested an increased risk with LTG (RR 2.71, 95% CI 1.05 to 6.98; I² = 0%), with children exposed to LTG (N = 348) experiencing more major malformations to control children (N = 1658) (Analysis 5.3). However, the RD suggested no difference in the level of risk (RD 0.01, 95% CI −0.01 to 0.03; I² = 0%).

5.3.2 LTG versus no medication (in women with epilepsy): cohort studies

Pooled results from six studies suggested no evidence of a difference in risk (RR 0.97, 95% CI 0.28 to 3.32; I² = 0%), with children exposed to LTG (N = 541) experiencing comparable rates of major malformations to control children (N = 571) (Analysis 5.3). However, the RD suggested no difference in the level of risk (RD 0.00, 95% CI −0.02 to 0.02; I² = 0%).

5.3.3 LTG versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data for this outcome.

5.3.4 LTG versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data for this outcome.

5.4 Oro‐facial cleft/craniofacial malformations

5.4.1 LTG versus no medication (in women without epilepsy): cohort studies

We were unable to estimate RR from the four included studies due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to LTG (N = 197) or control children (N = 826) (Analysis 5.4).

5.4.2 LTG versus no medication (in women with epilepsy): cohort studies

Pooled results from five studies suggested no evidence of a difference in risk (RR 1.37, 95% CI 0.29 to 6.56; I² = 65%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to LTG (N = 491) and control children (N = 322) (Analysis 5.4). Due to high heterogeneity, a random‐effects RR was calculated which also found no difference (RR 0.90, 95% CI 0.03 to 32.04, P = 0.95, I² = 65%). The RD suggested no difference in the level of risk (RD 0.00, 95% CI −0.01 to 0.02; I² = 0%).

5.4.3 LTG versus no medication (in women without epilepsy): routine health record data studies

In the study using the US Medicaid Registers by Hernandez‐Diaz and colleagues, there was no evidence of a difference in the oral cleft rates for children exposed to LTG (N = 2796) in comparison to the children born to women without epilepsy (N = 1,322,955) (RR 1.89, 95% CI 0.85 to 4.21).

5.4.4 LTG versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data for this outcome.

5.5 Skeletal/limb malformations

5.5.1 LTG versus no medication (in women without epilepsy): cohort studies

Pooled results from five studies suggested an increased risk with LTG (RR 11.29, 95% CI 2.37 to 53.91; I² = 0%), with children exposed to LTG (N = 311) experiencing more major malformations to control children (N = 1654) (Analysis 5.5). However, the RD suggested no difference in the level of risk (RD 0.01, 95% CI −0.00 to 0.03; I² = 0%).

5.5.2 LTG versus no medication (in women with epilepsy): cohort studies

Pooled results from five studies suggested no evidence of a difference in risk (RR 0.75, 95% CI 0.20 to 2.89; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to LTG (N = 521) and control children (N = 563) (Analysis 5.5). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.02 to 0.01; I² = 0%).

5.5.3 LTG versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data for this outcome.

5.5.4 LTG versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data for this outcome.

Lamotrigine dose

The EURAP 2018 collaboration has reported on a large, uniformly assessed, group of children exposed to LTG (N = 2514). It reported a higher MCM rate with higher doses of LTG. Doses =/< 325 mg/d were found to have an MCM risk of 2.5% (95% CI 1.8% to 3.3%), whilst doses > 325 mg/d were associated with MCM in 4.3% of children (95% CI 2.9% to 6.2%); a difference which suggested a dose association (OR 1.68, 95% CI 1.01 to 2.80, P = 0.0463).

When EURAP 2018 compared lower dose LTG (=/< 325 mg/d) to other monotherapy ASMs, they found evidence suggesting a lower MCM risk in comparison to CBZ at =/< 700 mg/d (OR 1.71 95% CI 1.12 to 2.61, P = 0.0143) and lower risk than CBZ doses > 700 mg/d (OR 2.68, 95% CI 1.71 to 4.19, P = 0.0002). In comparison to LEV, there was no evidence of a difference between lower doses of LTG (</= 325 mg/d) and LEV doses between =/> 250‐4000 mg/d (OR 1.11, 95% CI 0.62 to 2.00, P = 0.7282). Comparisons to VPA demonstrated lower MCM risks for lower LTG dose (=/< 325 mg/d) in comparison to VPA doses at =/< 650 mg/d (OR 2.70, 95% CI 1.67 to 4.38, P = 0.0002), > 650 mg/d to =/< 1450 mg/d (OR 4.72, 95% CI 3.11 to 7.18, P < 0.0002), or at doses of VPA > 1450 mg/d (OR 13.52, 95% CI 7.73 to 23.64, P = 0.0002). Exposure to LTG at a dose =/< 325 mg daily was associated with a lower MCM risk than PB exposure at doses of between > 80 and =/< 130 mg/d (OR 2.46, 95% CI 1.16 to 5.23, P = 0.0196) and at PB doses > 130 mg/d (OR 5.81, 95% CI 2.40 to 14.08, P = 0.0002). There was, however, no evidence of a difference in comparison of LTG doses =/< 325 mg/d to the lowest investigated PB dose of =/< 80 mg/d (OR 1.07, 95% CI 0.25 to 4.60, P = 0.923). Rates of PHT, TPM, and OXC‐exposed pregnancies were lower in the EURAP study which should be considered with regard to findings suggesting that there is no dose association here. In comparison to lower dose LTG (=/< 325 mg/d), there was no evidence of difference for PHT doses between =/30 mg/d and 730 mg/d (OR 1.93, 95% CI 0.78 to 4.75, P = 0.1554) or TPM doses =/> 25 mg/d to 500 mg/d (OR 1.67, 95% CI 0.69 to 4.04, P = 0.2524) or OXC doses between =/> 75 to 4500 mg/d (OR 1.13, 95% CI 0.55 to 2.31, P = 0.7358).

The EURAP 2018 collaboration also compared higher doses of LTG (> 325 mg/d) and found a comparable level of risk to higher doses of CBZ (> 700 mg/d, OR 0.63, 95% CI 0.38 to 1.05, P = 0.0766), to LEV doses between =/> 250‐4000 mg/d (OR 1.51, 95% CI 0.79 to 2.88, P = 0.2077) and to OXC doses between 75‐4500 mg/d (OR 1.49, 95% CI 0.70 to 3.17, P = 0.3051). Higher doses of LTG (> 325 mg/d) were not associated with lower rates of MCM compared to the lowest investigated dose range for VPA (=/< 650 mg/d, OR 0.62, 95% CI 0.36 to 1.09, P = 0.0959) but there was evidence suggesting higher doses of LTG were associated with a lower MCM risk than VPA doses between > 650 to =/< 1450 mg/d (OR 2.81, 95% 1.70 to 4.65, P = 0.0002).

In contrast to the data from EURAP, the UK and Ireland Epilepsy and Pregnancy Register (N = 2198) found no evidence of risk with increasing doses of LTG (0 to 200 mg/d vs 200 to 400 mg/d; 0 to 200 mg/d vs > 400 mg/d). The North American Epilepsy and Pregnancy Register (N = 1562), the Australian Epilepsy and Pregnancy Register (N = 406), and the Israeli Teratogen Service (N = 114) studies did not identify dose‐related risks associated with LTG. The frequency of MCM was too low in other included studies to allow reliable investigation of dose.

Oxcarbazepine

The prevalence of major malformations (any type) in cohort studies for children exposed to oxcarbazepine (OXC) (N = 378), based on data from 11 studies, was 2.8% (95% CI 1.1 to 6.6). The prevalence of major malformations in routine health record studies for children exposed to OXC (N = 507), based on data from four studies, was 4.8% (95% CI 0.7 to 31.5).

6 OXC versus controls

6.1 All major malformations

6.1.1 OXC versus no medication (in women without epilepsy): cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 2.20, 95% CI 0.67 to 7.27; I² = 18%), with children exposed to OXC (N = 184) experiencing comparable rates of major malformations to control children (N = 767) (Analysis 6.1). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.02 to 0.04; I² = 0%).

6.1.2 OXC versus no medication (in women with epilepsy): cohort studies

Pooled results from six studies suggested no evidence of a difference in risk (RR 1.40, 95% CI 0.68 to 2.91; I² = 23%), with children exposed to OXC (N = 134) experiencing comparable rates of major malformations to control children (N = 788) (Analysis 6.1). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.03 to 0.07; I² = 0%).

6.1.3 OXC versus no medication (in women without epilepsy): routine health record data studies

Results from one study found no evidence of a difference in risk (RR 0.70, 95% CI 0.10 to 4.86; I² = N/A), with children exposed to OXC (N = 57) experiencing comparable rates of major malformations to control children (N = 369,267) (Analysis 6.1). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.04 to 0.03; I² = N/A).

6.1.4 OXC versus no medication (in women with epilepsy): routine health record data studies

Pooled results from three studies suggested an increased risk with OXC (RR 1.75, 95% CI 1.22 to 2.52; I² = 94%), with children exposed to OXC (N = 503) experiencing higher rates of major malformations than control children (N = 11,316) (Analysis 6.1). Due to high heterogeneity, a random‐effects RR was calculated and found no evidence of a difference in risk (RR 1.61, 95% CI 0.26 to 9.86; I² = 94%). The RD suggested a higher risk for OXC (RD 0.03, 95% CI 0.01 to 0.05; I² = 94%); however, a random‐effects RD due to heterogeneity found no difference in the level of risk (RD 0.04, 95% CI ‐0.05 to 0.12; I² = 94%).

6.2 Neural tube malformations

6.2.1 OXC versus no medication (in women without epilepsy): cohort studies

No included studies reported data for this outcome.

6.2.2 OXC versus no medication (in women with epilepsy): cohort studies

We were unable to estimate a RR from the two included studies due to there being no reported neural tube malformations in children exposed to OXC (N = 102) or control children (N = 361) (Analysis 6.2).

6.2.3 OXC versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data for this outcome.

6.2.4 OXC versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

6.3 Cardiac malformations

6.3.1 OXC versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR in the included study due to there being no reported cardiac malformations in children exposed to OXC (N = 1) or control children (N = 128) (Analysis 6.3).

6.3.2 OXC versus no medication (in women with epilepsy): cohort studies

Pooled results from four studies suggested no evidence of a difference in risk (RR 1.10, 95% CI 0.36 to 3.35; I² = 22%), with children exposed to OXC (N = 106) experiencing comparable rates of major malformations to control children (N = 373) (Analysis 6.3). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.04 to 0.05; I² = 0%).

6.3.3 OXC versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

6.3.4 OXC versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

6.4 Oro‐facial cleft/craniofacial malformations

6.4.1 OXC versus no medication (in women without epilepsy): cohort studies

No included studies reported data on this outcome.

6.4.2 OXC versus no medication (in women with epilepsy): cohort studies

Included studies did not reach the threshold for reporting the meta‐analysis (Analysis 6.4). However, available data showed there were 0/34 cases of oro‐facial cleft/craniofacial malformations in children exposed to OXC and 1/29 cases in control children, based on data from two studies (AlSheikh 2020; Hosny 2021).

6.4.3 OXC versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

6.4.4 OXC versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

6.5 Skeletal/limb malformations

6.5.1 OXC versus no medication (in women without epilepsy): cohort studies

No included studies reported data on this outcome.

6.5.2 OXC versus no medication (in women with epilepsy): cohort studies

Pooled data from two studies suggested no evidence of a difference in risk (RR 2.39, 95% CI 0.22 to 26.05; I² = NA), with children exposed to OXC (N = 102) experiencing comparable rates of major malformations to control children (N = 361) (Analysis 6.5). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.02 to 0.03; I² = 0%).

6.5.3 OXC versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

6.5.4 OXC versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

Oxcarbazepine dose

The limited published experience of OXC in pregnancy limits dose comparisons, even in the EURAP 2018 study for different doses of OXC (N = 333). In EURAP 2018, there was no evidence that doses of OXC between =/> 75 to 4500 mg/d were different from those for lower dose LTG (=/< 325 mg/d) (OR 1.13, 95% CI 0.55 to 2.31, P = 0.7282) or dose of LTG > 325 mg/d (OR 1.49, 95% CI 0.70 to 3.17, P = 0.3051). Similarly, a lack of difference was also found in comparison to any dose of LEV exposure (OR 1.02, 95% CI 0.45 to 2.30, P = 0.9644). A lower prevalence of MCM was found for any dose of OXC (3.0%, 95% CI 1.4 to 5.4) in comparison to low dose VPA (=/< 650 mg/d, 6.3%, 95% CI 4.5 to 8.6) (OR 2.39, 95% CI 1.13 to 5.08, P = 0.0235), but was not reported for any higher‐dose VPA.

Other studies were limited to the number of OXC‐exposed pregnancies or had not published dose data.

Phenobarbital

The prevalence of major malformations (any type) in cohort studies for children exposed to phenobarbital (PB) (N = 840), based on data from 26 studies, was 6.3% (95% CI 4.8 to 8.3). The prevalence of major malformations in routine health record studies for children exposed to PB (N = 34), based on data from two studies, was 8.8% (95% CI 0.0 to 9722.4).

7 PB versus controls

7.1 All major malformations

7.1.1 PB versus no medication (in women without epilepsy): cohort studies

Pooled results from eight studies suggested an increased risk with PB (RR 3.22, 95% CI 1.84 to 5.65; I² = 0%), with children exposed to PB (N = 353) experiencing more major malformations than control children (N = 2042) (Analysis 7.1). The RD also suggested a higher risk for PB (RD 0.04, 95% CI 0.01 to 0.07; I² = 0%).

Samren 1997 reported five cases of major malformation out of 48 exposed infants (10%). Numbers were more limited in the comparison to control children (as not all centres in the study included control children), with just one malformation case out of six PB‐exposed children; analysis suggested no evidence of a difference between the groups (RR 2.4, 95% CI 0.3 to 23.0).

7.1.2 PB versus no medication (in women with epilepsy): cohort studies

Pooled results from 13 studies suggested no evidence of a difference in risk (RR 1.64, 95% CI 0.94 to 2.83; I² = 0%), with no difference in the number of major malformations in children exposed to PB (N = 438) and control children (N = 999) (Analysis 7.1). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.01 to 0.06; I² = 0%).

7.1.3 PB versus no medication (in women without epilepsy): routine health record data studies

The results from one study suggested no evidence of a difference in risk (RR 2.94, 95% CI 0.77 to 11.15; I² = NA), with children exposed to PB (N = 27) experiencing comparable rates of major malformations to control children (N = 369,267) (Analysis 7.1). The RD also suggested no difference in the level of risk (RD 0.05, 95% CI −0.05 to 0.15; I² = NA).

7.1.4 PB versus no medication (in women with epilepsy): routine health record data studies

The results from one study suggested no evidence of a difference in risk (RR 2.87, 95% CI 0.74 to 11.21; I² = NA), with children exposed to PB (N = 27) experiencing comparable rates of major malformations to control children (N = 1900) (Analysis 7.1). The RD also suggested no difference in the level of risk (RD 0.05, 95% CI −0.05 to 0.15; I² = NA).

7.2 Neural tube malformations

7.2.1 PB versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR from two studies due to there being no neural tube malformations in children exposed to PB (N = 7) or control children (N = 244) (Analysis 7.2).

7.2.2 PB versus no medication (in women with epilepsy): cohort studies

Pooled results from three studies suggested no evidence of a difference in risk (RR 3.85, 95% CI 0.47 to 31.26, I² = 0%), with no difference in the number of neural tube malformations in children exposed to PB (N = 146) and control children (N = 512) (Analysis 7.2). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.02 to 0.03; I² = 0%).

7.2.3 PB versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

7.2.4 PB versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

7.3 Cardiac malformations

7.3.1 PB versus no medication (in women without epilepsy): cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 7.80, 95% CI 0.36 to 168.52, I² = NA), with no difference in the number of cardiac malformations in children exposed to PB (N = 7) and control children (N = 244) (Analysis 7.3). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.27 to 0.26; I² = 0%).

7.3.2 PB versus no medication (in women with epilepsy): cohort studies

Pooled results from four studies suggested no evidence of a difference in risk (RR 1.80, 95% CI 0.69 to 4.71, I² = 0%), with no difference in the number of cardiac malformations in children exposed to PB (N = 149) and control children (N = 516) (Analysis 7.3). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.02 to 0.05; I² = 0%).

7.3.3 PB versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

7.3.4 PB versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

7.4 Oro‐facial cleft/craniofacial malformations

7.4.1 PB versus no medication (in women without epilepsy): cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 3.34, 95% CI 0.20 to 56.35, I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to PB (N = 7) and control children (N = 244) (Analysis 7.4). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.28 to 0.25; I² = 0%).

7.4.2 PB versus no medication (in women with epilepsy): cohort studies

We were unable to estimate a RR from two studies due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to PB (N = 9) or control children (N = 172) (Analysis 7.4).

7.4.3 PB versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

7.4.4 PB versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

7.5 Skeletal/limb malformations

7.5.1 PB versus no medication (in women without epilepsy): cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 7.80, 95% CI 0.36 to 168.52, I² = NA) with no difference in the number of skeletal/limb malformations in children exposed to PB (N = 7) in comparison to control children (N=244). (Analysis 7.5). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.27 to 0.26; I² = NA).

7.5.2 PB versus no medication (in women with epilepsy): cohort studies

Pooled results from three studies suggested no evidence of a difference in risk (RR 3.01, 95% CI 0.56 to 16.07; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to PB (N = 146) and control children (N = 512) (Analysis 7.5). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.02 to 0.03; I² = 0%).

7.5.3 PB versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

7.5.4 PB versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

Phenobarbital dose

Despite data being reported in 26 studies, most studies did not investigate dose or report the results of analyses of PB dose with regard to MCM risk or were too limited in terms of the number of included pregnancies. EURAP 2018 included 294 PB monotherapy‐exposed cases which is the largest cohort. They found that increasing PB dose was associated with an increasing prevalence of MCM risk. Doses =/< 80 mg/d had a prevalence of 2.7% (95% CI 0.3 to 9.5), doses > 80 to =/< 130 mg/d had a prevalence of 6.2% (95%CI 3.0 to 11.1), and doses > 130 mg/d had the highest prevalence of 11.7% (95% CI 4.8 to 22.6); there was evidence of a dose association the for comparison of the lowest and highest PB dose levels investigated (OR 5.41, 95% CI 1.05 to 27.89, P = 0.0436). PB doses > 130 mg/d were associated with a higher MCM risk than LTG at doses =/< 325 mg/d (OR 5.81, 95% CI 2.40 to 14.08, P = 0.0002). There were no comparisons of the different PB dose levels to other ASM doses, however. The Kerala Epilepsy and Pregnancy Registry reported on 137 pregnancies and demonstrated an increase in MCM risk with increasing dose; PB > 200 mg/d had a prevalence of 10.3% whilst PB doses > 45 to 60 mg/d had a prevalence of 3.5%. However, it is possible that there was some case overlap with the EURAP 2018 cases as the Kerala Epilepsy and Pregnancy Registry is a EURAP collaborator. The collaboration reported by Samren 1997 and colleagues reported a likely dose association with PB. The North American Epilepsy and Pregnancy Register included 199 PB‐exposed pregnancies and did not find an association with dose. Kaneko 1999 did find an association between PB exposure (N = 79) and increased malformation rate. Other studies were too small or did not investigate an association between PB dose and MCM risk.

Phenytoin

The prevalence of major malformations (any type) in cohort studies for children exposed to phenytoin (PHT) (N = 1327), based on data from 26 studies, was 5.4% (95% CI 3.6 to 8.1). The prevalence of major malformations in routine health record studies for children exposed to PHT (N = 103), based on data from one study, was 6.8% (95% CI 0.1 to 91.3).

8 PHT versus controls

8.1 All major malformations

8.1.1 PHT versus no medication (in women without epilepsy): cohort studies

Pooled results from eight studies suggested an increased risk with PHT (RR 3.81, 95% CI 1.91 to 7.57; I² = 35%), with children exposed to PHT (N = 496) experiencing more major malformations than control children (N = 1397) (Analysis 8.1). The RD also suggested a higher risk for PHT (RD 0.03, 95% CI 0.01 to 0.06; I² = 44%).

Samren 1997 reported nine cases of major malformation in 141 (6%) PHT‐exposed children. Outcomes at centres with a control group in this study were limited to five cases from 33 exposed children, which gave a non‐significant difference (RR 2.2, 95% CI 0.7 to 6.7).

8.1.2 PHT versus no medication (in women with epilepsy): cohort studies

Pooled results from 15 studies suggested an increased risk with PHT (RR 2.01, 95% CI 1.29 to 3.12; I² = 0%), with children exposed to PHT (N = 750) experiencing more major malformations than control children (N = 1588) (Analysis 8.1). The RD also suggested a higher risk for PHT (RD 0.03, 95% CI 0.01 to 0.05; I² = 0%).

8.1.3 PHT versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

8.1.4 PHT versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

8.2 Neural tube malformations

8.2.1 PHT versus no medication (in women without epilepsy): cohort studies

Pooled results from four studies suggested no evidence of a difference in risk (RR 13.17, 95% CI 0.58 to 299.00, I² = NA) with no difference in the number of neural tube malformations in children exposed to PHT (N = 48) and control children (N = 590) (Analysis 8.2). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.06 to 0.06; I² = 0%).

8.2.2 PHT versus no medication (in women with epilepsy): cohort studies

Pooled results from six studies suggested no evidence of a difference in risk (RR 2.56, 95% CI 0.64 to 10.17; I² = 28%), with no difference in the number of neural tube malformations in children exposed to PHT (N = 252) and control children (N = 595) (Analysis 8.2). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.01 to 0.03; I² = 0%).

8.2.3 PHT versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

8.2.4 PHT versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

8.3 Cardiac malformations

8.3.1 PHT versus no medication (in women without epilepsy): cohort studies

Pooled results from four studies suggested no evidence of a difference in risk (RR 6.31, 95% CI 0.75 to 52.91, I² = 0%), with no difference in the number of cardiac malformations in children exposed to PHT (N = 48) and control children (N = 590) (Analysis 8.3). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.05 to 0.08; I² = 0%).

8.3.2 PHT versus no medication (in women with epilepsy): cohort studies

Pooled results from seven studies suggested no evidence of a difference in risk (RR 1.86, 95% CI 0.72 to 4.80; I² = 0%), with no difference in the number of cardiac malformations in children exposed to PHT (N = 253) and control children (N = 599) (Analysis 8.3). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.01 to 0.04; I² = 0%).

8.3.3 PHT versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

8.3.4 PHT versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

8.4 Oro‐facial cleft/craniofacial malformations

8.4.1 PHT versus no medication (in women without epilepsy): cohort studies

Pooled results from four studies suggested no evidence of a difference in risk (RR 0.67, 95% CI 0.04 to 12.54, I² = NA), with no difference in the number of oro‐facial cleft/ craniofacial malformations in children exposed to PHT (N = 48) and control children (N = 590) (Analysis 8.4). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.08 to 0.05; I² = 0%).

8.4.2 PHT versus no medication (in women with epilepsy): cohort studies

We were unable to estimate a RR from five studies due to no reported oro‐facial cleft/craniofacial malformations in children exposed to PHT (N = 133) and control children (N = 530) (Analysis 8.4).

8.4.3 PHT versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

8.4.4 PHT versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

8.5 Skeletal/limb malformations

8.5.1 PHT versus no medication (in women without epilepsy)

Pooled results from four studies suggested no evidence of a difference in risk (RR 1.56, 95% CI 0.07 to 37.19, I² = NA), with no difference in the number of skeletal/limb malformations in children exposed to PHT (N = 48) and control children (N = 590) (Analysis 8.5). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.07 to 0.06; I² = 0%).

8.5.2 PHT versus no medication (in women with epilepsy)

Pooled results from six studies suggested no evidence of a difference in risk (RR 1.57, 95% CI 0.31 to 7.95; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to PHT (N = 252) and control children (N = 595) (Analysis 8.5). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.02 to 0.02; I² = 0%).

8.5.3 PHT versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

8.5.4 PHT versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

Phenytoin dose

The majority of included studies did not investigate or formally report on the relationship between the dose of PHT and malformation outcome, with many being limited by included numbers of PHT‐exposed pregnancies. The North American Epilepsy and Pregnancy Register, based on 416 exposed children, did not find an increased MCM with higher doses of PHT. Kaaja 2003 with 124 PHT‐exposed children also reported no association with dose. However, in contrast, Kaneko 1999 reported evidence of an association between PHT dose and MCM prevalence, based on 132 children exposed to monotherapy PHT (no further details given). EURAP 2018 included 125 pregnancies with PHT exposure and reported a prevalence of 6.4% (95% CI 2.8 to 12.2). They did not investigate within‐group dose associations because of group size but they did report that, in comparison to LTG at doses =/< 325 mg/d, children exposed to PHT at doses between =/> 30 mg/d to 730 mg/d demonstrated no evidence of a difference in risk (OR 1.93, 95% CI 0.78 to 4.75); but this should be considered with caution due to the wide range of PHT doses included. Data from other included studies were limited by group size or dose associations were not reported.

Primidone

The prevalence of major malformations (any type) in cohort studies for children exposed to primidone (PRM) (N = 112), based on data from seven studies, was 7.9% (95% CI 2.6 to 21.5). The prevalence of major malformations in routine health record studies for children exposed to PRM (N = 3), was based on data from one study and therefore was not calculated.

9 PRM versus controls

9.1 All major malformations

9.1.1 PRM versus no medication (in women without epilepsy): cohort studies

The results from one study (Koch 1992) suggested no evidence of a difference in risk (RR 0.48, 95% CI 0.03 to 8.43, I² = NA) (Analysis 9.1) with no difference in the number of major malformations in children exposed to PRM (N = 21) and control children (N = 116). The RD also suggested no difference in the level of risk (RD ‐0.04, 95% CI −0.12 to 0.03; I² = NA).

Samren 1997 reported four cases of major malformations out of 43 PRM‐exposed children (9%). When limited to centres with control children, there were three cases out of 39 exposed children, which suggested no evidence of difference from control children (RR 1.0, 95% CI 0.3 to 3.8).

9.1.2 PRM versus no medication (in women with epilepsy): cohort studies

Pooled results from six studies suggested an increased risk with PRM (RR 3.61, 95% CI 1.41 to 9.23; I² = 8%), with children exposed to PRM (N = 108) experiencing more major malformations than control children (N = 573) (Analysis 9.1). The RD also suggested a higher risk for PRM (RD 0.07, 95% CI 0.00 to 0.14; I² = 11%).

9.1.3 PRM versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

9.1.4 PRM versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

Specific malformation types were not reviewed due to no reported data on these outcomes.

Primidone dose

No included studies investigated the dose of PRM and malformation risk.

Topiramate

The prevalence of major malformations (any type) in cohort studies for children exposed to topiramate (TPM) (N = 510), based on data from eight studies, was 3.9% (95% CI 2.3 to 6.5). The prevalence of major malformations in routine health record studies for children exposed to TPM (N = 49), based on data from two studies, was 4.1% (95% CI 0.0 to 27,060).

10 TPM versus controls

10.1 All major malformations

10.1.1 TPM versus no medication (in women without epilepsy): cohort studies

Pooled data from three studies suggested an increased risk with TPM (RR 4.07, 95% CI 1.64 to 10.14; I² = 0%), with children exposed to TPM (N = 367) experiencing more major malformations than control children (N = 825) (Analysis 10.1). The RD also suggested a higher risk for TPM (RD 0.03, 95% CI 0.01 to 0.06; I² = 0).

There was just one case of MCM in 41 monotherapy cases described by the Israeli Teratogen Service, giving a prevalence of 4.9%, which suggested no difference in risk to control children (3.4%, P value not reported).

10.1.2 TPM versus no medication (in women with epilepsy): cohort studies

Pooled results from five studies suggested no evidence of a difference in risk (RR 1.37, 95% CI 0.57 to 3.27; I² = 0%), with no difference in the number of major malformations in children exposed to TPM (N = 139) and control children (N = 1080) (Analysis 10.1). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.03 to 0.04; I² = 0%).

10.1.3 TPM versus no medication (in women without epilepsy): routine health record data studies

The results from one study suggested no evidence of a difference in risk (RR 1.65, 95% CI 0.43 to 6.42; I² = NA), with children exposed to TPM (N = 48) experiencing comparable rates of major malformations to control children (N = 369,267) (Analysis 10.1). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.04 to 0.07; I² = NA).

10.1.4 TPM versus no medication (in women with epilepsy): routine health record data studies

The results from one study suggested no evidence of a difference in risk (RR 1.62, 95% CI 0.40 to 6.45; I² = NA), with children exposed to TPM (N = 48) experiencing comparable rates of major malformations to control children (N = 1900) (Analysis 10.1). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.04 to 0.07; I² = NA).

10.2 Neural tube malformations

10.2.1 TPM versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR from two studies due to there being no reported neural tube malformations in children exposed to TPM (N = 8) or control children (N = 383) (Analysis 10.2).

10.2.2 TPM versus no medication (in women with epilepsy): cohort studies

We were unable to estimate a RR from three studies due to there being no reported neural tube malformations in children exposed to TPM (N = 59) and control children (N = 383) (Analysis 10.2).

10.2.3 TPM versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

10.2.4 TPM versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

10.3 Cardiac malformations

10.3.1 TPM versus no medication (in women without epilepsy): cohort studies

Pooled data from two included studies suggested evidence of a difference in risk (RR 20.71, 95% CI 2.64 to 162.72, I² = 0%), with children exposed to TPM (N = 8) experiencing more cardiac malformations than control children (N = 383) (Analysis 10.3). However, the RD suggested no difference in the level of risk (RD 0.12, 95% CI −0.16 to 0.39; I² = 0%).

10.3.2 TPM versus no medication (in women with epilepsy): cohort studies

Pooled data from four included studies suggested no evidence of a difference in risk (RR 2.48, 95% CI 0.49 to 12.49; I² = NA), with no difference in the number of cardiac malformations in children exposed to TPM (N = 60) and control children (N = 510) (Analysis 10.3). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.05 to 0.06; I² = 0%).

10.3.3 TPM versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

10.3.4 TPM versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

10.4 Oro‐facial cleft/craniofacial malformations

10.4.1 TPM versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR from two studies due to there being no reported oro‐facial cleft/ craniofacial malformations in children exposed to TPM (N = 8) or control children (N = 383) (Analysis 10.4).

10.4.2 TPM versus no medication (in women with epilepsy): cohort studies

Pooled data from three included studies suggested no evidence of a difference in risk (RR 1.50, 95% CI 0.09 to 24.92; I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to TPM (N = 51) and control children (N = 170) (Analysis 10.4). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.05 to 0.04; I² = 0%).

10.4.3 TPM versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome to be included in the meta‐analysis. However, the study by Hernandez‐Diaz and colleagues using US Medicaid Registers could not be included in the meta‐analysis due to a lack of reporting of specific numbers of oral clefts. In comparison to children born to women without epilepsy (N = 1,322,955), the children exposed to TPM (N = 2425) had higher rates of oral clefts of 4.1 per 1000 live births (RR 3.63, 95% CI 1.95 to 6.76).

10.4.4 TPM versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

10.5 Skeletal/limb malformations

10.5.1 TPM versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR from two studies due to there being no reported skeletal/limb malformations in children exposed to TPM (N = 8) or control children (N = 383) (Analysis 10.5).

10.5.2 TPM versus no medication (in women with epilepsy): cohort studies

Pooled data from three included studies suggested no evidence of a difference in risk (RR 2.06, 95% CI 0.24 to 17.42; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to TPM (N = 59) and control children (N = 502) (Analysis 10.5). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.05 to 0.04; I² = 0%).

10.5.3 TPM versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

10.5.4 TPM versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

Topiramate dose

The largest included cohort of TPM‐exposed pregnancies came from the study by Hernandez‐ Diaz using data from the US Medicaid Registers (N = 2425). This register reported the risk of oral clefts for doses ≤ 100 mg/d as 2.4 per 1000 live births, and for doses > 100 mg/d, as 7.3 per 1000 live births. The adjusted values of corresponding adjusted RRs for daily doses ≤ 100 and > 100 mg were 1.64 (95% CI 0.53 to 5.07) and 5.16 (95% CI 1.94 to 13.73) for lower and higher doses, respectively. The data were too limited to provide dose investigations specifically for women with epilepsy, but they did report that higher doses tended to be used for women requiring TPM for the treatment of epilepsy.

North American Epilepsy and Pregnancy Register found no evidence of a difference in the median dose between TPM‐exposed children (N = 359) who had MCM versus those who did not (P value not reported). The Australian Epilepsy and Pregnancy Register (N = 53), did not find a dose association for monotherapy TPM but did see an increase in risk with polytherapy (prevalence not given). The UK and Ireland Epilepsy and Pregnancy Register cohort (N = 70) also failed to find an association between the dose of TPM and the risk of overall MCM. However, caution is required due to smaller numbers from the Epilepsy and Pregnancy Register cohorts currently for monotherapy TPM exposure in pregnancy.

Valproate

The prevalence of major malformations (any type) in cohort studies for children exposed to valproate (VPA) (N = 3018), based on data from 31 studies, was 9.8% (95% CI 8.1 to 11.9). The prevalence of major malformations in routine health record studies for children exposed to VPA (N = 1364), based on data from six studies, was 9.7% (95% CI 7.1 to 13.4).

11 VPA versus controls

11.1. All major malformations

11.1.1 VPA versus no medication (in women without epilepsy): cohort studies

Pooled results from 10 studies suggested an increased risk with VPA (RR 5.53, 95% CI 3.29 to 9.29; I² = 0%), with children exposed to VPA (N = 501) experiencing more major malformations than control children (N = 2634) (Analysis 11.1). The RD also suggested a higher risk for VPA (RD 0.07, 95% CI 0.04 to 0.10; I² = 40%).

Data from the Israeli Teratogen Service study, including women treated with VPA for epilepsy and other indications (restricted to monotherapy), reported major congenital malformations (MCM) in 3/89 (3.4%) VPA‐treated cases compared with 31/1236 (2.5%) of control children. Samren 1997 reported 16 cases of major malformations out of 184 (9%) VPA‐exposed children. When limited to the two sites with control children, investigators reported six cases with malformation out of 21 children exposed to VPA, which was higher than control children (RR 4.9, 95% CI 1.6 to 15.0).

11.1.2 VPA versus no medication (in women with epilepsy): cohort studies

Pooled results from 17 studies suggested an increased risk with VPA (RR 2.77, 95% CI 2.03 to 3.79; I² = 0%), with children exposed to VPA (N = 2288) experiencing more major malformations than control children (N = 1710) (Analysis 11.1). The RD also suggested a higher risk for VPA (RD 0.06, 95% CI 0.04 to 0.07; I² = 32%).

11.1.3 VPA versus no medication (in women without epilepsy): routine health record data studies

Pooled results from three studies suggested an increased risk with VPA (RR 2.29, 95% CI 1.71 to 3.08; I² = 0%), with children exposed to VPA (N = 621) experiencing more major malformations than control children (N = 373,028) (Analysis 11.1). The RD also suggested a higher risk for VPA (RD 0.04, 95% CI 0.02 to 0.06; I² = 0%).

11.1.4 VPA versus no medication (in women with epilepsy): routine health record data studies

Pooled results from four studies suggested an increased risk with VPA (RR 3.01, 95% CI 2.42 to 3.75; I² = 55%), with children exposed to VPA (N = 1151) experiencing more major malformations than control children (N = 12,218) (Analysis 11.1). Due to high heterogeneity, a random‐effects RR was calculated which found a similar result (RR 2.97, 95% CI 2.08 to 4.24, I² =55%). The RD also suggested a higher risk for VPA (RD 0.06, 95% CI 0.05 to 0.08; I² = 81%). Due to high heterogeneity, a random‐effects RD was calculated which found a similar result (RD 0.06, 95% CI 0.02 to 0.10, I² =85%).

11.2 Neural tube malformations

11.2.1 VPA versus no medication (in women without epilepsy): cohort studies

Pooled results from four studies suggested no evidence of a difference in risk (RR 6.05, 95% CI 0.94 to 38.81; I² = 20%), with no difference in the number of neural tube malformations in children exposed to VPA (N = 104) and control children (N = 836) (Analysis 11.2). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.02 to 0.04; I² = 0%).

11.2.2 VPA versus no medication (in women with epilepsy): cohort studies

Pooled results from eight studies suggested an increased risk with VPA (RR 5.64, 95% CI 1.37 to 23.24; I² = 0%), with a higher number of neural tube malformations in children exposed to VPA (N = 814) than in control children (N = 664) (Analysis 11.2). The RD also suggested a higher risk for VPA (RD 0.02, 95% CI 0.01 to 0.03; I² = 6%).

11.2.3 VPA versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

11.2.4 VPA versus no medication (in women with epilepsy): routine health record data studies

Data from one study suggested an increased risk with VPA (RR 8.02, 95% CI 1.48 to 43.50, I² = NA), with a higher number of neural tube malformations in children exposed to VPA (N = 225) than in control children (N = 902) (Analysis 11.2). The RD also suggested a higher risk for VPA (RD 0.02, 95% CI −0.00 to 0.03; I² = NA).

11.3 Cardiac malformations

11.3.1 VPA versus no medication (in women without epilepsy): cohort studies

Pooled results from four studies suggested an increased risk with VPA (RR 11.89 95% CI 2.88 to 49.08; I² = 0%), with children exposed to VPA (N = 104) experiencing more cardiac malformations than control children (N = 836) (Analysis 11.3). However, the RD suggested no difference in the level of risk (RD 0.04, 95% CI −0.00 to 0.09; I² = 28%).

11.3.2 VPA versus no medication (in women with epilepsy): cohort studies

Pooled results from 10 studies suggested an increased risk with VPA (RR 2.71, 95% CI 1.42 to 5.17; I² = 0%), with a higher number of cardiac malformations in children exposed to VPA (N = 821) than in control children (N = 676) (Analysis 11.3). The RD also suggested a higher risk for VPA (RD 0.03, 95% CI 0.01 to 0.05; I² = 0%).

11.2.3 VPA versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

11.2.4 VPA versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

11.4 Oro‐facial cleft/craniofacial malformations

11.4.1 VPA versus no medication (in women without epilepsy): cohort studies

Pooled results from four studies suggested no evidence of a difference in risk (RR 2.76, 95% CI 0.31 to 24.78; I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to VPA (N = 104) and control children (N = 836) (Analysis 11.4). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.02 to 0.04; I² = 0%).

11.4.2 VPA versus no medication (in women with epilepsy): cohort studies

Pooled results from eight studies suggested an increased risk with VPA (RR 4.44, 95% CI 1.14 to 17.27; I² = 2%), with more children exposed to VPA (N = 474) experiencing oro‐facial cleft/craniofacial malformations than control children (N = 332) (Analysis 11.4). The RD also suggested a higher risk for VPA (RD 0.02, 95% CI 0.00 to 0.05; I² = 0%).

11.4.3 VPA versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

11.4.4 VPA versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

11.5 Skeletal/limb malformations

11.5.1 VPA versus no medication (in women without epilepsy): cohort studies

Pooled results from four studies suggested an increased risk with VPA (RR 16.48, 95% CI 2.46 to 110.49; I² = 0%), with children exposed to VPA (N = 104) experiencing more skeletal/limb malformations than control children (N = 836) (Analysis 11.5). However, the RD suggested no difference in the level of risk (RD 0.03, 95% CI −0.01 to 0.07; I² = 0%).

11.5.2 VPA versus no medication (in women with epilepsy): cohort studies

Pooled results from eight studies suggested no evidence of a difference in risk (RR 2.38, 95% CI 0.93 to 6.12; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to VPA (N = 814) and control children (N = 664) (Analysis 11.5). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.00 to 0.03; I² = 0%).

11.5.3 VPA versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

11.5.4 VPA versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

Valproate dose

In contrast to the results on dosage for the other AEDs, for VPA there appears to be a consistently documented and clear association between increased dose and the risk for MCM in VPA‐exposed children. EURAP 2018 reported evidence that suggested a dose‐related MCM risk for VPA exposure. In 1381 exposed pregnancies, the MCM risk ranged from 6.3% (95% CI 4.5 to 8.6%) for doses =/< 650 mg/d, to 11.3% for doses > 650 mg/d to =/< 1450 mg/day and, most concerning, 25.2% (95% CI 17.6 to 34.2) for doses > 1450 mg/d. Doses of VPA =/< 650 mg/d (OR 2.70, 95% CI 1.67 to 4.38, P = 0.0002), doses > 650 mg/d to =/< 1450 mg/day (OR 4.72, 95% CI 3.11 to 7.18, P = 0.0002) and doses > 1450 mg/d (OR 13.52, 95% CI 7.73 to 23.64, P = 0.0002) were all associated with higher risk than LTG exposure at doses < 325 mg/d. Similarly, doses of > 650 mg/d to =/< 1450 mg/day (OR 2.81, 95% CI 1.70 to 4.65, P = 0.0002) had higher risk than LTG > 325 mg/d. The highest level of VPA exposure was not statistically compared to LTG doses > 325 mg/d, but there was a large difference in prevalence (25.2% vs 4.3%). The lowest doses of VPA investigated (=/< 650 mg/d) were not associated with a lower MCM risk than higher doses (> 325 mg/d) of LTG (OR 0.62, 95% CI 0.36 to 1.09, P = 0.0959).

In the UK and Ireland Epilepsy and Pregnancy Register (N = 1220), an increase in malformation from 5.0% at doses < 600 mg/d to 10.4% for doses > 1000 mg/d (OR 2.20 95% CI 1.26 to 3.82, P = 0.0045) was reported. The Australian Epilepsy and Pregnancy Register cohort also demonstrated an association with VPA (N = 290), as did the North American Epilepsy and Pregnancy Register (N = 323), where investigators reported the median daily dose in VPA‐exposed children with a malformation to be 1000 mg/d compared with children exposed to VPA without an MCM (750 mg/d). The Kerala Epilepsy and Pregnancy Registry reported a prevalence of MCM of 3.0% for doses of VPA =/< 400 mg/d, 9.5% for doses between > 400 to 800 mg/d, and 28.6% for doses over 800 mg/d. Smaller studies including VPA‐exposed children also reported data showing an association between VPA dose or serum levels and increased MCM rate (Israeli Teratogen Service; Kaneko 1999; Lindhout 1992; Meador 2006; Milan Study 1999; Samren 1997). Kaaja 2003 was the only smaller study that investigated a dose‐response association without finding a positive correlation (N = 61 VPA‐exposed pregnancies).

Investigations from studies using population health record data are fewer, due to the lack of dose information available for the Norwegian Health Record Registers, Sweden Health Record Registers, and the absence of dose information for the Denmark Health Record Registers or UK Clinical Research Practice Database; UK Health Record THIN Register at this time. Putignano and colleagues 2019, using the Italian Lombardy Region Health Register, reported that children with MCMs had a higher dose of VPA.

Zonisamide

The prevalence of major malformations (any type) in cohort studies for children exposed to zonisamide (ZNS) (N = 130), based on data from four studies, was 2.7% (95% CI 0.1 to 47.3). There were no children exposed to ZNS in routine health record studies, therefore, the prevalence of major malformations rated could not be calculated.

12 ZNS versus controls

12.1. All major malformations

12.1.1 ZNS versus no medication (in women without epilepsy): cohort studies

Pooled data from two studies suggested no evidence of a difference in risk (RR 1.13, 95% CI 0.21 to 6.11; I² = 36%), with no difference in the number of major malformations in children exposed to ZNS (N = 103) and control children (N = 548) (Analysis 12.1). The RD also suggested no difference in the level of risk (RD −0.00, 95% CI −0.03 to 0.02; I² = 39%).

12.1.2 ZNS versus no medication (in women with epilepsy): cohort studies

Pooled data from two studies suggested an increased risk with ZNS (RR 3.20, 95% CI 1.09 to 9.43; I² = 0%), with a higher number of major malformations in children exposed to ZNS (N = 39) than in control children (N = 556) (Analysis 12.1). However, the RD suggested no difference in the level of risk (RD 0.07, 95% CI −0.03 to 0.18; I² = 0%).

12.1.3 ZNS versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

12.1.4 ZNS versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

12.2. Neural tube malformations

12.2.1 ZNS versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR from one included study due to there being no reported neural tube malformations in children exposed to ZNS (N = 13) or control children (N = 106) (Analysis 12.2).

12.2.2 ZNS versus no medication (in women with epilepsy): cohort studies

We were unable to estimate a RR from one included study due to there being no reported neural tube malformations in children exposed to ZNS (N = 13) or control children (N = 15) (Analysis 12.2).

12.2.3 ZNS versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

12.2.4 ZNS versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

12.3 Cardiac malformations

12.3.1 ZNS versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR from one included study due to there being no reported cardiac malformations in children exposed to ZNS (N = 13) or control children (N = 106) (Analysis 12.3).

12.3.2 ZNS versus no medication (in women with epilepsy): cohort studies

We were unable to estimate a RR from one included study due to there being no reported cardiac malformations in children exposed to ZNS (N = 13) or control children (N = 15) (Analysis 12.3).

12.3.3 ZNS versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

12.3.4 ZNS versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

12.4 Oro‐facial cleft/craniofacial malformations

12.4.1 ZNS versus no medication (in women without epilepsy): cohort studies

We were unable to estimate a RR from one included study due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to ZNS (N = 13) or control children (N = 106) (Analysis 12.4).

12.4.2 ZNS versus no medication (in women with epilepsy): cohort studies

We were unable to estimate a RR from one included study due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to ZNS (N = 13) or control children (N = 15) (Analysis 12.4).

12.4.3 ZNS versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

12.4.4 ZNS versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

12.5 Skeletal/limb malformations

12.5.1 ZNS versus no medication (in women without epilepsy)

We were unable to estimate a RR from one included study due to there being no skeletal/limb malformations in children exposed to ZNS (N = 13) or control children (N = 106) (Analysis 12.5).

12.5.2 ZNS versus no medication (in women with epilepsy)

We were unable to estimate a RR from one included study due to there being no reported skeletal/limb malformations in children exposed to ZNS (N = 13) or control children (N = 15) (Analysis 12.5).

12.5.3 ZNS versus no medication (in women without epilepsy): routine health record data studies

No included studies reported data on this outcome.

12.5.4 ZNS versus no medication (in women with epilepsy): routine health record data studies

No included studies reported data on this outcome.

Zonisamide dose

No included study investigated a potential association between ZNS and malformation risk.

ASM versus ASM comparisons

13 CBZ versus CZP

13.1. All major malformations

13.1.1 Cohort studies

Pooled results from four studies suggested no evidence of a difference in risk (RR 1.82, 95% CI 0.63 to 5.26; I² = 0%), with no difference in the number of major malformations in children exposed to CBZ (N = 1311) and children exposed to CZP (N = 95) (Analysis 13.1). However, the RD suggested a higher risk for CBZ (RD 0.04, 95% CI ‐0.00 to 0.08; I² = 0%).

13.1.2 Routine health record data studies

Pooled data from two studies suggested no evidence of a difference in risk (RR 1.29, 95% CI 0.47 to 3.51; I² = 0%), with no difference in the number of major malformations in children exposed to CBZ (N = 1388) and children exposed to CZP (N = 161). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI ‐0.00 to 0.04; I² = 0%).

13.2 Neural tube malformations

13.2.1 Cohort studies

No included studies reported data on this outcome.

13.2.2 Routine health record data studies

No included studies reported data on this outcome.

13.3 Cardiac malformations

13.3.1 Cohort studies

No included studies reported data on this outcome.

13.3.2 Routine health record data studies

No included studies reported data on this outcome.

13.4 Oro‐facial cleft/craniofacial malformations

13.4.1 Cohort studies

No included studies reported data on this outcome.

13.4.2 Routine health record data studies

No included studies reported data on this outcome.

13.5 Skeletal/limb malformations

13.5.1 Cohort studies

No included studies reported data on this outcome.

13.5.2 Routine health record data studies

No included studies reported data on this outcome.

14 CBZ versus GBP

14.1. All major malformations

14.1.1 Cohort studies

Pooled results from four studies suggested no evidence of a difference in risk(RR 1.55, 95% CI 0.57 to 4.26; I² = 47%), with no difference in the number of major malformations in children exposed to CBZ (N = 3112) and children exposed to GBP (N = 192) (Analysis 14.1). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI ‐0.00 to 0.04; I² = 42%).

14.1.2 Routine health record data studies

Data from one study suggested no evidence of a difference in risk (RR 1.54, 95% CI 0.10 to 24.27; I² = NA), with no difference in the number of major malformations in children exposed to CBZ (N = 703) and children exposed to GBP (N = 18) (Analysis 14.1). The RD also suggested no difference in the level of risk (RD 0.04, 95% CI ‐0.03 to 0.11; I² = NA).

14.2 Neural tube malformations

14.2.1 Cohort studies

Data from one included study suggested no evidence of a difference in risk (RR 0.12, 95% CI 0.01 to 2.93, I² = NA) with no difference in the number of neural tube malformations in children exposed to CBZ (N= 361) and GBP‐exposed children (N = 14) (Analysis 14.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.09 to 0.09; I² = NA).

14.2.2 Routine health record data studies

No included studies reported data on this outcome.

14.3 Cardiac malformations

14.3.1 Cohort studies

Pooled results from two studies suggest n increased in risk (RR 0.13, 95% CI 0.02 to 0.95, I² = 0%) with children exposed to GBP (N = 16) being at a higher risk of cardiac malformation than children exposed to CBZ (N = 374)(Analysis 14.3). However, the RD also suggested no difference in the level of risk (RD ‐0.05, 95% CI −0.18 to 0.08; I² = 74%).

14.3.2 Routine health record data studies

No included studies reported data on this outcome.

14.4 Oro‐facial cleft/craniofacial malformations

14.4.1 Cohort studies

Results from one included study suggested no evidence of a difference in risk (RR 0.37, 95% CI 0.02 to 6.62, I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to CBZ (N = 361) and children exposed to GBP (N = 14) (Analysis 14.4). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.08 to 0.10; I² = NA).

14.4.2 Routine health record data studies

No included studies reported data on this outcome.

14.5 Skeletal/limb malformations

14.5.1 Cohort studies

Results from one included study suggest no evidence of a difference in risk (RR 0.21, 95% CI 0.01 to 4.13, I² = NA), with no difference in the number of skeletal/limb malformations in children exposed to CBZ (N = 361) and children exposed to GBP (N = 14) (Analysis 14.5). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.09 to 0.10; I² = NA).

14.5.2 Routine health record data studies

No included studies reported data on this outcome.

15 CBZ versus LEV

15.1. All major malformations

15.1.1 Cohort studies

Pooled results from 11 studies suggested an increased risk with CBZ(RR 1.51, 95% CI 1.01 to 2.26; I² = 0%), with more children exposed to CBZ (N = 3814) experiencing major malformations than children exposed to LEV (N = 1242) (Analysis 15.1). The RD also suggested a higher risk for CBZ (RD 0.01, 95% CI 0.00 to 0.02; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 5.5% (95% CI 4.5 to 6.6%) for children exposed to CBZ and 2.8% (95% CI 1.7 to 4.5) for children exposed to LEV. No direct statistical comparison was made at the group level, investigations were made across different doses of the two ASMs (see Carbamazepine dose and Levetiracetam dose sections).

15.1.2 Routine health record data studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 1.73, 95% CI 0.78 to 3.83; I² = 0%), with no difference in the number of major malformations in children exposed to CBZ (N = 1000) and children exposed to LEV (N = 248) (Analysis 15.1). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI ‐0.00 to 0.04; I² = 0%).

15.2 Neural tube malformations

15.2.1 Cohort studies

Pooled results from 10 studies suggested no evidence of a difference in risk (RR 1.57, 95% CI 0.41 to 6.08; I² = 0%), with no difference in the number of neural tube malformations in children exposed to CBZ (N = 3731) and children exposed to LEV (N = 1148) (Analysis 15.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.00 to 0.01; I² = 0%).

EURAP 2018 reported a prevalence of 0.35% (7/1957) for cases of neural tube anomaly in children exposed to CBZ and 0% (0/599) in children exposed to LEV.

15.2.2 Routine health record data studies

No included studies reported data on this outcome.

15.3 Cardiac malformations

15.3.1 Cohort studies

Pooled results from 11 studies suggested no evidence of a difference in risk (RR 1.20, 95% CI 0.57 to 2.52; I² = 0%), with no difference in the number of cardiac malformations in children exposed to CBZ (N = 3736) and children exposed to LEV (N = 1156) (Analysis 15.3). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.01 to 0.01; I² = 0%).

EURAP 2018 reported a prevalence of 1.4% (28/1957) for cases of a cardiac anomaly in children exposed to CBZ and 0.8% (5/599) in children exposed to LEV.

15.3.2 Routine health record data studies

No included studies reported data on this outcome.

15.4 Oro‐facial cleft/craniofacial malformations

15.4.1 Cohort studies

Pooled results from 10 studies suggested no evidence of a difference in risk (RR 1.79, 95% CI 0.43 to 7.41; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to CBZ (N = 3246) and children exposed to LEV (N = 1050) (Analysis 15.4). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.00 to 0.01; I² = 0%).

EURAP 2018 reported two cases of cleft lip or palate anomaly out of 1957 children exposed to CBZ and one case out of 599 children exposed to LEV.

15.4.2 Routine health record data studies

No included studies reported data on this outcome.

15.5 Skeletal/limb malformations

15.5.1 Cohort studies

Pooled results from 10 studies suggested no evidence of a difference in risk (RR 0.99, 95% CI 0.37 to 2.68; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to CBZ (N = 3731) and children exposed to LEV (N = 1147) (Analysis 15.5). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.00 to 0.01; I² = 0%).

15.5.2 Routine health record data studies

No included studies reported data on this outcome.

16 CBZ versus LTG

16.1. All major malformations

16.1.1: Cohort studies

Pooled results from 13 cohort studies suggested an increased risk with CBZ (RR 1.37, 95% CI 1.06 to 1.77; I² = 0%), with children exposed to CBZ (N = 4018) experiencing more major malformations than children exposed to LTG (N = 4550) (Analysis 16.1). The RD also suggested a higher risk for CBZ (RD 0.01, 95% CI 0.00 to 0.02; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 5.5% (95% CI 4.5 to 6.6%) for children exposed to CBZ and 2.9% (95% CI 2.3 to 3.7) for children exposed to LTG. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Carbamazepine dose and Lamotrigine dose sections).

16.1.2: Routine health record studies

Pooled results from four routine health record studies suggested no evidence of a difference in risk (RR 1.21, 95% CI 0.88 to 1.67; I² = 21%), with no difference in the number of major malformations in children exposed to CBZ (N = 2001) and LTG (N = 2502) (Analysis 16.1). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI ‐0.01 to 0.02; I² = 25%).

16.2. Neural tube malformations

16.2.1: Cohort studies

Pooled results from 12 cohort studies suggested no evidence of a difference in risk (RR 2.19, 95% CI 0.76 to 6.33; I² = 0%), with no difference in the number of neural tube malformations in children exposed to CBZ (N = 3935) and children exposed to LTG (N = 4406) (Analysis 16.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.00 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube malformations in those exposed to CBZ was 0.36% (7/1957) and 0.04% for those exposed to LTG (1/2514).

16.2.2: Routine health record studies

No included studies reported data on this outcome.

16.3. Cardiac malformations

16.3.1: Cohort studies

Pooled results from 12 cohort studies suggested no evidence of a difference in risk (RR 1.48, 95% CI 0.87 to 2.51; I² = 0%), with no difference in the number of cardiac malformations in children exposed to CBZ (N = 3933) and children exposed to LTG (N = 4407) (Analysis 16.3). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.00 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac malformations in those exposed to CBZ was 1.43% (28/1957) and 0.59% (15/2514) for those exposed to lamotrigine.

16.3.2: Routine health record studies

No included studies reported data on this outcome.

16.4. Oro‐facial cleft/craniofacial malformations

16.4.1: Cohort studies

Pooled results from 11 studies suggested no evidence of a difference in risk (RR 1.22, 95% CI 0.57 to 2.61; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to CBZ (N = 3443) and children exposed to LTG (N = 4357) (Analysis 16.4). However, only three studies contained occurrences of oro‐facial cleft/craniofacial malformations. The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.00 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to CBZ was 0.10% (2/1957) and 0.11% for children exposed to LTG (3/2514).

16.4.2: Routine health record studies

No included studies reported data on this outcome.

16.5. Skeletal/limb malformations

16.5.1: Cohort studies

Pooled results from 12 cohort studies suggested no evidence of a difference in risk (RR 1.86, 95% CI 0.82 to 4.22; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to CBZ (N = 3935) and children exposed to LTG (N = 4406) (Analysis 16.5). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.00 to 0.00; I² = 0%).

16.5.2: Routine health record studies

No included studies reported data on this outcome.

17 CBZ versus OXC

17.1. All major malformations

17.1.1 Cohort studies

Pooled results from 11 cohort studies suggested no evidence of a difference in risk (RR 1.26, 95% CI 0.74 to 2.15; I² = 20%), with no difference in the number of major malformations in children exposed to CBZ (N = 2499) and children exposed to OXC (N = 378) (Analysis 17.1). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.01 to 0.03; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 5.5% (95% CI 4.5 to 6.6%) for children exposed to CBZ and 3.0% (95% CI 1.4 to 5.4) for children exposed to OXC. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Carbamazepine dose and Oxcarbazepine dose sections).

17.1.2 Routine health record data studies

Pooled results from four routine data suggested an increased risk with CBZ (RR 0.64, 95% CI 0.44 to 0.91; I² = 89%), with children exposed to CBZ (N = 2508) experiencing more malformations than the children exposed to OXC (N = 507) (Analysis 17.1). Due to heterogeneity, a random‐effects RR was calculated which found no difference in risk (RR 0.75, 95% CI 0.15 to 3.72; I² = 89%). The RD also suggested no difference in the level of risk (RD ‐0.03, 95% CI ‐0.06 to 0.00, I² = 89%). Due to heterogeneity, a random‐effects RD was calculated which upheld similar findings (RD ‐0.02, 95% CI ‐0.11 to 0.07; I² = 89%).

17.2 Neural tube malformations

17.2.1 Cohort studies

Pooled results from nine cohort studies suggested no evidence of a difference in risk (RR 0.93, 95% CI 0.22 to 3.96; I² = 0%), with no difference in the number of neural tube malformations in children exposed to CBZ (N = 2403) and children exposed to OXC (N = 364) (Analysis 17.2). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI ‐0.01 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to CBZ was 0.36% (7/1957) and 0% for children exposed to OXC (0/333).

17.2.2 Routine health record data studies

Included studies did not report any data on this outcome.

17.3 Cardiac malformations

17.3.1 Cohort studies

Pooled results from 11 cohort studies suggested no evidence of a difference in risk (RR 0.56, 95% CI 0.23 to 1.38; I² = 0%), with no difference in the number of cardiac malformations in children exposed to CBZ (N = 2421) and children exposed to OXC (N = 368) (Analysis 17.3). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.02 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to CBZ was 1.46% (28/1957) and 1.2% for children exposed to OXC (4/333).

17.3.2 Routine health record data studies

Included studies did not report any data on this outcome.

17.4 Oro‐facial cleft/craniofacial malformations

17.4.1 Cohort studies

Pooled results from nine cohort studies suggested no evidence of a difference in risk (RR 0.52, 95% CI 0.12 to 2.26; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to CBZ (N = 1918) and children exposed to OXC (N = 296) (Analysis 17.4). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.02 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to CBZ was 0.10% (2/1957) and 0.30% for children exposed to OXC (1/333).

17.4.2 Routine health record data studies

Included studies did not report any data on this outcome.

17.5 Skeletal/limb malformations

17.5.1 Cohort studies

Pooled results from nine studies suggested no evidence of a difference in risk (RR 0.53, 95% CI 0.17 to 1.66; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to CBZ (N = 2403) and children exposed to OXC (N = 364) (Analysis 17.5). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.01 to 0.01; I² = 0%).

17.5.2 Routine health record data studies

Included studies did not report any data on this outcome.

18 CBZ versus PB

18.1 All major malformations

18.1.1. Cohort studies

Pooled results from 24 cohort studies suggested no evidence of a difference in risk (RR 0.83, 95% CI 0.61 to 1.13; I² = 3%), with no difference in the number of major malformations in children exposed to CBZ (N = 3235) and children exposed to PB (N = 832) (Analysis 18.1). The RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.03 to 0.01; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 5.5% (95% CI 4.5 to 6.6%) for children exposed to CBZ and 6.5% (95% CI 4.2 to 9.9) for children exposed to PB. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Carbamazepine dose and Phenobarbital dose sections). Samren 1997 reported 22 major malformation cases in 280 (8%) CBZ‐exposed children and five cases in 48 (10%) PB‐exposed children.

18.1.2 Routine health record data studies

Pooled results from two routine health record studies suggested no evidence of a difference in risk (RR 0.35, 95% CI 0.12 to 1.09; I² = 0%), with no difference in the number of major malformations in children exposed to CBZ (N = 1388) and children exposed to PB (N = 34) (Analysis 18.1). The RD also suggested no difference in the level of risk (RD −0.06, 95% CI −0.15 to 0.04; I² = 0%).

18.2 Neural tube malformations

18.2.1 Cohort studies

Pooled results from 15 cohort studies suggested no evidence of a difference in risk (RR 1.28, 95% CI 0.35 to 4.75; I² = 32%), with no difference in the number of neural tube malformations in children exposed to CBZ (N = 2340) and children exposed to PB (N = 550) (Analysis 18.2). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to CBZ was 0.36% (7/1957) and 0.68% for children exposed to PB (2/294).

18.2.2 Routine health record data studies

Included studies did not report any data on this outcome.

18.3 Cardiac malformations

18.3.1 Cohort studies

Pooled results from 15 cohort studies suggested an increased risk with PB (RR 0.26, 95% CI 0.14 to 0.47; I² = 0%), with children exposed to CBZ (N = 2340) experiencing fewer cardiac malformations than children exposed to PB (N = 550) (Analysis 18.5). The RD also suggested a higher risk for PB (RD ‐0.03, 95% CI ‐0.05 to ‐0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to CBZ was 1.46% (28/1957) and 2.7% for children exposed to PB (8/333).

18.3.2 Routine health record data studies

Included studies did not report any data on this outcome.

18.4 Oro‐facial cleft/craniofacial malformations

18.4.1 Cohort studies

Pooled results from 15 cohort studies suggested an increased risk with PB (RR 0.18, 95% CI 0.07 to 0.48; I² = 0%), with children exposed to CBZ (N = 1857) experiencing fewer oro‐facial cleft/craniofacial malformations than children exposed to PB (N = 422) (Analysis 18.4). The RD suggested no difference in the level of risk for PB (RD ‐0.01, 95% CI ‐0.03 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to CBZ was 0.10% (2/1957) and 0.34% for children exposed to PB (1/294).

18.4.2 Routine health record data studies

Included studies did not report any data on this outcome.

18.5 Skeletal/limb malformations

18.5.1 Cohort studies

Pooled results from 15 cohort studies suggested no evidence of a difference in risk (RR 1.08, 95% CI 0.45 to 2.61; I² = 6%), with no difference in the number of skeletal/limb malformations in children exposed to CBZ (N = 2340) and children exposed to PB (N = 550) (Analysis 18.5). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.01 to 0.02; I² = 0%).

18.5.2 Routine health record data studies

Included studies did not report any data on this outcome.

19 CBZ versus PHT

19.1 All major malformations

19.1.1 Cohort studies

Pooled results from 23 cohort studies suggested no evidence of a difference in risk (RR 0.83, 95% CI 0.62 to 1.11; I² = 0%), with no difference in the number of major malformations in children exposed to CBZ (N = 4759) and children exposed to PHT (N = 1287) (Analysis 19.1). The RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.02 to 0.01; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 5.5% (95% CI 4.5 to 6.6%) for children exposed to CBZ and 6.4% (95% CI 2.8 to 12.2) for children exposed to PHT. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Carbamazepine dose and Phenytoin dose sections). Samren 1997 reported 22 cases of major malformation out of 280 (8%) CBZ‐exposed children and 9 cases from 141 PHT‐exposed children (9%).

19.1.2 Routine health record data studies

Results from one routine health record data study suggested no evidence of a difference in risk (RR 0.59, 95% CI 0.26 to 1.31; I² = NA), with no difference in the number of major malformations in children exposed to CBZ (N = 703) and children exposed to PHT (N = 103) (Analysis 19.1). The RD also suggested no difference in the level of risk (RD −0.03, 95% CI −0.08 to 0.02; I² = 0%).

19.2 Neural tube malformations

19.2.1 Cohort studies

Pooled results from 16 cohort studies suggested no evidence of a difference in risk (RR 1.12, 95% CI 0.45 to 2.83; I² = 0%), with no difference in the number of neural tube malformations in children exposed to CBZ (N = 4341) and children exposed to PHT (N = 1005) (Analysis 19.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.00 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to CBZ was 0.36% (7/1957) and 0.80% for children exposed to PB (1/125).

19.2.2 Routine health record data studies

Included studies did not report any data on this outcome.

19.3 Cardiac malformations

19.3.1 Cohort studies

Pooled results from 16 cohort studies suggested an increased risk with PHT (RR 0.44, 95% CI 0.23 to 0.84; I² = 8%), with fewer cardiac malformations in children exposed to CBZ (N = 4341) than in children exposed to PHT (N = 1005) (Analysis 19.3). However, the RD suggested no difference in the level of risk (RD ‐0.01, 95% CI ‐0.02 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to CBZ was 1.46% (28/1957) and 4% for children exposed to PHT (5/125).

19.3.2 Routine health record data studies

Included studies did not report any data on this outcome.

19.4 Oro‐facial cleft/craniofacial malformations

19.4.3 Cohort studies

Pooled results from 16 cohort studies suggested no evidence of a difference in risk (RR 0.81, 95% CI 0.32 to 2.08; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to CBZ (N = 3858) and children exposed to PHT (N = 891) (Analysis 19.4). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to CBZ was 0.10% (2/1957) and 0% for children exposed to PHT (0/125).

19.4.2 Routine health record data studies

Included studies did not report any data on this outcome.

19.5 Skeletal/limb malformations

19.5.1 Cohort studies

Pooled results from 16 cohort studies suggested no evidence of a difference in risk (RR 0.88, 95% CI 0.43 to 1.82; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to CBZ (N = 4341) and children exposed to PHT (N = 1005) (Analysis 19.5). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.01 to 0.01; I² = 0%).

19.5.2 Routine health record data studies

Included studies did not report any data on this outcome.

20 CBZ versus PRM

20.1 All major malformations

20.1.1 Cohort studies

Pooled results from seven cohort studies suggested no evidence of a difference in risk (RR 0.59, 95% CI 0.23 to 1.56; I² = 40%), with no difference in the number of major malformations in children exposed to CBZ (N = 1076) and children with PRM (N = 112) (Analysis 20.1). The RD also suggested no difference in the level of risk (RD −0.02, 95% CI −0.09 to 0.05; I² = 8%).

Samren 1997 reported 22 cases of major malformation out of 280 (8%) CBZ‐exposed children and 4 cases out of 43 (9%) PRM‐exposed children.

20.1.2 Routine health record data studies

Data from one included study suggested no evidence of a difference in risk (RR 0.32, 95% CI 0.02 to 4.44, I² = NA), with no difference in the number of major malformations in children exposed to CBZ (N = 703) and children exposed to PRM (N = 3) (Analysis 20.1). The RD also suggested no difference in the level of risk (RD 0.04, 95% CI −0.28 to 0.36; I² = NA).

20.2 Neural tube malformations

20.2.1 Cohort studies

Pooled data from two studies suggested no evidence of a difference in risk (RR 0.95, 95% CI 0.04 to 22.75, I² = NA), with no difference in the number of neural tube malformations in children exposed to CBZ (N = 119) and children exposed to PRM (N = 39) (Analysis 20.2). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.04 to 0.06; I² = 0%.

20.2.2 Routine health record data studies

Included studies did not report any data on this outcome.

20.3 Cardiac malformations

20.3.1 Cohort studies

Pooled data from two studies suggested no evidence of a difference in risk (RR 0.11, 95% CI 0.00 to 2.53, I² = NA), with no difference in the number of cardiac malformations in children exposed to CBZ (N = 119) and children exposed to PRM (N = 39) (Analysis 20.3). The RD also suggested no difference in the level of risk (RD ‐0.03, 95% CI −0.10 to 0.04; I² = 0%).

20.3.2 Routine health record data studies

Included studies did not report any data on this outcome.

20.4 Oro‐facial cleft/craniofacial malformations

20.4.1 Cohort studies

We were unable to estimate a RR from two studies due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to CBZ (N = 119) or children exposed to PRM (N = 39) (Analysis 20.4).

20.4.2 Routine health record data studies

Included studies did not report any data on this outcome.

20.5 Skeletal/limb malformations

20.5.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 2.84, 95% CI 0.16 to 51.53, I² = NA), with no difference in the number of skeletal/limb malformations in children exposed to CBZ (N = 119) and children exposed to PRM (N = 39) (Analysis 20.5). The RD also suggested no difference in the level of risk (RD 0.03, 95% CI −0.03 to 0.09; I² = 0%).

20.5.2 Routine health record data studies

Included studies did not report any data on this outcome.

21 CBZ versus TPM

21.1 All major malformations

21.1.1 Cohort studies

Pooled results from eight cohort studies suggested no evidence of a difference in risk (RR 0.83, 95% CI 0.51 to 1.33; I² = 0%), with no difference in the number of major malformations in children exposed to CBZ (N = 3651) and children exposed to TPM (N = 505) (Analysis 21.1). The RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.02 to 0.01; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 5.5% (95% CI 4.5 to 6.6%) for children exposed to CBZ and 3.9% (95% CI 1.5 to 8.4) for children exposed to TPM. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Carbamazepine dose and Topiramate dose sections).

21.1.2 Routine health record data studies

Pooled results from two routine health records suggested no evidence of a difference in risk (RR 0.59, 95% CI 0.17 to 2.06; I² = 12%), with children exposed to CBZ (N = 1388) experiencing more major malformations than children exposed to TPM (N = 49) (Analysis 21.1). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.07 to 0.05; I² = 0%).

21.2 Neural tube malformations

21.2.1 Cohort studies

Pooled results from seven cohort studies suggested no evidence of a difference in risk (RR 0.91, 95% CI 0.18 to 4.51; I² = 0%), with no difference in the number of neural tube malformations in children exposed to CBZ (N = 3568) and children exposed to TPM (N = 496) (Analysis 21.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to CBZ was 0.36% (7/1957) and 0% for children exposed to TPM (0/152).

21.2.2 Routine health record data studies

Included studies did not report any data on this outcome.

21.3 Cardiac malformations

21.3.1 Cohort studies

Pooled results from eight cohort studies suggested no evidence of a difference in risk (RR 0.73, 95% CI 0.25 to 2.12; I² = 0%), with no difference in the number of cardiac malformations in children exposed to CBZ (N = 3573) and children exposed to TPM (N = 497) (Analysis 21.3). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to CBZ was 1.46% (28/1957) and 1.97% for children exposed to TPM (3/152).

21.3.2 Routine health record data studies

Included studies did not report any data on this outcome.

21.4 Oro‐facial cleft/craniofacial malformations

21.4.1 Cohort studies

Pooled results from seven cohort studies suggested an increased risk with CBZ (RR 0.33, 95% CI 0.13 to 0.82; I² = 40%), with children exposed to CBZ (N = 3083) experiencing more oro‐facial cleft/craniofacial malformations than children exposed to TPM (N = 488) (Analysis 21.4). However, the RD suggested no difference in the level of risk (RD ‐0.01, 95% CI ‐0.02 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to CBZ was 0.10% (2/1957) and 0% for children exposed to TPM (0/152).

21.4.2 Routine health record data studies

Included studies did not report any data on this outcome.

21.5 Skeletal/limb malformations

21.5.1 Cohort studies

Pooled results from seven cohort studies suggested an increased risk with CBZ (RR 0.34, 95% CI 0.12 to 0.94; I² = 0%), with children exposed to CBZ (N = 3568) experiencing more skeletal/limb malformations than children exposed to TPM (N = 496). However, the RD suggested no difference in the level of risk (RD ‐0.01, 95% CI ‐0.02 to 0.01; I² = 0%).

21.5.2 Routine health record data studies

There were no studies that provided data for this outcome.

22 CBZ versus VPA

22.1. All major malformations

22.1.1 Cohort studies

Pooled results from 29 cohort studies suggested an increased risk with VPA (RR 0.44, 95% CI 0.37 to 0.53; I² = 0%), with children exposed to CBZ (N = 5133) experiencing fewer major malformations than children exposed to VPA (N = 2957) (Analysis 22.1). The RD also suggested a higher risk for VPA (RD −0.05, 95% CI −0.06 to −0.04; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 5.5% (95% CI 4.5 to 6.6%) for children exposed to CBZ and 10.3% (95% CI 8.8 to 12.0) for children exposed to VPA. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Carbamazepine dose and Valproate dose sections). Samren 1997 reported 22 cases of major malformation out of 280 (8%) CBZ‐exposed children and six cases out of 184 (9%) VPA‐exposed children.

22.1.2 Routine health record data studies

Pooled results from five routine health record studies suggested an increased risk with VPA (RR 0.42, 95% CI 0.33 to 0.54; I² = 14%), with children exposed to CBZ (N = 2806) experiencing fewer major malformations than children exposed to VPA (N = 1351) (Analysis 22.1). The RD also suggested a higher risk for VPA (RD −0.06, 95% CI −0.07 to −0.04; I² = 49%). Due to heterogeneity, a random‐effects RD was calculated which found a similar effect (RD ‐0.08, 95% CI ‐0.08 to ‐0.03, I² = 49%).

22.2 Neural tube malformations

22.2.1 Cohort studies

Pooled results from 21 cohort studies suggested an increased risk with VPA (RR 0.124, 95% CI 0.14 to 0.41; I² = 7%), with children exposed to CBZ (N = 4738) experiencing fewer neural tube malformations than children exposed to VPA (N = 2721) (Analysis 22.2). The RD also suggested a higher risk for VPA (RD ‐0.01, 95% CI ‐0.02 to ‐0.01; I² = 14%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to CBZ was 0.36% (7/1957) and 1.6% for children exposed to VPA (16/1381).

22.2.2 Routine health record data studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 0.19, 95% CI 0.02 to 2.09; I² = NA), with children exposed to CBZ (N = 703) experiencing comparable neural tube malformations to children exposed to VPA (N = 268) (Analysis 22.2). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI ‐0.02 to 0.00; I² = NA).

22.3 Cardiac malformations

22.3.1 Cohort studies

Pooled results from 22 cohort studies suggested an increased risk with VPA (RR 0.40, 95% CI 0.28 to 0.58; I² = 12%), with children exposed to CBZ (N = 4743) experiencing fewer cardiac malformations than children exposed to VPA (N = 2722) (Analysis 22.3). The RD also suggested a higher risk for VPA (RD ‐0.02, 95% CI ‐0.02 to ‐0.01; I² = 20%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to CBZ was 1.46% (28/1957) and 2.46% for children exposed to VPA (34/1381).

22.3.2 Routine health record data studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 0.38, 95% CI 0.13 to 1.08; I² = NA), with children exposed to CBZ (N = 703) experiencing comparable cardiac malformations to children exposed to VPA (N = 268) (Analysis 22.3). The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI ‐0.04 to 0.00; I² = NA).

22.4 Oro‐facial cleft/craniofacial malformations

22.4.1 Cohort studies

Pooled results from 22 cohort studies suggested an increased risk with VPA (RR 0.31, 95% CI 0.18 to 0.54; I² = 0%), with children exposed to CBZ (N = 4260) experiencing fewer oro‐facial cleft/craniofacial malformations than children exposed to VPA (N = 2387) (Analysis 22.4). The RD also suggested a higher risk for VPA (RD ‐0.01, 95% CI ‐0.02 to ‐0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to CBZ was 0.10% (2/1957) and 0.43% for children exposed to VPA (6/1381).

22.4.2 Routine health record data studies

Results from one routine health record study suggested an increased risk with VPA (RR 0.15, 95% CI 0.03 to 0.78; I² = NA), with children exposed to CBZ (N = 703) experiencing fewer major malformations than children exposed to VPA (N = 268) (Analysis 22.4). The RD also suggested a higher risk for VPA (RD ‐0.02, 95% CI ‐0.03 to 0.01; I² = NA).

22.5 Skeletal/limb malformations

22.5.1 Cohort studies

Pooled results from 21 cohort studies suggested an increased risk with VPA (RR 0.31, 95% CI 0.19 to 0.51; I² = 0%), with children exposed to CBZ (N = 4748) experiencing fewer skeletal/limb malformations than children exposed to VPA (N = 2711) (Analysis 22.5). The RD also suggested a higher risk for VPA (RD ‐0.01, 95% CI ‐0.02 to ‐0.01; I² = 0%).

22.5.2 Routine health record data studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 0.38, 95% CI 0.02 to 6.07; I² = NA), with children exposed to CBZ (N = 703) experiencing comparable skeletal/limb malformations to children exposed to VPA (N = 268) (Analysis 22.5). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.01 to 0.01; I² = 0%).

23 CBZ versus ZNS

23.1 All major malformations

23.1.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 0.94, 95% CI 0.36 to 2.44; I² = 75%), with no difference in the number of major malformations in children exposed to CBZ (N = 2711) and children exposed to ZNS (N = 130) (Analysis 23.1). Due to high heterogeneity, a random‐effects RR was calculated which also found no difference (RR 0.86, 95% CI 0.07 to 10.35, I² =75%). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.03 to 0.03; I² = 74%). Due to heterogeneity, a random‐effects RD was calculated which upheld a similar result (RD ‐0.02, 95% CI ‐0.15 to 0.12).

23.1.2 Routine health record data studies

Included studies did not report any data on this outcome.

23.2 Neural tube malformations

23.2.1 Cohort studies

Pooled results from three studies suggested evidence of a difference in risk (RR 0.06, 95% CI 0.01 to 0.54, I² = NA), with children exposed to CBZ (N = 1678) experiencing more neural tube malformations than children exposed to ZNS (N = 40) (Analysis 23.2). However, the RD suggested no difference in the level of risk (RD ‐0.03, 95% CI −0.10 to 0.04; I² = 0%).

23.2.2 Routine health record data studies

No included studies reported data on this outcome.

23.3 Cardiac malformations

23.3.1 Cohort studies

Pooled results from three studies suggested no evidence of a difference in risk (RR 0.47, 95% CI 0.03 to 7.72, I² = NA), with no difference in the number of cardiac malformations between children exposed to CBZ (N = 1678) and children exposed to ZNS (N = 40) (Analysis 23.3). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.05 to 0.06; I² = 0%).

23.3.2 Routine health record data studies

No included studies reported data on this outcome.

23.4 Oro‐facial cleft/craniofacial malformations

23.4.1 Cohort studies

Pooled results from three studies suggested no evidence of a difference in risk (RR 0.15, 95% CI 0.01 to 2.66, I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations between children exposed to CBZ (N = 1678) and children exposed to ZNS (N = 40) (Analysis 23.4). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.06 to 0.06; I² = 0%).

23.4.2 Routine health record data studies

No included studies reported data on this outcome.

23.5 Skeletal/limb malformations

23.5.1 Cohort studies

Pooled results from three studies suggested no evidence of a difference in risk (RR 0.15, 95% CI 0.01 to 2.66, I² = NA), with no difference in the number of skeletal/limb malformations between children exposed to CBZ (N = 1678) and children exposed to ZNS (N = 40) (Analysis 23.5). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.06 to 0.06; I² = 0%).

23.5.2 Routine health record data studies

No included studies reported data on this outcome.

24 GBP versus LTG

24.1 All major malformations

24.1.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 0.92, 95% CI 0.34 to 2.47; I² = 58%), with no difference in the number of major malformations in children exposed to GBP (N = 192) and children exposed to LTG (N = 4103) (Analysis 24.1). Due to high heterogeneity, a random‐effects RR was calculated which also found no difference (RR 1.54, 95% CI 0.25 to 9.55, I² =85%). The RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.03 to 0.01; I² = 37%).

24.1.2 Routine health record data studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 0.53, 95% CI 0.03 to 9.48; I² = NA), with children exposed to GBP (N = 18) experiencing more major malformations than children exposed to LTG (N = 90) (Analysis 24.1). The RD also suggested no difference in the level of risk (RD −0.04, 95% CI −0.13 to 0.01; I² = 37%).

24.2 Neural tube malformations

24.2.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported neural tube malformations in children exposed to GBP (N = 14) or in children exposed to LTG (N = 314) (Analysis 24.2).

24.2.2 Routine health record data studies

No included studies reported data on this outcome.

24.3 Cardiac malformations

24.3.1 Cohort studies

Pooled results from two studies suggested evidence of a difference in risk (RR 9.57, 95% CI 1.69 to 54.15, I² = 30%), with children exposed to GBP (N = 16) experiencing more cardiac malformations than children exposed to LTG (N = 352) (Analysis 24.3). However, the RD suggested no difference in the level of risk (RD 0.05, 95% CI −0.08 to 0.19; I² = 76%).

24.3.2 Routine health record data studies

No included studies reported data on this outcome.

24.4 Oro‐facial cleft/craniofacial malformations

24.4.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 1.92, 95% CI 0.11 to 33.05, I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations between children exposed to GBP (N = 14) and children exposed to LTG (N = 315) (Analysis 24.4). The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI −0.11 to 0.08; I² = NA).

24.4.2 Routine health record data studies

No included studies reported data on this outcome.

24.5 Skeletal/limb malformations

24.5.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported skeletal/limb malformations in children exposed to GBP (N = 14) or children exposed to LTG (N = 315) (Analysis 24.5).

24.5.2 Routine health record data studies

No included studies reported data on this outcome.

25 GBP versus OXC

25.1 All major malformations

25.1.1 Cohort studies

Pooled results from three cohort studies suggested no evidence of a difference in risk (RR 0.53, 95% CI 0.13 to 2.17; I² = 0%), with no difference in the number of major malformations in children exposed to GBP (N = 161) and children exposed to OXC (N = 202) (Analysis 25.1). The RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.04 to 0.01; I² = 0%).

25.1.2 Routine health record data studies

We were unable to estimate a RR from one study due to there being no reported major malformations in children exposed to GBP (N = 18) or children exposed to OXC (N = 4) (Analysis 25.1).

25.2 Neural tube malformations

25.2.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported neural tube malformations in children exposed to GBP (N = 14) or children exposed to OXC (N = 12) (Analysis 25.2).

25.2.2 Routine health record data studies

No included studies reported data on this outcome.

25.3 Cardiac malformations

25.3.1 Cohort studies

Included studies did not reach the threshold for reporting of the meta‐analysis (Analysis 25.3). However, available data show that there were 1/15 cases of cardiac malformation in children exposed to GBP and 0/13 in OXC children, based on data from two studies (Australian Epilepsy and Pregnancy Register; Miskov 2016).

25.3.2 Routine health record data studies

No included studies reported data on this outcome.

25.4 Oro‐facial cleft/craniofacial malformations

25.4.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to GBP (N = 14) or children exposed to OXC (N = 12) (Analysis 25.4).

25.4.2 Routine health record data studies

No included studies reported data on this outcome.

25.5 Skeletal/limb malformations

25.5.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported skeletal/limb malformations in children exposed to GBP (N = 14) or children exposed to OXC (N = 12) (Analysis 25.5).

25.5.2 Routine health record data studies

No included studies reported data on this outcome.

26 GBP versus PB

26.1 All major malformations

26.1.1 Cohort studies

Pooled results from three cohort studies suggested no evidence of a difference in risk (RR 0.30, 95% CI 0.08 to 1.14; I² = 75%), with children exposed to GBP (N = 161) experiencing no difference in major malformations to children exposed to PB (N = 204) (Analysis 26.1). Due to high heterogeneity, a random‐effects RR was calculated which also found no difference (RR 0.61, 95% CI 0.02 to 19.36, I² =85%). However, the RD suggested a higher risk for PB (RD −0.04, 95% CI −0.08 to ‐0.00; I² = 31%).

26.1.2 Routine health record data studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 0.14, 95% CI 0.01 to 3.09; I² = NA), with children exposed to GBP (N = 18) experiencing more major malformations than children exposed to PB (N = 1) (Analysis 26.1). The RD also suggested no difference in the level of risk (RD ‐0.14, 95% CI ‐0.42 to 0.14).

26.2 Neural tube malformations

26.2.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported neural tube malformations in children exposed to GBP (N = 14) or children exposed to PB (N = 5) (Analysis 26.2).

26.2.2 Routine health record data studies

No included studies reported data on this outcome.

26.3 Cardiac malformations

26.3.1 Cohort studies

Included studies did not reach the threshold for reporting of the meta‐analysis (Analysis 26.3). However, available data show that there were 1/16 cases of cardiac malformation in children exposed to GBP and 0/8 in children exposed to PB, based on data from two studies (Australian Epilepsy and Pregnancy Register; Miskov 2016).

26.3.2 Routine health record data studies

No included studies reported data on this outcome.

26.4 Oro‐facial cleft/craniofacial malformations

26.4.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to GBP (N = 14) or children exposed to PB (N = 5) (Analysis 26.4).

26.4.2 Routine health record data studies

No included studies reported data on this outcome.

26.5 Skeletal/limb malformations

26.5.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported skeletal/limb malformations in children exposed to GBP (N = 14) or children exposed to PB (N = 5) (Analysis 26.5).

26.5.2 Routine health record data studies

No included studies reported data on this outcome.

27 GBP versus PRM

27.1 All major malformations

27.1.1 Cohort studies

No included studies reported data on this outcome.

27.1.2 Routine health record data studies

We were unable to estimate a RR from one study due to there being no reported major malformations in children exposed to GBP (N = 18) or children exposed to PRM (N = 8) (Analysis 27.1).

27.2 Neural tube malformations

27.2.1 Cohort studies

No included studies reported data on this outcome.

27.2.2 Routine health record data studies

No included studies reported data on this outcome.

27.3 Cardiac malformations

27.3.1 Cohort studies

No included studies reported data on this outcome.

27.3.2 Routine health record data studies

No included studies reported data on this outcome.

27.4 Oro‐facial cleft/craniofacial malformations

27.4.1 Cohort studies

No included studies reported data on this outcome.

27.4.2 Routine health record data studies

No included studies reported data on this outcome.

27.5 Skeletal/limb malformations

27.5.1 Cohort studies

No included studies reported data on this outcome.

27.5.2 Routine health record data studies

No included studies reported data on this outcome.

28 GBP versus TPM

28.1 All major malformations

28.1.1 Cohort studies

Pooled results from three cohort studies suggested no evidence of a difference in risk (RR 0.32, 95% CI 0.09 to 1.19; I² = 0%), with no difference in the number of major malformations in children exposed to GBP (N = 190) and children exposed to TPM (N = 482) (Analysis 28.1). However, the RD suggested a higher risk for TPM (RD −0.03, 95% CI −0.05 to −0.01; I² = 0%).

28.1.2 Routine health record data studies

We were unable to estimate a RR from one study due to there being no reported major malformations in children exposed to GBP (N = 18) or children exposed to TPM (N = 1) (Analysis 28.1).

28.2 Neural tube malformations

28.2.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported neural tube malformations in children exposed to GBP (N = 14) or children exposed to TPM (N = 44) (Analysis 28.2).

28.2.2 Routine health record data studies

No included studies reported data on this outcome.

28.3 Cardiac malformations

28.3.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported cardiac malformations in children exposed to GBP (N = 14) or children exposed to TPM (N = 44) (Analysis 28.3).

28.3.2 Routine health record data studies

No included studies reported data on this outcome.

28.4 Oro‐facial cleft/craniofacial malformations

28.4.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to GBP (N = 14) or children exposed to TPM (N = 44) (Analysis 28.4).

28.4.2 Routine health record data studies

No included studies reported data on this outcome.

28.5 Skeletal/limb malformations

28.5.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported skeletal/limb malformations in children exposed to GBP (N = 14) or children exposed to TPM (N = 44) (Analysis 28.5).

28.5.2 Routine health record data studies

No included studies reported data on this outcome.

29 GBP versus ZNS

29.1 All major malformations

29.1.1 Cohort studies

Data from two cohort studies suggested no evidence of a difference in risk (RR 0.53, 95% CI 0.10 to 2.76; I² = 0%), with no difference in the number of major malformations in children exposed to GBP (N = 176) and children exposed to ZNS (N = 116) (Analysis 29.1). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.04 to 0.02; I² = 72%). Due to heterogeneity, a random‐effects RD was calculated and upheld a similar finding (RD ‐0.03, 95% CI ‐0.15 to 0.10, I² =72%).

29.1.2 Routine health record data studies

No included studies reported data on this outcome.

29.2 Neural tube malformations

29.2.1 Cohort studies

No included studies reported data on this outcome.

29.2.2 Routine health record data studies

No included studies reported data on this outcome.

29.3 Cardiac malformations

29.3.1 Cohort studies

No included studies reported data on this outcome.

29.3.2 Routine health record data studies

No included studies reported data on this outcome.

29.4 Oro‐facial cleft/craniofacial malformations

29.4.1 Cohort studies

No included studies reported data on this outcome.

29.4.2 Routine health record data studies

No included studies reported data on this outcome.

29.5 Skeletal/limb malformations

29.5.1 Cohort studies

No included studies reported data on this outcome.

29.5.2 Routine health record data studies

No included studies reported data on this outcome.

30 LEV versus GBP

30.1 All major malformations

30.1.1 Cohort studies

Pooled results from three studies suggested no evidence of a difference in risk (RR 1.61, 95% CI 0.46 to 5.63; I² = 43%), with no difference in the number of major malformations in children exposed to LEV (N = 893) and children exposed to GBP (N = 190) (Analysis 30.1). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.01 to 0.03; I² = 0%).

30.1.2 Routine health record data studies

No included studies reported data on this outcome.

30.2 Neural tube malformations

30.2.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported neural tube malformations in children exposed to LEV (N = 63) or children exposed to GBP (N = 14) (Analysis 30.2).

30.2.2 Routine health record data studies

No included studies reported data on this outcome.

30.3 Cardiac malformations

30.3.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 0.70, 95% CI 0.03 to 16.42, I² = NA), with no difference in the number of cardiac malformations in children exposed to LEV (N = 63) and children exposed to GBP (N = 14) (Analysis 30.3). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.08 to 0.11; I² = NA).

30.3.2 Routine health record data studies

No included studies reported data on this outcome.

30.4 Oro‐facial cleft/craniofacial malformations

30.4.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 0.70, 95% CI 0.03 to 16.42, I² = NA), with no difference in the number of cardiac malformations in children exposed to LEV (N = 63) and children exposed to GBP (N = 14) (Analysis 30.4). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.08 to 0.11; I² = NA).

30.4.2 Routine health record data studies

No included studies reported data on this outcome.

30.5 Skeletal/limb malformations

30.5.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported skeletal/limb malformations in children exposed to LEV (N = 63) or children exposed to GBP (N = 14) (Analysis 30.5).

30.5.2 Routine health record data studies

No included studies reported data on this outcome.

31 LEV versus LTG

31.1. All major malformations

31.1.1 Cohort studies

Pooled results from 10 cohort studies suggested no evidence of a difference in risk (RR 0.90, 95% CI 0.58 to 1.39; I² = 16%), with no difference in the number of major malformations in children exposed to LEV (N = 1223) and children exposed to LTG (N = 4389) (Analysis 31.1). The RD also suggested no difference in the level of risk (RD −0.00, 95% CI −0.01 to 0.01; I² = 10%).

The EURAP 2018 collaboration reported the prevalence of MCM was 2.8% (95% CI 1.7 to 4.5%) for children exposed to LEV and 2.9% (95% CI 2.3 to 3.7) for children exposed to LTG. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Levetiracetam dose and Lamotrigine dose sections).

31.1.2 Routine health record data studies

Pooled results from two routine health record studies suggested no evidence of a difference in risk (RR 0.79, 95% CI 0.37 to 1.69; I² = 0%), with no difference in the number of major malformations in children exposed to LEV (N = 248) and children exposed to LTG (N = 2068) (Analysis 31.1). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.03 to 0.01; I² = 0%).

31.2 Neural tube malformations

31.2.1 Cohort studies

Pooled results from nine cohort studies suggested no evidence of a difference in risk (RR 1.59, 95% CI 0.24 to 10.38; I² = 0%), with no difference in the number of neural tube malformations in children exposed to LEV (N = 1128) and children exposed to LTG (N = 4245) (Analysis 31.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.00 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to LEV was 0% (0/599) and 0.04% for children exposed to LTG (1/2514).

31.2.2 Routine health record data studies

No included studies reported data on this outcome.

31.3 Cardiac malformations

31.3.1 Cohort studies

Pooled results from nine cohort studies suggested no evidence of a difference in risk (RR 1.20, 95% CI 0.51 to 2.85; I² = 0%), with no difference in the number of cardiac malformations in children exposed to LEV (N = 1125) and children exposed to LTG (N = 4246) (Analysis 31.3). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.00 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to LEV was 0.83% (5/599) and 0.59% for children exposed to LTG (15/2514).

32.3.2 Routine health record data studies

No included studies reported data on this outcome.

31.4 Oro‐facial cleft/craniofacial malformations

31.4.1 Cohort studies

Pooled results from eight cohort studies suggested no evidence of a difference in risk (RR 0.63, 95% CI 0.15 to 2.68; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to LEV (N = 1019) and children exposed to LTG (N = 4196) (Analysis 31.4). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.01 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to LEV was 0.16% (1/599) and 0.43% for children exposed to LTG (3/2514).

31.4.2 Routine health record data studies

No included studies reported data on this outcome.

31.5 Skeletal/limb malformations

31.5.1 Cohort studies

Pooled results from nine cohort studies suggested no evidence of a difference in risk (RR 1.36, 95% CI 0.45 to 4.13; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to LEV (N = 1128) and children exposed to LTG (N = 4245) (Analysis 31.5). The RD also suggested no difference in the level of risk (RD −0.00, 95% CI −0.01 to 0.00; I² = 0%).

31.5.2 Routine health record data studies

No included studies reported data on this outcome.

32 LEV versus OXC

32.1 All major malformations

32.1.1 Cohort studies

Pooled results from eight cohort studies suggested no evidence of a difference in risk (RR 1.04, 95% CI 0.51 to 2.09; I² = 0%), with no difference in the number of major malformations in children exposed to LEV (N = 833) and children exposed to OXC (N = 333) (Analysis 32.1). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.02 to 0.03; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 2.8% (95% CI 1.7 to 4.5%) for children exposed to LEV and 3.0% (95% CI 1.4 to 5.4) for children exposed to OXC. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Levetiracetam dose and Oxcarbazepine dose sections).

32.1.2 Routine health record data studies

Pooled results from two routine health record studies suggested no evidence of a difference in risk (RR 1.17, 95% CI 0.45 to 3.06; I² = 0%), with no difference in the number of major malformations in children exposed to LEV (N = 248) and children exposed to OXC (N = 373) (Analysis 32.1). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.02 to 0.03; I² = 0%).

32.2 Neural tube malformations

32.2.1 Cohort studies

Pooled results from seven cohort studies suggested no evidence of a difference in risk (RR 1.22, 95% CI 0.05 to 29.74; I² = NA), with no difference in the number of neural tube malformations in children exposed to LEV (N = 738) and children exposed to OXC (N = 320) (Analysis 32.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to LEV was 0% (0/599) and 0% for children exposed to OXC (0/333).

32.2.2 Routine health record data studies

No included studies reported data on this outcome.

32.3 Cardiac malformations

32.3.1 Cohort studies

Pooled results from eight cohort studies suggested no evidence of a difference in risk (RR 0.93, 95% CI 0.31 to 2.76; I² = 0%), with no difference in the number of cardiac malformations in children exposed to LEV (N = 747) and children exposed to OXC (N = 323) (Analysis 32.3). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.01 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to LEV was 0.83% (5/599) and 1.2% for children exposed to OXC (4/333).

32.3.2 Routine health record data studies

No included studies reported data on this outcome.

32.4 Oro‐facial cleft/craniofacial malformations

32.4.1 Cohort studies

Pooled results from seven cohort studies suggested no evidence of a difference in risk (RR 0.25, 95% CI 0.03 to 2.12; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to LEV (N = 641) and children exposed to OXC (N = 252) (Analysis 32.4). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.02 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to LEV was 0.83% (5/599) and 0.30% for children exposed to OXC (1/333).

32.4.2 Routine health record data studies

No included studies reported data on this outcome.

32.5 Skeletal/limb malformations

32.5.1 Cohort studies

Pooled results from seven cohort studies suggested no evidence of a difference in risk (RR 0.80, 95% CI 0.20 to 3.29; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to LEV (N = 738) children exposed to OXC (N = 320) (Analysis 32.5). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.02 to 0.02; I² = 0%).

32.5.2 Routine health record data studies

No included studies reported data on this outcome.

33 LEV versus PB

33.1 All major malformations

33.1.1 Cohort studies

Results from five cohort studies suggested no evidence of a difference in risk (RR 0.54, 95% CI 0.29 to 1.02; I² = 0%), with children exposed to LEV (N = 726) experiencing comparable major malformations to children exposed to PB (N = 341) (Analysis 33.1). The RD also suggested no difference in the level of risk (RD −0.02, 95% CI −0.05 to 0.01; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 2.8% (95% CI 1.7 to 4.5%) for children exposed to LEV and 6.5% (95% CI 4.2 to 9.9) for children exposed to PB. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Levetiracetam dose and Phenobarbital dose sections).

33.1.2 Routine health record data studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 0.23, 95% CI 0.03 to 1.55; I² = NA), with children exposed to LEV (N = 118) experiencing comparable major malformation rates to children exposed to PB (N = 27). The RD also suggested no difference in the level of risk (RD −0.06, 95% CI −0.16 to 0.04; I² = NA).

33.2 Neural tube malformations

33.2.1 Cohort studies

Results from five cohort studies suggested no evidence of a difference in risk (RR 0.74, 95% CI 0.08 to 6.51; I² = 0%), with no difference in the number of neural tube malformations in children exposed to LEV (N = 650) and children exposed to PB (N = 344) (Analysis 33.2). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to LEV was 0% (0/599) and 0.68% for children exposed to PB (2/294).

33.2.2 Routine health record data studies

No included studies reported data on this outcome.

33.3 Cardiac malformations

33.3.1 Cohort studies

Pooled results from five cohort studies suggested no evidence of a difference in risk (RR 0.33, 95% CI 0.12 to 0.88; I² = 17%), with no difference in the number of cardiac malformations in children exposed to LEV (N = 650) and PB (N = 344) (Analysis 33.3). The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI ‐0.04 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to LEV was 0.83% (5/599) and 2.72% for children exposed to PB (8/294%).

33.3.2 Routine health record data studies

No included studies reported data on this outcome.

33.4 Oro‐facial cleft/craniofacial malformations

33.4.1 Cohort studies

Pooled results from four cohort studies suggested an increased risk with PB (RR 0.08, 95% CI 0.01 to 0.67; I² = 0%), with children exposed to LEV (N = 544) experiencing fewer oro‐facial cleft/craniofacial malformations than children exposed to PB (N = 207) (Analysis 33.4). The RD suggested no difference in the level of risk (RD ‐0.02, 95% CI ‐0.04 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to LEV was 0.16% (1/599) and 0.34% for children exposed to PB (1/294).

33.4.2 Routine health record data studies

No included studies reported data on this outcome.

33.5 Skeletal/limb malformations

33.5.1 Cohort studies

Pooled results from five cohort studies suggested no evidence of a difference in risk (RR 0.67, 95% CI 0.15 to 2.94; I² = 23%), with no difference in the number of skeletal/limb malformations in children exposed to LEV (N = 650) and children exposed to PB (N = 344) (Analysis 33.5). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.02 to 0.01; I² = 0%).

33.5.2 Routine health record data studies

No included studies reported data on this outcome.

34 LEV versus PHT

34.1 All major malformations

34.1.1 Cohort studies

Pooled results from five cohort studies suggested an increased risk with PHT (RR 0.58, 95% CI 0.34 to 0.97; I² = 58%), with children exposed to LEV (N = 1018) experiencing fewer major malformations than children exposed to PHT (N = 687) (Analysis 34.1). Due to high heterogeneity, we undertook a random‐effects analysis, which changed found no evidence of a difference in risk (RR 0.46, 95% CI 0.17 to 1.28; I² = 58%). The RD also suggested no difference in the level of risk (RD −0.02, 95% CI −0.04 to −0.00; I² = 52%). Due to high heterogeneity, we undertook a random‐effects analysis, which also found no difference (RD −0.02, 95% CI −0.05 to 0.02; I² = 52%).

The EURAP 2018 collaboration reported the prevalence of MCM was 2.8% (95% CI 1.7 to 4.5%) for children exposed to LEV and 6.4% (95% CI 2.8 to 12.2) for children exposed to PHT. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Levetiracetam dose and Phenytoin dose sections).

34.1.2 Routine health record data studies

No included studies reported data on this outcome.

34.2 Neural tube malformations

34.2.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 0.68, 95% CI 0.13 to 3.44; I² = 0%), with no difference in the number of neural tube malformations in children exposed to LEV (N = 913) and children exposed to PHT (N = 661) (Analysis 34.2). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to LEV was 0% (0/599) and 0.80% for children exposed to PHT (1/125).

34.2.2 Routine health record data studies

No included studies reported data on this outcome.

34.3 Cardiac malformations

34.3.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 0.43, 95% CI 0.16 to 1.13; I² = 0%), with no difference in the number of cardiac malformations in children exposed to LEV (N = 911) and children exposed to PHT (N = 611) (Analysis 34.3). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI ‐0.02 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to LEV was 0.83% (5/599) and 4.0% for children exposed to PHT (5/125).

34.3.2 Routine health record data studies

No included studies reported data on this outcome.

34.4 Oro‐facial cleft/craniofacial malformations

34.4.1 Cohort studies

Pooled results from three studies suggested no evidence of a difference in risk (RR 0.37, 95% CI 0.09 to 1.61; I² = 4%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to LEV (N = 807) and children exposed to PHT (N = 542) (Analysis 34.4). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to LEV was 0.16% (1/599) and 0% for children exposed to PHT (0/125).

34.4.2 Routine health record data studies

No included studies reported data on this outcome.

34.5 Skeletal/limb malformations

34.5.1 Cohort studies

Pooled results from four studies suggested no evidence of a difference in risk (RR 0.46, 95% CI 0.11 to 1.96; I² = 63%), with no difference in the number of skeletal/limb malformations in children exposed to LEV (N = 913) and children exposed to PHT (N = 661) (Analysis 34.5). Due to high heterogeneity, a random‐effects RR was calculated, which also found no difference (RR 0.54, 95% CI 0.02 to 11.85, I² =63%). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI ‐0.01 to 0.00; I² = 0%).

34.5.2 Routine health record data studies

No included studies reported data on this outcome.

35 LEV versus PRM

35.1 All major malformations

35.1.1. Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 0.24, 95% CI 0.02 to 3.37, I² = NA), with no difference in the number of major malformations in children exposed to LEV (N = 139) and children exposed to PRM (N = 2) (Analysis 35.1). The RD also suggested no difference in the level of risk (RD 0.04, 95% CI −0.39 to 0.46; I² = NA).

35.1.2. Routine health record studies

No included studies reported data on this outcome.

35.2 Neural tube malformations

35.2.1. Cohort studies

No included studies reported data on this outcome.

35.2.2. Routine health record studies

No included studies reported data on this outcome.

35.3 Cardiac malformations

35.3.1. Cohort studies

No included studies reported data on this outcome.

35.3.2. Routine health record studies

No included studies reported data on this outcome.

35.4 Oro‐facial cleft/craniofacial malformations

35.4.1. Cohort studies

No included studies reported data on this outcome.

35.4.2. Routine health record studies

No included studies reported data on this outcome.

35.5 Skeletal/limb malformations

35.5.1. Cohort studies

No included studies reported data on this outcome.

35.5.2. Routine health record studies

No included studies reported data on this outcome.

36 LEV versus TPM

36.1 All major malformations

36.1.1 Cohort studies

Pooled results from eight cohort studies suggested no evidence of a difference in risk (RR 0.57, 95% CI 0.32 to 1.04; I² = 0%), with children exposed to LEV (N = 1124) experiencing comparable major malformations to children exposed to TPM (N = 505) (Analysis 36.1). The RD also suggested no difference in the level of risk (RD −0.02, 95% CI −0.04 to 0.00; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 2.8% (95% CI 1.7 to 4.5%) for children exposed to LEV and 3.9% (95% CI 1.5 to 8.4) for children exposed to TPM. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Levetiracetam dose and Topiramate dose sections).

36.1.2 Routine health record studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 0.41, 95% CI 0.06 to 2.81; I² = NA), with children exposed to LEV (N = 118) experiencing comparable major malformation rate to children exposed to TPM (N = 48) (Analysis 36.1). The RD also suggested no difference in the level of risk (RD −0.02, 95% CI −0.09 to 0.04; I² = NA).

36.2 Neural tube malformations

36.2.1 Cohort studies

Pooled results from seven cohort studies suggested no evidence of a difference in risk (RR 2.39, 95% CI 0.10 to 58.61; I² = NA), with no difference in the number of neural tube malformations in children exposed to LEV (N = 1030) and children exposed to TPM (N = 496) (Analysis 36.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to LEV was 0% (0/599) and 0% for children exposed to TPM (0/152).

36.2.2 Routine health record studies

No included studies reported data on this outcome.

36.3 Cardiac malformations

36.3.1 Cohort studies

Pooled results from eight studies suggested no evidence of a difference in risk (RR 0.72, 95% CI 0.21 to 2.53; I² = 0%), with no difference in the number of cardiac malformations in children exposed to LEV (N = 1039) and children exposed to TPM (N = 497) (Analysis 36.2). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to LEV was 0.83% (5/599) and 1.97% for children exposed to TPM (3/152).

36.3.2 Routine health record studies

No included studies reported data on this outcome.

36.4 Oro‐facial cleft/craniofacial malformations

36.4.1 Cohort studies

Pooled results from seven cohort studies suggested an increased risk with TPM (RR 0.19, 95% CI 0.05 to 0.70; I² = 48%), with children exposed to LEV (N = 933) experiencing fewer oro‐facial/craniofacial malformations than children exposed to TPM (N = 488) (Analysis 36.4). However, the RD suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.03 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to LEV was 0.16% (1/599) and 0% for children exposed to TPM (0/125).

36.4.2 Routine health record studies

No included studies reported data on this outcome.

36.5 Skeletal/limb malformations

36.5.1 Cohort studies

Pooled results from seven cohort studies suggested an increased risk with TPM (RR 0.12, 95% CI 0.02 to 0.98; I² = 0%), with children exposed to LEV (N = 1030) experiencing fewer skeletal/limb malformations than children exposed to TPM (N = 496) (Analysis 36.5). However, the RD suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.02 to 0.00; I² = 0%).

36.5.2 Routine health record studies

No included studies reported data on this outcome.

37 LEV versus ZNS

37.1 All major malformations

37.1.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 0.66, 95% CI 0.25 to 1.71; I² = 79%), with no difference in the number of major malformations in children exposed to LEV (N = 865) and children exposed to ZNS (N = 130) (Analysis 37.1). Due to high heterogeneity, a random‐effects RR was calculated, which also found no difference (RD 0.48, 95% CI 0.03 to 7.24, I² = 79%). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI ‐0.04 to 0.03; I² = 76%). Due to heterogeneity, a random‐effects RD was calculated, which maintained similar findings (RD ‐0.03, 95% CI ‐0.16 to 0.10, I² = 76%).

37.1.2 Routine health record studies

No included studies reported data on this outcome.

37.2 Neural tube malformations

37.2.1 Cohort studies

Pooled results from three included studies suggested evidence of a difference in risk (RR 0.03, 95% CI 0.00 to 0.71, I² = NA), with children exposed to ZNS (N = 40) experiencing more neural tube malformations than children exposed to LEV (N = 415) (Analysis 37.2). However, the RD suggested no difference in the level of risk (RD ‐0.03, 95% CI −0.10 to 0.05; I² = 0%).

37.2.2 Routine health record studies

No included studies reported data on this outcome.

37.3 Cardiac malformations

37.3.1 Cohort studies

Pooled results from three included studies suggested no evidence of a difference in risk (RR 0.98, 95% CI 0.05 to 17.99, I² = NA), with no difference in cardiac malformations between children exposed to LEV (N = 415) and children exposed to ZNS (N = 40) (Analysis 37.3). The RD suggested no difference in the level of risk (RD 0.01, 95% CI −0.05 to 0.07; I² = 0%).

37.3.2 Routine health record studies

No included studies reported data on this outcome.

37.4 Oro‐facial cleft/craniofacial malformations

37.4.1 Cohort studies

We were unable to estimate a RR from three studies due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to LEV (N = 415) and ZNS (N = 40) (Analysis 37.4).

37.4.2 Routine health record studies

No included studies reported data on this outcome.

37.5 Skeletal/limb malformations

37.5.1 Cohort studies

We were unable to estimate a RR from three studies due to there being no reported skeletal/limb malformations in children exposed to LEV (N = 415) and ZNS (N = 40) (Analysis 37.5).

37.5.2 Routine health record studies

No included studies reported data on this outcome.

38 LTG versus CZP

38.1 All major malformations

38.1.1 Cohort studies

Pooled results from three cohort studies suggested no evidence of a difference in risk (RR 0.92, 95% CI 0.29 to 2.91; I² = 0%), with no difference in the number of major malformations in children exposed to LTG (N = 2018) and children exposed to CZP (N = 94) (Analysis 38.1). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI ‐0.03 to 0.04; I² = 33%).

38.1.2 Routine health record studies

Pooled results from two routine health record studies suggested no evidence of a difference in risk (RR 1.54, 95% CI 0.53 to 4.54; I² = 0%), with no difference in the number of major malformations in children exposed to LTG (N = 923) and children exposed to CZP (N = 161) (Analysis 38.1). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.01 to 0.04; I² = 0%).

38.2 Neural tube malformations

38.2.1 Cohort studies

No included studies reported data on this outcome.

38.2.2 Routine health record studies

No included studies reported data on this outcome.

38.3 Cardiac malformations

38.3.1 Cohort studies

No included studies reported data on this outcome.

38.3.2 Routine health record studies

No included studies reported data on this outcome.

38.4 Oro‐facial cleft/craniofacial malformations

38.4.1 Cohort studies

No included studies reported data on this outcome.

38.4.2 Routine health record studies

No included studies reported data on this outcome.

38.5 Skeletal/limb malformations

38.5.1 Cohort studies

No included studies reported data on this outcome.

38.5.2 Routine health record studies

No included studies reported data on this outcome.

39 LTG versus LAC

39.1 All major malformations

39.1.1 Cohort studies

We were unable to estimate the RR from one study due to there being no major malformations observed in children exposed to LTG (N = 19) or LAC (N = 1) (Analysis 39.1).

39.1.2 Routine health record studies

No included studies reported data on this outcome.

39.2 Neural tube malformations

39.2.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported neural tube malformations in children exposed to LTG (N = 19) or LAC (N = 1) (Analysis 39.2).

39.2.2 Routine health record studies

No included studies reported data on this outcome.

39.3 Cardiac malformations

39.3.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported cardiac malformations in children exposed to LTG (N = 19) or LAC (N = 1) (Analysis 39.3).

39.3.2 Routine health record studies

No included studies reported data on this outcome.

39.4 Oro‐facial cleft/craniofacial malformations

39.4.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to LTG (N = 19) or LAC (N = 1) (Analysis 39.3).

39.4.2 Routine health record studies

No included studies reported data on this outcome.

39.5 Skeletal/limb malformations

39.5.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported skeletal/limb malformations in children exposed to LTG (N = 19) or LAC (N = 1) (Analysis 39.3).

39.5.2 Routine health record studies

No included studies reported data on this outcome.

40 LTG versus OXC

40.1 All major malformations

40.1.1 Cohort studies

Pooled results from eight cohort studies suggested no evidence of a difference in risk(RR 0.73, 95% CI 0.33 to 1.62; I² = 0%), with no difference in the number of major malformations in children exposed to LTG (N = 2208) and children exposed to OXC (N = 333) (Analysis 40.1). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.03 to 0.02; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 2.9% (95% CI 2.3 to 3.7%) for children exposed to LTG and 3.0% (95% CI 1.4 to 5.4) for children exposed to OXC. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Lamotrigine dose and Oxcarbazepine dose sections).

40.1.2 Routine health record studies

Pooled results from three routine health record studies suggested no evidence of a difference in risk (RR 1.24, 95% CI 0.67 to 2.30; I² = 0%), with no difference in the number of major malformations in children exposed to LTG (N = 2158) and children exposed to OXC (N = 377) (Analysis 40.1). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.01 to 0.03; I² = 0%).

40.2 Neural tube malformations

40.2.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.59, 95% CI 0.03 to 12.15; I² = NA), with no difference in the number of neural tube malformations in children exposed to LTG (N = 2027) and children exposed to OXC (N = 319) (Analysis 40.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to LTG was 0.3% (1/2514) and 0% for children exposed to OXC (0/333).

40.2.2 Routine health record studies

No included studies reported data on this outcome.

40.3 Cardiac malformations

40.3.1 Cohort studies

Pooled results from eight cohort studies suggested no evidence of a difference in risk (RR 0.59, 95% CI 0.15 to 2.30; I² = 0%), with no difference in the number of cardiac malformations in children exposed to LTG (N = 2084) and children exposed to OXC (N = 323) (Analysis 40.3). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.02 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to LTG was 5.9% (15/2514) and 1.2% for children exposed to OXC (4/333).

40.3.2 Routine health record studies

No included studies reported data on this outcome.

40.4 Oro‐facial cleft/craniofacial malformations

40.4.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.64, 95% CI 0.12 to 3.46; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to LTG (N = 1997) and children exposed to OXC (N = 251) (Analysis 40.4). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.02 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to LTG was 0.11% (3/2514) and 0.30% for children exposed to OXC (1/333).

40.4.2 Routine health record studies

No included studies reported data on this outcome.

40.5 Skeletal/limb malformations

40.5.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.29, 95% CI 0.06 to 1.56; I² = NA), with no difference in the number of skeletal/limb malformations in children exposed to LTG (N = 2027) and children exposed to OXC (N = 319) (Analysis 40.5). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.02 to 0.01; I² = 0%).

40.5.2 Routine health record studies

No included studies reported data on this outcome.

41 LTG versus PB

41.1 All major malformations

41.1.1 Cohort studies

Pooled results from seven cohort studies suggested an increased risk with PB (RR 0.32, 95% CI 0.17 to 0.59; I² = 0%), with children exposed to LTG (N = 2156) experiencing fewer major malformations than children exposed to PB (N = 421) (Analysis 41.1). The RD also suggested a higher risk for PB (RD −0.04, 95% CI −0.07 to −0.01; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 2.9% (95% CI 2.3 to 3.7%) for children exposed to LTG and 6.5% (95% CI 4.2 to 9.9) for children exposed to PB. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Lamotrigine dose and Phenobarbital dose sections).

41.1.2 Routine health record studies

Pooled results from two routine health record studies suggested no evidence of a difference in risk (RR 0.41, 95% CI 0.13 to 1.28; I² = 0%), with children exposed to LTG (N = 923) experiencing comparable major malformations to children exposed to PB (N = 34) (Analysis 41.1). The RD also suggested no difference in the level of risk (RD −0.05, 95% CI −0.15 to 0.04; I² = 0%).

41.2 Neural tube malformations

41.2.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.76, 95% CI 0.09 to 6.88; I² = 0%), with no difference in the number of neural tube malformations in children exposed to LTG (N = 2009) and children exposed to PB (N = 412) (Analysis 41.2). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to LTG was 0.3% (1/2514) and 0.68% for children exposed to PB (2/294).

41.2.2 Routine health record studies

No included studies reported data on this outcome.

41.3 Cardiac malformations

41.3.1 Cohort studies

Pooled results from five cohort studies suggested an increased risk with PB (RR 0.21, 95% CI 0.08 to 0.56; I² = 0%), with children exposed to LTG (N = 1990) experiencing fewer cardiac malformations than children exposed to PB (N = 411) (Analysis 41.3). The RD also suggested a higher risk for PB (RD ‐0.02, 95% CI −0.04 to ‐0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to LTG was 5.9% (15/2514) and 2.72% for children exposed to PB (8/294).

41.3.2 Routine health record studies

No included studies reported data on this outcome.

41.4 Oro‐facial cleft/craniofacial malformations

41.4.1 Cohort studies

Pooled results from four cohort studies suggested an increased risk with PB (RR 0.22, 95% CI 0.07 to 0.68; I² = 0%), with fewer oro‐facial cleft/craniofacial malformations in children exposed to LTG (N = 1940) compared to PB (N = 274) (Analysis 41.4). However, the RD suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.03 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to LTG was 0.11% (3/2514) and 0.34% for children exposed to PB (1/294).

41.4.2 Routine health record studies

No included studies reported data on this outcome.

41.5 Skeletal/limb malformations

41.5.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.38, 95% CI 0.06 to 2.58; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to LTG (N = 2009) and children exposed to PB (N = 413) (Analysis 41.5). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.02 to 0.01; I² = 0%).

41.5.2 Routine health record studies

No included studies reported data on this outcome.

42 LTG versus PHT

42.1 All major malformations

42.1.1 Cohort studies

Pooled results from six cohort studies suggested an increased risk with PHT (RR 0.55, 95% CI 0.35 to 0.87; I² = 24%), with children exposed to LTG (N = 4251) experiencing fewer major malformations than children exposed to PHT (N = 742) (Analysis 42.1). The RD also suggested a higher risk for PHT (RD −0.02, 95% CI −0.03 to −0.00; I² = 31%).

The EURAP 2018 collaboration reported the prevalence of MCM was 2.9% (95% CI 2.3 to 3.7%) for children exposed to LTG and 6.4% (95% CI 2.8 to 12.2) for children exposed to PHT. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Lamotrigine dose and Phenytoin dose sections).

42.1.2 Routine health record studies

One routine health record study suggested no evidence of a difference in risk (RR 0.65, 95% CI 0.20 to 2.16; I² = NA%), with children exposed to LTG (N = 90) experiencing comparable major malformations to children exposed to PHT (N = 103) (Analysis 42.1). The RD also suggested no difference in the level of risk (RD −0.02, 95% CI −0.09 to 0.04; I² = NA).

42.2 Neural tube malformations

43.2.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.40, 95% CI 0.11 to 1.51; I² = 0%), with no difference in the number of neural tube malformations in children exposed to LTG (N = 4127) and children exposed to PHT (N = 718) (Analysis 42.2). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to LTG was 0.3% (1/2514) and 0.80% for children exposed to PHT (1/125).

42.2.2 Routine health record studies

No included studies reported data on this outcome.

42.3 Cardiac malformations

42.3.1 Cohort studies

Pooled results from six cohort studies suggested an increased risk with PHT (RR 0.41, 95% CI 0.17 to 0.98; I² = 0%), with children exposed to LTG (N = 4127) experiencing fewer cardiac malformations than children exposed to PHT (N = 718) (Analysis 42.3). However, the RD suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.02 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to LTG was 5.9% (15/2514) and 4.0% for children exposed to PHT (5/125).

42.3.2 Routine health record studies

No included studies reported data on this outcome.

42.4 Oro‐facial cleft/craniofacial malformations

42.4.1 Cohort studies

Pooled results from five cohort studies suggested no evidence of a difference in risk (RR 0.73, 95% CI 0.23 to 2.28; I² = 45%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to LTG (N = 4077) and children exposed to PHT (N = 599) (Analysis 42.4). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to LTG was 0.11% (3/2514) and 0% for children exposed to PHT (0/125).

42.4.2 Routine health record studies

No included studies reported data on this outcome.

42.5 Skeletal/limb malformations

42.5.1 Cohort studies

Pooled results from six cohort studies suggested an increased risk with PHT (RR 0.28, 95% CI 0.09 to 0.86; I² = 0%), with children exposed to LTG (N = 4127) experiencing fewer skeletal/limb malformations than children exposed to PHT (N = 718) (Analysis 42.5). However, the RD suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.01 to 0.00; I² = 0%).

42.5.2 Routine health record studies

No included studies reported data on this outcome.

43 LTG versus PRM

43.1 All major malformations

43.1.1 Cohort studies

One cohort study suggested no evidence of a difference in risk (RR 0.30, 95% CI 0.02 to 3.93; I² = NA), with children exposed to LTG (N = 406) experiencing comparable major malformations to children exposed to PRM (N = 2) (Analysis 43.1). The RD also suggested no difference in the level of risk (RD 0.05, 95% CI −0.37 to 0.47; I² = NA).

43.1.2 Routine health record studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 0.40, 95% CI 0.03 to 6.16; I² = NA), with children exposed to LTG (N = 90) experiencing comparable major malformations to children exposed to PRM (N = 2) (Analysis 43.1). The RD also suggested no difference in the level of risk (RD 0.04, 95% CI −0.28 to 0.37; I² = NA).

43.2 Neural tube malformations

43.2.1 Cohort studies

No included studies reported data on this outcome.

43.2.2 Routine health record studies

No included studies reported data on this outcome.

43.3 Cardiac malformations

43.3.1 Cohort studies

No included studies reported data on this outcome.

43.3.2 Routine health record studies

No included studies reported data on this outcome.

43.4 Oro‐facial cleft/craniofacial malformations

43.4.1 Cohort studies

No included studies reported data on this outcome.

43.4.2 Routine health record studies

No included studies reported data on this outcome.

43.5 Skeletal/limb malformations

43.5.1 Cohort studies

No included studies reported data on this outcome.

43.5.2 Routine health record studies

No included studies reported data on this outcome.

44 LTG versus TPM

44.1 All major malformations

44.1.1 Cohort studies

Pooled results from eight cohort studies suggested an increased risk with TPM (RR 0.59, 95% CI 0.36 to 0.96; I² = 0%), with children exposed to LTG (N = 4275) experiencing fewer major malformations than children exposed to TPM (N = 505) (Analysis 44.1). However, the RD suggested no difference in the level of risk (RD −0.02, 95% CI −0.03 to 0.00; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 2.9% (95% CI 2.3 to 3.7%) for children exposed to LTG and 3.9% (95% CI 1.5 to 8.4) for children exposed to TPM. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Lamotrigine dose and Topiramate dose sections).

44.1.2 Routine health record studies

Pooled results from two routine health record studies suggested no evidence of a difference in risk (RR 0.68, 95% CI 0.20 to 2.37; I² = 0%), with children exposed to LTG (N = 923) experiencing comparable major malformation rates to children exposed to TPM (N = 49) (Analysis 44.1). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.07 to 0.06; I² = 0%).

44.2 Neural tube malformations

44.2.1 Cohort studies

Pooled results from seven cohort studies suggested no evidence of a difference in risk (RR 0.62, 95% CI 0.08 to 4.94; I² = 0%), with no difference in the number of neural tube malformations in children exposed to LTG (N = 4131) and children exposed to TPM (N = 496) (Analysis 44.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to LTG was 0.3% (1/2514) and 0% for children exposed to TPM (0/152).

44.2.2 Routine health record studies

No included studies reported data on this outcome.

44.3 Cardiac malformations

44.3.1 Cohort studies

Pooled results from eight cohort studies suggested no evidence of a difference in risk (RR 0.58, 95% CI 0.19 to 1.81; I² = 0%), with no difference in the number of cardiac malformations in children exposed to LTG (N = 4151) and children exposed to TPM (N = 497) (Analysis 44.3). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.01 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to LTG was 5.9% (15/2514) and 1.97% for children exposed to TPM (3/152).

44.3.2 Routine health record studies

No included studies reported data on this outcome.

44.4 Oro‐facial cleft/craniofacial malformations

44.4.1 Cohort studies

Pooled results from seven cohort studies suggested evidence of a difference in risk (RR 0.31, 95% CI 0.13 to 0.74; I² = 68%), with children exposed to LTG (N = 4101) experiencing less oro‐facial cleft/craniofacial malformations than children exposed to TPM (N = 488) (Analysis 44.4). Due to high heterogeneity, we undertook a random‐effects analysis, which found no difference (RR 0.22, 95% CI 0.03 to 1.48; I² = 68%). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.02 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to LTG was 0.11% (3/2514) and 0% for children exposed to TPM (0/152).

44.4.2 Routine health record studies

No included studies reported data on this outcome that could be included in the meta‐analysis. One study by Hernandez‐Diaz and colleagues using US Medicaid Registers could not be included in the meta‐analysis due to a lack of reporting of specific numbers of oral clefts. In this study, children born to women taking TPM had higher rates of oral clefts (N = 2425, 4.1 per 1000 live births) than the children born to women taking LTG (N = 2796, 1.5 per 1000 live births), but this was not reported to be statistically significant (RR 2.30, 95% CI 0.69 to 7.64).

44.5 Skeletal/limb malformations

44.5.1 Cohort studies

Pooled results from seven cohort studies suggested no evidence of a difference in risk (RR 0.17, 95% CI 0.06 to 0.52; I² = 0%), with children exposed to LTG (N = 4131) experiencing fewer skeletal/limb malformations than children exposed to TPM (N = 496) (Analysis 44.5). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.02 to 0.00; I² = 0%).

44.5.2 Routine health record studies

No included studies reported data on this outcome.

45 LTG versus ZNS

45.1 All major malformations

45.1.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 0.66, 95% CI 0.26 to 1.65; I² = 66%), with no difference in the number of major malformations in children exposed to LTG (N = 3792) and children exposed to ZNS (N = 130) (Analysis 45.1). Due to high heterogeneity, a random‐effects RR was calculated which also found no difference (RD 0.57, 95% CI 0.09 to 3.81, I² = 66%). The RD also suggested no difference in the level of riskThe RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.04 to 0.02; I² = 77%). Due to heterogeneity, a random‐effects RD was calculated, which upheld similar findings (RD ‐0.03, 95% CI ‐0.16 to 0.11, I² = 77%).

45.1.2 Routine health record studies

No included studies reported data on this outcome.

45.2 Neural tube malformations

45.2.1 Cohort studies

Pooled data from three included studies suggested evidence of a difference in risk (RR 0.02, 95% CI 0.00 to 0.26, I² = NA), with children exposed to ZNS (N = 40) experiencing more neural tube malformations than children exposed to LTG (N = 2230) (Analysis 45.2). However, the RD suggested no difference in the level of risk (RD ‐0.03, 95% CI −0.09 to 0.04; I² = 0%).

45.2.2 Routine health record studies

No included studies reported data on this outcome.

45.3 Cardiac malformations

45.3.1 Cohort studies

Pooled data from two included studies suggested no evidence of a difference in risk (RR 0.30, 95% CI 0.04 to 2.52, I² = 0%), with no difference in the number of cardiac malformations in children exposed to LTG (N = 2211) and children exposed to ZNS (N = 39) (Analysis 45.2). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.04 to 0.05; I² = 0%).

45.3.2 Routine health record studies

No included studies reported data on this outcome.

45.4 Oro‐facial cleft/craniofacial malformations

45.4.1 Cohort studies

Pooled data from two included studies suggested no evidence of a difference in risk (RR 0.06, 95% CI 0.00 to 1.31, I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to LTG (N = 2211) and children exposed to ZNS (N = 39) (Analysis 45.4). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.05 to 0.05; I² = 0%).

45.4.2 Routine health record studies

No included studies reported data on this outcome.

45.5 Skeletal/limb malformations

45.5.1 Cohort studies

Pooled data from three included studies suggested no evidence of a difference in risk (RR 0.22, 95% CI 0.03 to 1.93, I² = 0%, with no difference in the number of skeletal/limb malformations in children exposed to LTG (N = 2230) and children exposed to ZNS (N = 40) (Analysis 45.5). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.05 to 0.06; I² = 0%).

45.5.2 Routine health record studies

No included studies reported data on this outcome.

46 PHT versus GBP

46.1 All major malformations

46.1.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 2.15, 95% CI 0.69 to 6.73; I² = 0%), with no difference in the number of major malformations in children exposed to PHT (N = 567) and children exposed to GBP (N = 192) (Analysis 46.1). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI ‐0.00 to 0.04; I² = 0%).

46.1.2 Routine health record studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 2.74, 95% CI 0.16 to 46.00; I² = 0%), with children exposed to PHT (N = 103) experiencing comparable major malformations to children exposed to GBP (N = 18) (Analysis 46.1). The RD also suggested no difference in the level of risk (RD 0.07, 95% CI ‐0.02 to 0.16; I² = NA).

46.2 Neural tube malformations

46.2.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 46.2). However, available data showed there were 1/45 cases of neural tube malformations in children exposed to PHT and 0/16 cases in children exposed to GBP, based on data from two studies (Australian Epilepsy and Pregnancy Register; Miskov 2016).

46.2.2 Routine health record studies

No included studies reported data on this outcome.

46.3 Cardiac malformations

46.3.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 46.3). However, available data showed there were 1/45 cases of cardiac malformations in children exposed to PHT and 1/16 cases in children exposed to GBP, based on data from two studies (Australian Epilepsy and Pregnancy Register; Miskov 2016).

46.3.2 Routine health record studies

No included studies reported data on this outcome.

46.4 Oro‐facial cleft/craniofacial malformations

46.4.1 Cohort studies

We were unable to estimate a RR from two studies due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to PHT (N = 45) or GBP (N = 16) (Analysis 46.4).

46.4.2 Routine health record studies

No included studies reported data on this outcome.

46.5 Skeletal/limb malformations

46.5.1 Cohort studies

We were unable to estimate a RR due to there being no reported cases of skeletal/limb malformations in children exposed to PHT (N = 45) or GBP (N = 16) (Analysis 46.5).

46.5.2 Routine health record studies

No included studies reported data on this outcome.

47 PHT versus OXC

47.1 All major malformations

47.1.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.94, 95% CI 0.48 to 1.85; I² = 0%), with no difference in the number of major malformations in children exposed to PHT (N = 706) and children exposed to OXC (N = 283) (Analysis 47.1). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.03 to 0.03; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 6.4% (95% CI 2.8 to 12.2) for children exposed to PHT and 3.0% (95% CI 1.4 to 5.4) for children exposed to OXC. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Phenytoin dose and Oxcarbazepine dose sections).

47.1.2 Routine health record studies

Results from one routine health record study suggested an increased risk with PHT (RR 0.72, 95% CI 0.05 to 0.93; I² = NA), with children exposed to PHT (N = 103) experiencing more major malformations than children exposed to OXC (N = 4) (Analysis 47.1). However, the RD suggested no difference in the level of risk (RD 0.07, 95% CI −0.20 to 0.34; I² = NA).

47.2 Neural tube malformations

47.2.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 1.16, 95% CI 0.13 to 10.29; I² = 0%), with no difference in the number of neural tube malformations in children exposed to PHT (N = 703) and children exposed to OXC (N = 271) (Analysis 47.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.01 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to PHT was 0.80% (1/125) and 0% for children exposed to OXC (0/333).

47.2.2 Routine health record studies

No included studies reported data on this outcome.

47.3 Cardiac malformations

47.3.1 Cohort studies

Pooled results from five cohort studies suggested no evidence of a difference in risk (RR 1.33, 95% CI 0.43 to 14.17; I² = 0%), with no difference in the number of cardiac malformations in children exposed to PHT (N = 704) and children exposed to OXC (N = 272) (Analysis 47.3). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.01 to 0.03; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to PHT was 4.0% (5/125) and 1.2% for children exposed to OXC (4/333).

47.3.2 Routine health record studies

No included studies reported data on this outcome.

47.4 Oro‐facial cleft/craniofacial malformations

47.4.1 Cohort studies

Pooled results from three cohort studies suggested no evidence of a difference in risk (RR 0.62, 95% CI 0.10 to 4.05; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to PHT (N = 584) and children exposed to OXC (N = 200) (Analysis 47.4). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.02 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to PHT was 0% (0/125) and 0.30% for children exposed to OXC (1/333).

47.4.2 Routine health record studies

No included studies reported data on this outcome.

47.5 Skeletal/limb malformations

47.5.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 1.20, 95% CI 0.23 to 6.35; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to PHT (N = 703) and children exposed to OXC (N = 271) (Analysis 47.5). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.01 to 0.02; I² = 0%).

47.5.2 Routine health record studies

No included studies reported data on this outcome.

48 PHT versus PB

48.1 All major malformations

48.1.1 Cohort studies

Pooled results from 20 cohort studies suggested no evidence of a difference in risk (RR 0.84, 95% CI 0.57 to 1.23; I² = 0%), with no difference in the number of major malformations in children exposed to PHT (N = 1095) and children exposed to PB (N = 634) (Analysis 48.1). The RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.03 to 0.02; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 6.4% (95% CI 2.8 to 12.2) for children exposed to PHT and 6.5% (95% CI 4.2 to 9.9) for children exposed to PHT. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Phenytoin dose and Phenobarbital dose sections). Samren 1997 reported nine cases of major malformation in 141 (6%) PHT cases and five cases in 48 (10%) PB‐exposed children.

48.1.2 Routine health record studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 0.48, 95% CI 0.07 to 3.35; I² = NA), with children exposed to PHT (N = 103) experiencing comparable major malformations to children exposed to PB (N = 7) (Analysis 48.1). The RD also suggested no difference in the level of risk (RD −0.07, 95% CI −0.34 to 0.19; I² = NA).

48.2 Neural tube malformations

48.2.1 Cohort studies

Pooled results from 11 studies suggested no evidence of a difference in risk (RR 0.79, 95% CI 0.10 to 5.94; I² = 0%), with no difference in the number of neural tube malformations in children exposed to PHT (N = 707) and children exposed to PB (N = 475) (Analysis 48.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to PHT was 0.80% (1/125) and 0.68% for children exposed to PB (2/294).

48.2.2 Routine health record studies

No included studies reported data on this outcome.

48.3 Cardiac malformations

48.3.1 Cohort studies

Pooled results from 11 cohort studies suggested no evidence of a difference in risk (RR 0.56, 95% CI 0.29 to 1.07; I² = 0%), with children exposed to PHT (N = 707) experiencing no more cardiac malformations than children exposed to PB (N = 475) (Analysis 48.3). The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI −0.04 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to PHT was 4.0% (5/125) and 2.72% for children exposed to PB (8/294).

48.3.2 Routine health record studies

No included studies reported data on this outcome.

48.4 Oro‐facial cleft/craniofacial malformations

48.4.1 Cohort studies

Pooled results from 11 cohort studies suggested no evidence of a difference in risk (RR 0.25, 95% CI 0.07 to 0.82; I² = 0%), with children exposed to PHT (N = 593) experiencing fewer oro‐facial cleft/craniofacial malformations than children exposed to PB (N = 347) (Analysis 48.4). The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI −0.04 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to PHT was 0% (0/125) and 0.34% for children exposed to PB (1/294).

48.4.2 Routine health record studies

No included studies reported data on this outcome.

48.5 Skeletal/limb malformations

48.5.1 Cohort studies

Pooled results from 11 cohort studies suggested no evidence of a difference in risk (RR 1.31, 95% CI 0.39 to 4.39; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to PHT (N = 707) and children exposed to PB (N = 475) (Analysis 48.5). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.01 to 0.02; I² = 0%).

48.5.2 Routine health record studies

No included studies reported data on this outcome.

49 PHT versus PRM

49.1 All major malformations

49.1.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.78, 95% CI 0.39 to 1.56; I² = 19%), with no difference in the number of major malformations in children exposed to PHT (N = 360) and children exposed to PRM (N = 103) (Analysis 49.1). The RD also suggested no difference in the level of risk (RD −0.02, 95% CI −0.09 to 0.06; I² = 0%).

49.1.2 Routine health record studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 0.58, 95% CI 0.04 to 8.44; I² = NA), with children exposed to PHT (N = 103) experiencing no more major malformations than children exposed to PRM (N = 3) (Analysis 49.1). The RD also suggested no difference in the level of risk (RD 0.07, 95% CI −0.26 to 0.40; I² = NA).

49.2 Neural tube malformations

49.2.1 Cohort studies

We were unable to estimate a RR from two studies due to there being no cases of neural tube malformations in children exposed to PHT (N = 36) or PRM (N = 39) (Analysis 49.2).

49.2.2 Routine health record studies

No included studies reported data on this outcome.

49.3 Cardiac malformations

49.3.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 49.3). However, available data showed there were 0/36 cases of cardiac malformations in children exposed to PHT and 1/39 cases in children exposed to PRM, based on data from two studies (Milan Study 1999; Pardi 1982).

49.3.2 Routine health record studies

No included studies reported data on this outcome.

49.4 Oro‐facial cleft/craniofacial malformations

49.4.1 Cohort studies

We were unable to estimate a RR due to there being no reported cases of oro‐facial cleft/craniofacial malformations in children exposed to PHT (N = 36) or PRM (N = 39) (Analysis 49.2).

49.4.2 Routine health record studies

No included studies reported data on this outcome.

49.5 Skeletal/limb malformations

49.5.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 49.5). However, available data showed there were 1/36 cases of skeletal/limb malformations in children exposed to PHT and 0/39 cases in children exposed to PRM, based on data from two studies (Milan Study 1999; Pardi 1982).

49.5.2 Routine health record studies

No included studies reported data on this outcome.

50 PHT versus TPM

50.1 All major malformations

50.1.1 Cohort studies

Pooled results from four cohort studies showed no evidence of a difference in risk (RR 0.88, 95% CI 0.48 to 1.61; I² = 0%), with no difference in the number of major malformations in children exposed to PHT (N = 685) and children exposed to TPM (N = 491) (Analysis 50.1). The RD also suggested no difference in the level of risk (RD −0.00, 95% CI −0.03 to 0.02; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 2.9% (95% CI 2.3 to 3.7%) for children exposed to PHT and 3.9% (95% CI 1.5 to 8.4) for children exposed to TPM. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Phenytoin dose and Topiramate dose sections).

50.1.2 Routine health record studies

Results from one study suggested no evidence of a difference in risk (RR 0.29, 95% CI 0.02 to 3.51, I² = NA), with no difference in the number of major malformations in children exposed to PHT (N = 103) and children exposed to PRM (N = 1) (Analysis 50.1). The RD also suggested no difference in the level of risk (RD 0.07, 95% CI ‐0.53 to 0.67; I² = NA).

50.2 Neural tube malformations

50.2.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 1.23, 95% CI 0.17 to 8.87; I² = 24%), with no difference in the number of neural tube malformations in children exposed to PHT (N = 661) and children exposed to TPM (N = 483) (Analysis 50.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.01 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to PHT was 0.80% (1/125) and 0% for children exposed to TPM (0/152).

50.2.2 Routine health record studies

No included studies reported data on this outcome.

50.3 Cardiac malformations

50.3.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 2.46, 95% CI 0.65 to 9.36; I² = 0%), with no difference in the number of cardiac malformations in children exposed to PHT (N = 661) and children exposed to TPM (N = 483) (Analysis 50.3). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.00 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to PHT was 4.0% (5/125) and 1.97% for children exposed to TPM (3/152).

50.3.2 Routine health record studies

No included studies reported data on this outcome.

50.4 Oro‐facial cleft/craniofacial malformations

50.4.1 Cohort studies

Pooled results from three cohort studies suggested no evidence of a difference in risk (RR 0.37, 95% CI 0.10 to 1.42; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to PHT (N = 542) and children exposed to TPM (N = 474) (Analysis 50.4). The RD also suggested no difference in the level of risk(RD ‐0.01, 95% CI −0.02 to 0.00; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to PHT was 0% (0/125) and 0% for children exposed to TPM (0/152).

50.4.2 Routine health record studies

No included studies reported data on this outcome.

50.5 Skeletal/limb malformations

50.5.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 0.63, 95% CI 0.19 to 2.09; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to PHT (N = 661) and children exposed to TPM (N = 483) (Analysis 50.5). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.02 to 0.01; I² = 0%).

50.5.2 Routine health record studies

No included studies reported data on this outcome.

51 PHT versus ZNS

51.1 All major malformations

51.1.1 Cohort studies

Pooled results from two cohort studies suggested no evidence of a difference in risk (RR 1.28, 95% CI 0.42 to 3.93; I² = 61%), with no difference in the number of major malformations in children exposed to PHT (N = 522) and children exposed to ZNS (N = 116) (Analysis 51.1). Due to high heterogeneity, a random‐effects RR was calculated which found a similar effect (RR 1.28, 95% CI 0.42 to 3.93, I² = 61%). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.02 to 0.05; I² = 68%). Due to heterogeneity, a random‐effects RD was calculated which also found no difference in risk (RD 0.00, 95% CI ‐0.11 to 0.11, I² = 68%)

51.1.2 Routine health record studies

No included studies reported data on this outcome.

51.2 Neural tube malformations

51.2.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 0.11, 95% CI 0.00 to 2.58, I² = NA), with no difference in the number of neural tube malformations in children exposed to PHT (N = 82) and children exposed to ZNS (N = 26) (Analysis 51.2). The RD also suggested no difference in the level of risk (RD ‐0.04, 95% CI −0.13 to 0.05; I² = NA).

51.2.2 Routine health record studies

No included studies reported data on this outcome.

51.3 Cardiac malformations

51.3.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 0.98, 95% CI 0.04 to 23.26, I² = NA), with no difference in the number of cardiac malformations in children exposed to PHT (N = 82) and children exposed to ZNS (N = 26) (Analysis 51.3). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.05 to 0.07; I² = NA).

51.3.2 Routine health record studies

No included studies reported data on this outcome.

51.4 Oro‐facial cleft/craniofacial malformations

51.4.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 0.98, 95% CI 0.04 to 23.26, I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to PHT (N = 82) and children exposed to ZNS (N = 26) (Analysis 51.4). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.05 to 0.07; I² = NA).

51.4.2 Routine health record studies

No included studies reported data on this outcome.

51.5 Skeletal/limb malformations

51.5.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported skeletal/limb malformations in children exposed to PHT (N = 82) or ZNS (N = 26) (Analysis 51.5).

51.5.2 Routine health record studies

No included studies reported data on this outcome.

52 PB versus OXC

52.1 All major malformations

52.1.1 Cohort studies

Pooled results from eight cohort studies suggested no evidence of a difference in risk (RR 1.61, 95% CI 0.83 to 3.14; I² = 19%), with no difference in the number of major malformations in children exposed to PB (N = 354) and children exposed to OXC (N = 322) (Analysis 52.1). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.02 to 0.06; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 6.5% (95% CI 4.2 to 9.9) for children exposed to PB and 3.0% (95% CI 1.4 to 5.4) for children exposed to OXC. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Phenobarbital dose and Oxcarbazepine dose sections).

52.1.2 Routine health record studies

Pooled results from two routine health record studies suggested no evidence of a difference in risk (RR 3.07, 95% CI 0.50 to 18.92; I² = 0%), with no difference in the number of major malformations in children exposed to PB (N = 34) and children exposed to OXC (N = 61) (Analysis 52.1). The RD also suggested no difference in the level of risk (RD 0.07, 95% CI −0.04 to 0.17; I² = 0%).

52.2 Neural tube malformations

52.2.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 1.57, 95% CI 0.06 to 37.94; I² = NA), with no difference in the number of neural tube malformations in children exposed to PB (N = 349) and children exposed to OXC (N = 305) (Analysis 52.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.01 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to PB was 0.68% (2/294) and 0% for children exposed to OXC (0/333).

52.2.2 Routine health record studies

No included studies reported data on this outcome.

52.3 Cardiac malformations

52.3.1 Cohort studies

Pooled results from seven cohort studies suggested no evidence of a difference in risk (RR 2.58, 95% CI 0.94 to 7.09; I² = 51%), with children exposed to PB (N = 352) experiencing comparable cardiac malformations to children exposed to OXC (N = 306) (Analysis 52.3). Due to high heterogeneity, a random‐effects RR was calculated which also found no difference (RR 3.84, 95% CI 0.54 to 27.19, I² = 51%). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.01 to 0.05; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to PB was 2.72% (8/294) and 1.20% for children exposed to OXC (4/333).

52.3.2 Routine health record studies

No included studies reported data on this outcome.

52.4 Oro‐facial cleft/craniofacial malformations

52.4.1 Cohort studies

Pooled results from five cohort studies suggested no evidence of a difference in risk (RR 3.66, 95% CI 0.41 to 32.43; I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to PB (N = 212) and children exposed to TPM (N = 234) (Analysis 52.4). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.01 to 0.04; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to PB was 0.34% (1/294) and 0.30% for children exposed to OXC (1/333).

52.4.2 Routine health record studies

No included studies reported data on this outcome.

52.5 Skeletal/limb malformations

52.5.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.98, 95% CI 0.16 to 5.97; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to PB (N = 349) and children exposed to OXC (N = 305) (Analysis 52.5). The RD also suggested no difference in the level of risk (RD ‐0.00, 95% CI −0.02 to 0.02; I² = 0%).

52.5.2 Routine health record studies

No included studies reported data on this outcome.

53 PB versus PRM

53.1 All major malformations

53.1.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.50, 95% CI 0.21 to 1.16; I² = 0%), with no difference in the number of major malformations in children exposed to PB (N = 241) and children exposed to PRM (N = 110) (Analysis 53.1). The RD also suggested no difference in the level of risk (RD −0.05, 95% CI −0.12 to 0.02; I² = 0%).

53.1.2 Routine health record studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 53.1). However, available data showed there were 1/7 cases of major malformations in children exposed to PB and 0/3 cases in children exposed to PRM, based on data from one study (Sweden Health Record Registers).

53.2 Neural tube malformations

53.2.1 Cohort studies

We were unable to estimate a RR from two studies due to there being no reported no cases of neural tube malformations in children exposed to PB (N = 95) or PRM (N = 39) (Analysis 53.2).

53.2.2 Routine health record studies

No included studies reported data on this outcome.

53.3 Cardiac malformations

53.3.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.42, 95% CI 0.03 to 6.55, I² = NA), with no difference in the number of cardiac malformations in children exposed to PB (N = 95) and children exposed to PRM (N = 39) (Analysis 53.1). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.08 to 0.05; I² = 0%).

53.3.2 Routine health record studies

No included studies reported data on this outcome.

53.4 Oro‐facial cleft/craniofacial malformations

53.4.1 Cohort studies

We were unable to estimate a RR from two studies due to there being no reported cases of oro‐facial cleft/craniofacial malformations in children exposed to PB (N = 95) or PRM (N = 39) (Analysis 53.4).

53.4.2 Routine health record studies

No included studies reported data on this outcome.

53.5 Skeletal/limb malformations

53.5.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 1.29, 95% CI 0.05 to 30.82, I² = NA), with no difference in the number of skeletal/limb malformations in children exposed to PB (N = 95) and children exposed to PRM (N = 39) (Analysis 53.5). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.05 to 0.07; I² = 0%).

53.5.2 Routine health record studies

No included studies reported data on this outcome.

54 PB versus TPM

54.1 All major malformations

54.1.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 1.38, 95% CI 0.68 to 2.81; I² = 0%), with no difference in the number of major malformations in children exposed to PB (N = 340) and children exposed to TPM (N = 426) (Analysis 54.1). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.02 to 0.05; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 6.5% (95% CI 4.2 to 9.9) for children exposed to PB and 3.9% (95% CI 1.5 to 8.4) for children exposed to TPM. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Phenobarbital dose and Topiramate dose sections).

54.1.2 Routine health record studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 54.1). However, available data showed there were 3/34 cases of major malformations in children exposed to PB and 2/49 cases in children exposed to TPM, based on data from two studies (Norwegian Health Record Registers; Sweden Health Record Registers).

54.2 Neural tube malformations

54.2.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 0.22, 95% CI 0.01 to 5.00; I² = NA), with no difference in the number of neural tube malformations in children exposed to PB (N = 343) and children exposed to TPM (N = 417) (Analysis 54.2). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.02 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to PB was 0.68% (2/294) and 0% for children exposed to TPM (0/152).

54.2.2 Routine health record studies

No included studies reported data on this outcome.

54.3 Cardiac malformations

54.3.1 Cohort studies

Pooled results from four cohort studies suggested an increased risk with PB (RR 4.44, 95% CI 0.98 to 20.12; I² = 37%), with children exposed to PB (N = 343) experiencing more cardiac malformations than children exposed to TPM (N = 417) (Analysis 54.3). However, the RD suggested no difference in the level of risk (RD 0.02, 95% CI −0.00 to 0.05; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to PB was 2.72% (8/294) and 1.97% for children exposed to TPM (3/152).

54.3.2 Routine health record studies

No included studies reported data on this outcome.

54.4 Oro‐facial cleft/craniofacial malformations

54.4.1 Cohort studies

Pooled results from three cohort studies suggested no evidence of a difference in risk (RR 1.44, 95% CI 0.39 to 5.31; I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to PB (N = 206) and children exposed to TPM (N = 408) (Analysis 54.4). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.02 to 0.03; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to PB was 0.34% (1/294) and 0% for children exposed to TPM (0/152).

54.4.2 Routine health record studies

No included studies reported data on this outcome.

54.5 Skeletal/limb malformations

54.5.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 0.36, 95% CI 0.06 to 2.19; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to PB (N = 343) and children exposed to TPM (N = 417) (Analysis 54.5). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.03 to 0.01; I² = 0%).

54.5.2 Routine health record studies

No included studies reported data on this outcome.

55 PB versus ZNS

55.1 All major malformations

55.1.1 Cohort studies

Pooled results from two cohort studies suggested no evidence of a difference in risk (RR 10.46, 95% CI 0.62 to 175.67; I² = NA), with no difference in the number of major malformations in children exposed to PB (N = 201) and children exposed to ZNS (N = 91) (Analysis 55.1). The RD suggested a higher rate of major malformation observed in the PB‐exposed group (RD 0.05, 95% CI 0.02 to 0.09; I² = 0%).

55.1.2 Routine health record studies

No included studies reported data on this outcome.

55.2 Neural tube malformations

55.2.1 Cohort studies

We were unable to estimate a RR from one study due to there being no cases of neural tube malformations in children exposed to PB (N = 2) or ZNS (N = 1) (Analysis 55.2).

55.2.2 Routine health record studies

No included studies reported data on this outcome.

55.3 Cardiac malformations

55.3.1 Cohort studies

We were unable to estimate a RR from one study due to there being no cases of cardiac malformations in children exposed to PB (N = 2) or ZNS (N = 1) (Analysis 55.3).

55.3.2 Routine health record studies

No included studies reported data on this outcome.

55.4 Oro‐facial cleft/craniofacial malformations

55.4.1 Cohort studies

We were unable to estimate a RR from one study due to there being no cases of oro‐facial cleft /craniofacial malformations PB (N = 2) or ZNS (N = 1) (Analysis 55.4).

55.4.2 Routine health record studies

No included studies reported data on this outcome.

55.5 Skeletal/limb malformations

55.5.1 Cohort studies

We were unable to estimate a RR from one study due to there being no cases of skeletal/limb malformations in children exposed to PB (N = 2) or ZNS (N = 1) (Analysis 55.5).

55.5.2 Routine health record studies

No included studies reported data on this outcome.

56 TPM versus ZNS

56.1 All major malformations

56.1.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 1.59, 95% CI 0.54 to 4.66; I² = 58%), with no difference in the number of major malformations in children exposed to TPM (N = 440) and children exposed to ZNS (N = 130) (Analysis 56.1). Due to high heterogeneity, a random‐effects RR was calculated which also found no difference (RD 1.44, 95% CI 0.19 to 10.82, I² =58%). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI ‐0.02 to 0.06; I² = 35%).

56.1.2 Routine health record studies

No included studies reported data on this outcome.

56.2 Neural tube malformations

56.2.1 Cohort studies

We were unable to estimate a RR from two studies due to there being no cases of neural tube malformations in children exposed to TPM (N = 11) or ZNS (N = 14) (Analysis 56.2).

56.2.2 Routine health record studies

No included studies reported data on this outcome.

56.3 Cardiac malformations

56.3.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 6.00 95% CI 0.28 to 129.16, I² = NA), with no difference in the number of cardiac malformations in children exposed to TPM (N = 81) and children exposed to ZNS (N = 40) (Analysis 56.3). The RD also suggested no difference in the level of risk (RD 0.03, 95% CI −0.06 to 0.12; I² = 0%).

56.3.2 Routine health record studies

No included studies reported data on this outcome.

56.4 Oro‐facial cleft/craniofacial malformations

56.4.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 1.90 95% CI 0.09 to 38.34, I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to TPM (N = 81) and children exposed to ZNS (N = 40) (Analysis 56.4). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.06 to 0.10; I² = 0%).

56.4.2 Routine health record studies

No included studies reported data on this outcome.

56.5 Skeletal/limb malformations

56.5.1 Cohort studies

We were unable to estimate a RR from three studies due to there being no cases of skeletal/limb malformations in children exposed to TPM (N = 81) or ZNS (N = 40) (Analysis 56.5).

56.5.2 Routine health record studies

No included studies reported data on this outcome.

57 TPM vs LAC

57.1 All major malformations

57.1.1 Cohort studies

We were unable to estimate a RR from one cohort study due to there being no major malformations observed in children exposed to TPM (N = 5) or LAC (N = 1) (Analysis 57.1).

57.1.2 Routine health record studies

No included studies reported data on this outcome.

57.2 Neural tube malformations

57.2.1 Cohort studies

We were unable to estimate a RR from one study due to there being no cases of neural tube malformations in children exposed to TPM (N = 5) or LAC (N = 1) (Analysis 57.2).

57.2.2 Routine health record studies

No included studies reported data on this outcome.

57.3 Cardiac malformations

57.3.1 Cohort studies

We were unable to estimate a RR from one study due to there being no cases of cardiac malformations in children exposed to TPM (N = 5) or LAC (N = 1) (Analysis 57.3).

57.3.2 Routine health record studies

No included studies reported data on this outcome.

57.4 Oro‐facial cleft/craniofacial malformations

57.4.1 Cohort studies

We were unable to estimate a RR from one study due to there being no cases of oro‐facial cleft/ craniofacial malformations in children exposed to TPM (N = 5) or LAC (N = 1) (Analysis 57.4).

57.4.2 Routine health record studies

No included studies reported data on this outcome.

57.5 Skeletal/limb malformations

57.5.1 Cohort studies

We were unable to estimate a RR from one study due to there being no cases of skeletal/limb malformations in children exposed to TPM (N = 5) or LAC (N = 1) (Analysis 57.5).

57.5.2 Routine health record studies

No included studies reported data on this outcome.

58 VPA versus GBP

58.1 All major malformations

58.1.1 Cohort studies

Pooled results from four cohort studies suggested an increased risk with VPA (RR 4.27, 95% CI 1.60 to 11.35; I² = 58%), with children exposed to VPA (N = 1839) experiencing more major malformations than children exposed to GBP (N = 192) (Analysis 58.1). Due to high heterogeneity, a random‐effects RR was calculated which found no difference in the level of risk (RD 2.43, 95% CI 0.40 to 14.64, I² = 58%). However, both the fixed‐effect RD analysis (RD 0.08, 95% CI 0.04 to 0.11; I² = 60%) and a random‐effects RD also suggested a higher absolute risk for VPA (RD 0.08, 95% CI 0.01 to 0.14, I² = 60%).

58.1.2 Routine health record studies

Results from one study suggested no evidence of a difference in risk (RR 3.74, 95% CI 0.24 to 59.08, I² = NA), with no difference in the number of major malformations in children exposed to VPA (N = 268) and children exposed to GBP (N = 18) (Analysis 58.1). However, the RD suggested a difference in the level of risk (RD 0.10, 95% CI 0.02 to 0.18; I² = NA).

58.2 Neural tube malformations

58.2.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.83, 95% CI 0.05 to 13.81, I² = NA), with no difference in the number of neural tube malformations in children exposed to VPA (N = 277) and children exposed to GBP (N = 18) (Analysis 58.2). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI ‐0.09 to 0.13; I² = 0%).

58.2.2 Routine health record studies

No included studies reported data on this outcome.

58.3 Cardiac malformations

58.3.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.46, 95% CI 0.08 to 2.70, I² = 4%), with no difference in the number of cardiac malformations in children exposed to VPA (N = 277) and children exposed to GBP (N = 16) (Analysis 58.3). The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI ‐0.14 to 0.11; I² = 74%).

58.3.2 Routine health record studies

No included studies reported data on this outcome.

58.4 Oro‐facial cleft/craniofacial malformations

58.4.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 1.38, 95% CI 0.09 to 22.19, I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to VPA (N = 277) and children exposed to GBP (N = 16) (Analysis 58.4). The RD also suggested no difference in the level of risk (RD 0.04, 95% CI ‐0.07 to 0.15; I² = 0%).

58.4.2 Routine health record studies

No included studies reported data on this outcome.

58.5 Skeletal/limb malformations

58.5.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.72, 95% CI 0.04 to 12.14, I² = NA), with no difference in the number of skeletal/limb malformations in children exposed to VPA (N = 277) and children exposed to GBP (N = 16)(Analysis 58.5). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI ‐0.09 to 0.13; I² = 0%).

58.5.2 Routine health record studies

No included studies reported data on this outcome.

59 VPA vs LAC

59.1 All major malformations

59.1.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 59.1). However, available data showed there were 4/17 cases of major malformations in children exposed to VPA and 0/1 cases in children exposed to LAC, based on data from one study (Jimenez 2020).

59.1.2 Routine health record studies

No included studies reported data on this outcome.

59.2 Neural tube malformations

59.2.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 59.3). However, available data showed there were 1/17 cases of neural tube malformations in children exposed to VPA and 0/1 cases in children exposed to LAC, based on data from one study (Jimenez 2020).

59.2.2 Routine health record studies

No included studies reported data on this outcome.

59.3 Cardiac malformations

59.3.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 59.4). However, available data showed there were 1/17 cases of cardiac malformations in children exposed to VPA and 0/1 cases in children exposed to LAC, based on data from one study (Jimenez 2020).

59.3.2 Routine health record studies

No included studies reported data on this outcome.

59.4 Oro‐facial cleft/craniofacial malformations

59.4.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 59.5). However, available data showed there were 1/17 cases of oro‐facial cleft/craniofacial malformations in children exposed to VPA and 0/1 cases in children exposed to LAC, based on data from one study (Jimenez 2020).

59.4.2 Routine health record studies

No included studies reported data on this outcome.

59.5 Skeletal/limb malformations

59.5.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 59.2). However, available data showed that there were 0/17 cases of skeletal/limb malformations in children exposed to VPA (N = 17) and 0/1 in children exposed to LAC, based on data from one study (Jimenez 2020).

59.5.2 Routine health record studies

No included studies reported data on this outcome.

60 VPA versus LEV

60.1 All major malformations

60.1.1 Cohort studies

Pooled results from 10 cohort studies suggested an increased risk with VPA (RR 3.77, 95% CI 2.48 to 5.74; I² = 17%), with children exposed to VPA (N = 2342) experiencing more major malformations than children exposed to LEV (N = 1143) (Analysis 60.1). The RD also suggested a higher absolute risk for VPA (RD 0.07, 95% CI 0.05 to 0.08; I² = 11%).

The EURAP 2018 collaboration reported the prevalence of MCM was 10.3% (95% CI 8.8 to 12.0) for children exposed to VPA and 2.8% (95% CI 1.7 to 4.5) for children exposed to LEV. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Valproate dose and Levetiracetam dose sections).

60.1.2 Routine health record studies

Pooled results from two routine health record studies suggested an increased risk with VPA (RR 3.26, 95% CI 1.51 to 7.03; I² = 0%), with children exposed to VPA (N = 663) experiencing more major malformations than children exposed to LEV (N = 248) (Analysis 60.1). The RD also suggested a higher risk for VPA (RD 0.06, 95% CI 0.03 to 0.09; I² = 28%).

60.2 Neural tube malformations

60.2.1 Cohort studies

Pooled results from nine cohort studies suggested an increased risk with VPA (RR 3.76, 95% CI 1.22 to 11.55; I² = 0%), with children exposed to VPA (N = 2298) experiencing more neural tube malformations than children exposed to LEV (N = 1048) (Analysis 60.2). The RD also suggested a higher risk for VPA (RD 0.01, 95% CI 0.00 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to VPA was 1.15% (16/1381) and 0% for children exposed to LEV (0/599).

60.2.2 Routine health record studies

No included studies reported data on this outcome.

60.3 Cardiac malformations

60.3.1 Cohort studies

Pooled results from 10 cohort studies suggested an increased risk with VPA (RR 3.04, 95% CI 1.46 to 6.34; I² = 0%), with children exposed to VPA (N = 2299) experiencing more cardiac malformations than children exposed to LEV (N = 1057) (Analysis 60.3). The RD also suggested a higher risk for VPA (RD 0.02, 95% CI 0.01 to 0.03; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to VPA was 2.46% (34/1381) and 0.83% for children exposed to LEV 0.83% (5/599).

60.3.2 Routine health record studies

No included studies reported data on this outcome.

60.4 Oro‐facial cleft/craniofacial malformations

60.4.1 Cohort studies

Pooled results from nine studies suggested an increased risk with VPA (RR 3.75, 95% CI 1.19 to 11.77; I² = 0%), with children exposed to VPA (N = 1958) experiencing more oro‐facial cleft/craniofacial malformations than children exposed to LEV (N = 951) (Analysis 60.4). The RD also suggested a higher risk for VPA (RD 0.01, 95% CI 0.00 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to VPA was 0.43% (6/1381) and 0.16% for children exposed to LEV (1/599).

60.4.2 Routine health record studies

No included studies reported data on this outcome.

60.5 Skeletal/limb malformations

60.5.1 Cohort studies

Pooled results from nine cohort studies suggested an increased risk with VPA (RR 2.41, 95% CI 0.99 to 5.85; I² = 55%), with children exposed to VPA (N = 2298) experiencing more skeletal/limb malformations than children exposed to LEV (N = 1048) (Analysis 60.5). Due to high heterogeneity, a random‐effects RR was calculated which found no difference in the level of risk (RR 1.89, 95% CI 0.34 to 10.60, I² = 55%). However, both the fixed‐effect RD analysis (RD 0.01, 95% CI 0.00 to 0.02; I² = 12%) and the random‐effects RD analysis (RD 0.01, 95% CI 0.00 to 0.02; I² = 12%) also suggested a higher absolute risk for VPA.

60.5.2 Routine health record studies

No included studies reported data on this outcome.

61 VPA versus LTG

61.1 All major malformations

61.1.1 Cohort studies

Pooled results from 12 cohort studies suggested an increased risk with VPA (RR 3.50, 95% CI 2.76 to 4.46; I² = 0%), with children exposed to VPA (N = 2459) experiencing more major malformations than children exposed to LTG (N = 4437) (Analysis 61.1). The RD suggested an increased risk for VPA (RD 0.06, 95% CI 0.05 to 0.08; I² = 34%).

The EURAP 2018 collaboration reported the prevalence of MCM was 10.3% (95% CI 8.8 to 12.0) for children exposed to VPA and 2.9% (95% CI 2.3 to 3.7) for children exposed to LTG. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Valproate dose and Lamotrigine dose sections).

61.1.2 Routine health record studies

Pooled results from four routine health record studies suggested an increased risk with VPA (RR 2.49, 95% CI 1.86 to 3.35; I² = 0%), with children exposed to VPA (N = 1088) experiencing more major malformations than children exposed to LTG (N = 2502) (Analysis 61.1). The RD also suggested an increased level of risk for VPA (RD 0.05, 95% CI 0.03 to 0.07; I² = 42%).

61.2 Neural tube malformations

61.2.1 Cohort studies

Pooled results from 11 cohort studies suggested an increased risk with VPA (RR 7.48, 95% CI 3.27 to 17.13; I² = 0%), with children exposed to VPA (N = 2415) experiencing more neural tube malformations than children exposed to LTG (N = 4293) (Analysis 61.2). The RD also suggested an increased level of risk for VPA (RD 0.01, 95% CI 0.01 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to VPA was 1.15% (16/1381) and 0.3% for children exposed to LTG (1/2514).

61.2.2 Routine health record studies

No included studies reported data on this outcome.

61.3 Cardiac malformations

61.3.1 Cohort studies

Pooled results from 12 cohort studies suggested an increased risk with VPA (RR 3.39, 95% CI 2.06 to 5.60; I² = 0%), with children exposed to VPA (N = 2416) experiencing more neural tube malformations than children exposed to LTG (N = 4313) (Analysis 61.3). The RD also suggested an increased level of risk for VPA (RD 0.02, 95% CI 0.01 to 0.02; I² = 3%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to VPA was 2.46% (34/1381) and 5.9% for children exposed to LTG (15/2514).

61.3.2 Routine health record studies

No included studies reported data on this outcome.

61.4 Oro‐facial cleft/craniofacial malformations

61.4.1 Cohort studies

Pooled results from 11 cohort studies suggested an increased risk with VPA (RR 4.16, 95% CI 2.14 to 8.08; I² = 0%), with children exposed to VPA (N = 2075) experiencing more craniofacial malformations than children exposed to LTG (N = 4263) (Analysis 61.4). The RD also suggested an increased level of risk for VPA (RD 0.01, 95% CI 0.01 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to VPA was 0.43% (6/1381) and 0.11% for children exposed to LTG (3/2514).

61.4.2 Routine health record studies

No included studies reported data on this outcome.

61.5 Skeletal/limb malformations

61.5.1 Cohort studies

Pooled results from 11 cohort studies suggested an increased risk with VPA (RR 6.09, 95% CI 2.91 to 12.76; I² = 0%), with children exposed to VPA (N = 2415) experiencing more craniofacial malformations than children exposed to LTG (N = 4293) (Analysis 61.5). The RD also suggested an increased level of risk for VPA (RD 0.01, 95% CI 0.01 to 0.02; I² = 26%).

61.5.2 Routine health record studies

No included studies reported data on this outcome.

62 VPA versus TPM

62.1 All major malformations

62.1.1 Cohort studies

Pooled results from seven cohort studies suggested an increased risk with VPA (RR 2.47, 95% CI 1.50 to 4.08; I² = 0%), with children exposed to VPA (N = 2219) experiencing more major malformations than children exposed to TPM (N = 504) (Analysis 62.1). The RD also suggested a higher absolute risk for VPA (RD 0.06, 95% CI 0.03 to 0.09; I² = 41%). Due to high heterogeneity, we undertook a random‐effects analysis which found a similar effect (RD 0.07, 95% CI 0.02 to 0.11; I² = 41%).

The EURAP 2018 collaboration reported the prevalence of MCM was 10.3% (95% CI 8.8 to 12.0) for children exposed to VPA and 3.9% (95% CI 1.5 to 8.4) for children exposed to TPM. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Valproate dose and Topiramate dose sections).

62.1.2 Routine health record studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 1.27, 95% CI 0.36 to 4.39, I² = 0%), with no difference in the number of major malformations in children exposed to VPA (N = 601) and children exposed to TPM (N = 49) (Analysis 62.1). The RD also suggested no difference in the level of risk (RD 1.27, 95% CI 0.36 to 4.39; I² = 0%).

62.2 Neural tube malformations

62.2.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 2.39, 95% CI 0.73 to 7.80; I² = 2%), with no difference in the number of neural tube malformations in children exposed to VPA (N = 2175) and children exposed to TPM (N = 490) (Analysis 62.2). The RD suggested a higher risk for VPA (RD 0.01, 95% CI 0.00 to 0.03; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to VPA was 1.15% (16/1381) and 0% for children exposed to TPM (0/152).

62.2.2 Routine health record studies

No included studies reported data on this outcome.

62.3 Cardiac malformations

62.3.1 Cohort studies

Pooled results from six cohort studies suggested an increased risk with VPA (RR 3.48, 95% CI 1.16 to 10.48; I² = 0%), with children exposed to VPA (N = 2175) experiencing more cardiac malformations than children exposed to TPM (N = 495) (Analysis 62.3). The RD also suggested a higher absolute risk for VPA (RD 0.02, 95% CI 0.01 to 0.04; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to VPA was 2.46% (34/1381) and 1.97% for children exposed to TPM (3/152).

62.3.2 Routine health record studies

No included studies reported data on this outcome.

62.4 Oro‐facial cleft/craniofacial malformations

62.4.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.89 95% CI 0.37 to 2.13; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to VPA (N = 1835) and children exposed to TPM (N = 482) (Analysis 62.4). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.02 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to VPA was 0.43% (6/1381) and 0% for children exposed to TPM (0/152).

62.4.2 Routine health record studies

No included studies reported data on this outcome.

62.5 Skeletal/limb malformations

62.5.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 1.45, 95% CI 0.55 to 3.82; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to VPA (N = 2199) and children exposed to TPM (N = 490) (Analysis 62.5). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI ‐0.00 to 0.02; I² = 0%).

62.5.2 Routine health record studies

No included studies reported data on this outcome.

63 VPA versus OXC

63.1 All major malformations

63.1.1 Cohort studies

Pooled results from 11 cohort studies suggested an increased risk with VPA (RR 2.48, 95% CI 1.42 to 4.31; I² = 13%), with children exposed to VPA (N = 1183) experiencing more major malformations than children exposed to OXC (N = 378) (Analysis 63.1). The RD also suggested a higher risk for VPA (RD 0.06, 95% CI 0.03 to 0.09; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 10.3% (95% CI 8.8 to 12.0) for children exposed to VPA and 3.0% (95% CI 1.4 to 5.4) for children exposed to OXC. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Valproate dose and Oxcarbazepine dose sections).

63.1.2 Routine health record studies

Results from four routine health record studies suggested an increased risk with VPA (RR 1.60, 95% CI 1.11 to 2.29; I² = 80%), with children exposed to VPA (N = 1194) experiencing more major malformations than children exposed to OXC (N = 507) (Analysis 63.1). Due to heterogeneity, a random‐effects RR was calculated, which showed no evidence of a difference in risk (RR 1.80, 95% CI 0.57 to 5.67; I² = 80%). The RD suggested a higher risk for VPA (RD 0.04, 95% CI 0.01 to 0.08; I² = 65%). Due to heterogeneity, a random‐effects RD was calculated which showed no evidence of a difference in risk (RD 0.04, 95% CI ‐0.01 to 0.10, I² = 65%).

63.2.1 Cohort studies

Pooled results from nine cohort studies suggested no evidence of a difference in risk (RR 1.55, 95% CI 0.49 to 4.89; I² = 0%), with no difference in the number of neural tube malformations in children exposed to VPA (N = 1133) and children exposed to OXC (N = 364) (Analysis 63.2). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI ‐0.00 to 0.03; I² = 0%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to VPA was 1.15% (16/1381) and 0% for children exposed to OXC (0/333).

63.2.2 Routine health record studies

No included studies reported data on this outcome.

63.3 Cardiac malformations

63.3.1 Cohort studies

Pooled results from 11 cohort studies suggested no evidence of a difference in risk (RR 1.80, 95% CI 0.84 to 3.88; I² = 0%), with no difference in the number of cardiac malformations in children exposed to VPA (N = 1140) and children exposed to OXC (N = 457) (Analysis 63.3). The RD suggested a higher risk for VPA (RD 0.02, 95% CI 0.00 to 0.04; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to VPA was 2.46% (34/1381) and 1.20% for children exposed to OXC (4/333).

63.3.2 Routine health record studies

No included studies reported data on this outcome.

63.4 Oro‐facial cleft/craniofacial malformations

63.4.1 Cohort studies

Pooled results from nine cohort studies suggested no evidence of a difference in risk (RR 2.14, 95% CI 0.76 to 6.06; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to VPA (N = 793) and children exposed to OXC (N = 385) (Analysis 63.4). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI ‐0.00 to 0.04; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to VPA was 0.43% (6/1381) and 0.30% for children exposed to OXC (1/333).

63.4.2 Routine health record studies

No included studies reported data on this outcome.

63.5 Skeletal/limb malformations

63.5.1 Cohort studies

Pooled results from nine cohort studies suggested no evidence of a difference in risk (RR 1.37, 95% CI 0.42 to 4.49; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to VPA (N = 1133) and children exposed to OXC (N = 364) (Analysis 63.5). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI ‐0.01 to 0.02; I² = 0%).

63.5.2 Routine health record studies

No included studies reported data on this outcome.

64 VPA versus PB

64.1 All major malformations

64.1.1 Cohort studies

Pooled results from 23 cohort studies suggested an increased risk with VPA (RR 1.49, 95% CI 1.08 to 2.07; I² = 0%), with children exposed to VPA (N = 1557) experiencing more major malformations than children exposed to PB (N = 759) (Analysis 64.1). The RD also suggested a higher risk for VPA (RD 0.04, 95% CI 0.01 to 0.06; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 10.3% (95% CI 8.8 to 12.0) for children exposed to VPA and 6.5% (95% CI 4.2 to 9.9) for children exposed to PB. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Valproate dose and Phenobarbital dose sections). Samren 1997 reported six cases of major malformation out of 184 (9%) VPA‐exposed children and five cases from 48 (10%) PB‐exposed children.

64.1.2 Routine health record studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.79, 95% CI 0.26 to 2.42, I² = 0%), with no difference in the number of major malformations in children exposed to VPA (N = 601) and children exposed to PRM (N = 34) (Analysis 64.1). The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI −0.12 to 0.08; I² = 0%).

64.2 Neural tube malformations

64.2.1 Cohort studies

Pooled results from 14 cohort studies suggested evidence of a difference in risk (RR 3.04, 95% CI 1.27 to 7.30; I² = 1%), with children exposed to VPA (N = 1174) experiencing more neural tube malformations to children exposed to PB (N = 546) (Analysis 64.2). The RD also suggested a difference in the level of risk (RD 0.02, 95% CI 0.01 to 0.04; I² = 7%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to VPA was 1.15% (16/1381) and 0.68% for children exposed to PB (2/294).

64.2.2 Routine health record studies

No included studies reported data on this outcome.

64.3 Cardiac malformations

64.3.1 Cohort studies

Pooled results from 14 cohort studies suggested no evidence of a difference in risk (RR 0.84, 95% CI 0.50 to 1.43; I² = 0%), with no difference in the number of cardiac malformations in children exposed to VPA (N = 1174) and children exposed to PB (N = 546) (Analysis 64.3). The RD also suggested no difference in the level of risk (RD −0.00, 95% CI −0.03 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to VPA was 2.46% (34/1381) and 2.72% for children exposed to PB (8/294).

64.3.2 Routine health record studies

No included studies reported data on this outcome.

64.4 Oro‐facial cleft/craniofacial malformations

64.4.1 Cohort studies

Pooled results from 14 cohort studies suggested no evidence of a difference in risk (RR 0.54, 95% CI 0.23 to 1.27; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to VPA (N = 839) and children exposed to PB (N = 418) (Analysis 64.4). The RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.03 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to VPA was 0.43% (6/1381) and 0.34% for children exposed to PB (1/294).

64.4.2 Routine health record studies

No included studies reported data on this outcome.

64.5 Skeletal/limb malformations

64.5.1 Cohort studies

Pooled results from 14 cohort studies suggested no evidence of a difference in risk (RR 1.62, 95% CI 0.70 to 3.74; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to VPA (N = 1174) and children exposed to PB (N = 546) (Analysis 64.5). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI ‐0.01 to 0.03; I² = 0%).

64.5.2 Routine health record studies

No included studies reported data on this outcome.

65 VPA versus PHT

65.1 All major malformations

65.1.1 Cohort studies

Pooled results from 21 cohort studies suggested an increased risk with VPA (RR 1.92, 95% CI 1.44 to 2.56; I² = 0%), with children exposed to VPA (N = 2650) experiencing more major malformations than children exposed to PHT (N = 1247) (Analysis 65.1). The RD also suggested a higher risk for VPA (RD 0.05, 95% CI 0.03 to 0.07; I² = 4%).

The EURAP 2018 collaboration reported the prevalence of MCM was 10.3% (95% CI 8.8 to 12.0) for children exposed to VPA and 6.4% (95% CI 2.8 to 12.2) for children exposed to PHT. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see Valproate dose and Phenytoin dose sections). Samren 1997 reported six cases of major malformation in 184 (9%) children exposed to VPA and nine in 141 (6%) PHT‐exposed children.

65.1.2 Routine health record studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 1.43, 95% CI 0.64 to 3.19; I² = 0%), with no difference in the number of major malformations in children exposed to VPA (N = 268) and children exposed to PHT (N = 103) (Analysis 65.1). The RD also suggested no difference in the level of risk (RD 0.03, 95% CI −0.03 to 0.09; I² = NA).

65.2 Neural tube malformations

65.2.1 Cohort studies

Pooled results from 14 cohort studies suggested an increased risk with VPA (RR 3.75, 95% CI 1.57 to 8.94; I² = 0%), with children exposed to VPA (N = 2419) experiencing more neural tube malformations than children exposed to PHT (N = 974) (Analysis 65.2). The RD also suggested a higher risk for VPA (RD 0.02, 95% CI 0.01 to 0.03; I² = 2%).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to VPA was 1.15% (16/1381) and 0.80% for children exposed to PHT (1/125).

65.2.2 Routine health record studies

No included studies reported data on this outcome.

65.3 Cardiac malformations

65.3.1 Cohort studies

Pooled results from 14 cohort studies suggested an increased risk with VPA (RR 1.90, 95% CI 1.07 to 3.36; I² = 0%), with children exposed to VPA (N = 2419) experiencing more cardiac malformations than children exposed to PHT (N = 974) (Analysis 65.3). The RD also suggested a higher risk for VPA (RD 0.02, 95% CI 0.00 to 0.03; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to VPA was 2.46% (34/1381) and 4.0% for children exposed to PHT (5/125).

65.3.2 Routine health record studies

No included studies reported data on this outcome.

65.4 Oro‐facial cleft/craniofacial malformations

65.4.1 Cohort studies

Pooled results from 14 cohort studies suggested no evidence of a difference in risk (RR 2.24, 95% CI 0.89 to 5.58; I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to VPA (N = 2084) and children exposed to PHT (N = 860) (Analysis 65.4). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI ‐0.00 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to VPA was 0.43% (6/1381) and 0% for children exposed to PHT (0/125).

65.4.2 Routine health record studies

No included studies reported data on this outcome.

65.5 Skeletal/limb malformations

65.5.1 Cohort studies

Pooled results from 14 cohort studies suggested an increased risk with VPA (RR 2.12, 95% CI 1.01 to 4.45; I² = 0%), with children exposed to VPA (N = 2419) experiencing more skeletal/limb malformations than children exposed to PHT (N = 975) (Analysis 65.5). The RD also suggested a higher risk for VPA (RD 0.01, 95% CI 0.00 to 0.02; I² = 0%).

65.5.2 Routine health record studies

No included studies reported data on this outcome.

66 VPA versus ZNS

66.1 All major malformations

66.1.1 Cohort studies

Pooled results from three cohort studies suggested no evidence of a difference in risk (RR 2.34, 95% CI 0.95 to 5.80; I² = 77%), with children exposed to VPA (N = 1560) experiencing comparable major malformations to children exposed to ZNS (N = 117) (Analysis 66.1). Due to high heterogeneity, a random‐effects RR analysis was completed which also found no difference (RR 1.81, 95% CI 0.14 to 22.75; I² = 77%). The RD suggested a higher risk for VPA (RD 0.06, 95% CI 0.01 to 0.10; I² = 72%). However, due to heterogeneity, a random‐effects RD was calculated which found no difference in the level of risk (RD 0.04, 95% CI ‐0.11 to 0.19, I² = 72%).

66.1.2 Routine health record studies

No included studies reported data on this outcome.

66.2 Neural tube malformations

66.2.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.29, 95% CI 0.06 to 1.51, I² = 0%), with no difference in the number of neural tube malformations in children exposed to VPA (N = 1237) and children exposed to ZNS (N = 27) (Analysis 66.2). The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI −0.11 to 0.06; I² = 0%).

66.2.2 Routine health record studies

No included studies reported data on this outcome.

66.3 Cardiac malformations

66.3.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.49, 95% CI 0.07 to 3.65, I² = 0%), with no difference in the number of cardiac malformations in children exposed to VPA (N = 1237) and children exposed to ZNS (N = 27) (Analysis 66.3). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.05 to 0.08; I² = 0%).

66.3.2 Routine health record studies

No included studies reported data on this outcome.

66.4 Oro‐facial cleft/craniofacial malformations

66.4.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.47, 95% CI 0.06 to 3.49, I² = 0%), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to VPA (N = 1237) and children exposed to ZNS (N = 27) (Analysis 66.4). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.05 to 0.08; I² = 0%).

66.4.2 Routine health record studies

No included studies reported data on this outcome.

66.5 Skeletal/limb malformations

66.5.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.46, 95% CI 0.03 to 7.72, I² = NA), with no difference in the number of skeletal/limb malformations in children exposed to VPA (N = 1237) and children exposed to ZNS (N = 27) (Analysis 66.5). The RD also suggested no difference in the level of risk (RD 0.01, 95% CI −0.06 to 0.07; I² = 0%).

66.5.2 Routine health record studies

No included studies reported data on this outcome.

67 CZP vs VPA

67.1 All major malformations

67.1.1 Cohort studies

Pooled results from four cohort studies suggested an increased risk with VPA (RR 0.29, 95% CI 0.09 to 0.90; I² = 0%), with children exposed to VPA (N = 955) experiencing more major malformations than children exposed to CZP (N = 95) (Analysis 67.1). The RD also suggested a higher risk for VPA (RD −0.09, 95% CI −0.13 to ‐0.04; I² = 30%).

67.1.2 Routine health record studies

Pooled results from two routine health record studies suggested an increased risk with VPA (RR 0.34, 95% CI 0.13 to 0.94; I² = 0%), with children exposed to VPA (N = 601) experiencing more major malformations than children exposed to CZP (N = 161) (Analysis 67.1). The RD suggested no difference in the level of risk (RD −0.05, 95% CI −0.12 to 0.01; I² = 0%).

67.2 Neural tube malformations

67.2.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 9.77, 95% CI 0.58 to 165.35, I² = NA), with no difference in the number of neural tube malformations in children exposed to CZP (N = 4) and children exposed to VPA (N = 341) (Analysis 67.2). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.27 to 0.25; I² = NA).

67.2.2 Routine health record studies

No included studies reported data on this outcome.

67.3 Cardiac malformations

67.3.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 1.67, 95% CI 0.12 to 23.92, I² = NA), with no difference in the number of cardiac malformations in children exposed to CZP (N = 4) and children exposed to VPA (N = 341) (Analysis 67.3). The RD also suggested no difference in the level of risk (RD ‐0.06, 95% CI −0.32 to 0.21; I² = NA).

67.3.2 Routine health record studies

No included studies reported data on this outcome.

67.4 Skeletal/limb malformations

67.4.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 7.60, 95% CI 0.47 to 123.14, I² = NA), with no difference in the number of skeletal/limb malformations in children exposed to CZP (N = 4) and children exposed to VPA (N = 341) (Analysis 67.4). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI −0.27 to 0.25; I² = NA).

67.4.2 Routine health record studies

No included studies reported data on this outcome.

67.5 Oro‐facial cleft/craniofacial malformations

67.5.1 Cohort studies

No included studies reported data on this outcome.

67.5.2 Routine health record studies

No included studies reported data on this outcome.

68 CZP versus LEV

68.1 All major malformations

68.1.1 Cohort studies

Pooled results from three cohort studies suggested no evidence of a difference in risk (RR 1.06, 95% CI 0.32 to 3.44; I² = 0%), with children exposed to CZP (N = 94) experiencing more major malformations than children exposed to LEV (N = 695) (Analysis 68.1). The RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.05 to 0.03; I² = 0%).

68.1.2 Routine health record studies

Results from one routine health record study suggested no evidence of a difference in risk (RR 1.04, 95% CI 0.15 to 7.29; I² = NA), with no difference in the number of major malformations in children exposed to CZP (N = 113) and children exposed to LEV (N = 118) (Analysis 68.1). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI −0.03 to 0.03; I² = 0%).

68.2 Neural tube malformations

68.2.1 Cohort studies

No included studies reported data on this outcome.

68.2.2 Routine health record studies

No included studies reported data on this outcome.

68.3 Cardiac malformations

68.3.1 Cohort studies

No included studies reported data on this outcome.

68.3.2 Routine health record studies

No included studies reported data on this outcome.

68.4 Oro‐facial cleft/craniofacial malformations

68.4.1 Cohort studies

No included studies reported data on this outcome.

68.4.2 Routine health record studies

No included studies reported data on this outcome.

68.5 Skeletal/limb malformations

68.5.1 Cohort studies

No included studies reported data on this outcome.

68.5.2 Routine health record studies

No included studies reported data on this outcome.

69 OXC versus PRM

69.1 All major malformations

69.1.1 Cohort studies

Pooled results from two cohort studies suggested no evidence of a difference in risk (RR 0.58, 95% CI 0.08 to 4.03; I² = 0%), with no difference in the number of major malformations in children exposed to OXC (N = 28) and children exposed to PRM (N = 8) (Analysis 69.1). The RD also suggested no difference in the level of risk (RD −0.02, 95% CI −0.34 to 0.30; I² = 0%).

69.1.2 Routine health record studies

We were unable to estimate the RR for one routine health record study due to there being no reported major malformations observed in children exposed to OXC (N = 4) or PRM (N = 3) (Analysis 69.1).

69.2 Neural tube malformations

69.2.1 Cohort studies

No included studies reported data on this outcome.

69.2.2 Routine health record studies

No included studies reported data on this outcome.

69.3 Cardiac malformations

69.3.1 Cohort studies

No included studies reported data on this outcome.

69.3.2 Routine health record studies

No included studies reported data on this outcome.

69.4 Oro‐facial cleft/craniofacial malformations

69.4.1 Cohort studies

No included studies reported data on this outcome.

69.4.2 Routine health record studies

No included studies reported data on this outcome.

69.5 Skeletal/limb malformations

69.5.1 Cohort studies

No included studies reported data on this outcome.

69.5.2 Routine health record studies

No included studies reported data on this outcome.

70 OXC versus TPM

70.1 All major malformations

70.1.1 Cohort studies

Pooled results from five cohort studies suggested no evidence of a difference in risk (RR 0.71, 95% CI 0.28 to 1.77; I² = 0%), with no difference in the number of major malformations in children exposed to OXC (N = 279) and children exposed to TPM (N = 427) (Analysis 70.1). The RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.04 to 0.02; I² = 0%).

The EURAP 2018 collaboration reported the prevalence of MCM was 3.0% (95% CI 1.4 to 5.4) for children exposed to OXC and 3.9% (95% CI 1.5 to 8.4) for children exposed to TPM. No direct statistical comparison was made at the group level; investigations were made across different doses of the two ASMs (see oxcarbazepine dose and topiramate dose sections).

70.1.2 Routine health record studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 70.1). However, available data showed there were 1/61 cases of major malformations in children exposed to OXC and 2/49 cases in children exposed to TPM, based on data from two studies (Norwegian Health Record Registers; Sweden Health Record Registers).

70.2 Neural tube malformations

70.2.1 Cohort studies

We could not estimate a RR from four cohort studies due to there being no reported neural tube malformations in children exposed to OXC (N = 266) or children exposed to TPM (N = 418) (Analysis 70.2).

In the EURAP 2018 data, the prevalence of neural tube anomalies in those exposed to OXC was 0% (0/333) and 0% for children exposed to TPM (0/152).

70.2.2 Routine health record studies

No included studies reported data on this outcome.

70.3 Cardiac malformations

70.3.1 Cohort studies

Pooled results from five cohort studies suggested no evidence of a difference in risk (RR 0.80, 95% CI 0.09 to 6.81; I² = 0%), with no difference in the number of cardiac malformations in children exposed to OXC (N = 269) and children exposed to TPM (N = 419) (Analysis 70.3). The RD also suggested no difference in the level of risk (RD 0.00, 95% CI ‐0.02 to 0.02; I² = 0%).

In the EURAP 2018 data, the prevalence of cardiac anomalies in those exposed to OXC was 1.20% (4/333) and 1.97% for children exposed to TPM (3/152).

70.3.2 Routine health record studies

No included studies reported data on this outcome.

70.4 Oro‐facial cleft/craniofacial malformations

70.4.1 Cohort studies

Pooled results from four studies suggested no evidence of a difference in risk (RR 0.39, 95% CI 0.05 to 3.35, I² = NA), with no difference in the number of oro‐facial cleft/craniofacial malformations in children exposed to OXC (N = 198) and children exposed to TPM (N = 410) (Analysis 70.4). The RD also suggested no difference in the level of risk (RD ‐0.01, 95% CI ‐0.03 to 0.01; I² = 0%).

In the EURAP 2018 data, the prevalence of cleft malformations (other oro‐facial not specifically reported) in those exposed to OXC was 0.30% (1/333) and 0% for children exposed to TPM (0/152).

70.4.2 Routine health record studies

No included studies reported data on this outcome.

70.5 Skeletal/limb malformations

70.5.1 Cohort studies

Pooled results from four cohort studies suggested no evidence of a difference in risk (RR 0.40, 95% CI 0.07 to 2.44; I² = 0%), with no difference in the number of skeletal/limb malformations in children exposed to OXC (N = 266) and children exposed to TPM (N = 418) (Analysis 70.5). The RD also suggested no difference in the level of risk (RD −0.01, 95% CI −0.03 to 0.01; I² = 0%).

70.5.2 Routine health record studies

No included studies reported data on this outcome.

71 OXC versus ZNS

71.1 All major malformations

71.1.1 Cohort studies

Pooled results from two cohort studies suggested no evidence of a difference in risk (RR 4.48, 95% CI 0.24 to 82.23; I² = NA), with no difference in the number of major malformations in children exposed to OXC (N = 186) and children exposed to ZNS (N = 91) ( Analysis 71.1). The RD also suggested no difference in the level of risk (RD 0.02, 95% CI −0.01 to 0.05; I² = 0%).

71.1.2 Routine health record studies

No included studies reported data on this outcome.

71.2 Neural tube malformations

71.2.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported cases of neural tube malformations in children exposed to OXC (N = 4) or ZNS (N = 1) (Analysis 71.2).

71.2.2 Routine health record studies

No included studies reported data on this outcome.

71.3 Cardiac malformations

71.3.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported cases of cardiac malformations in children exposed to OXC (N = 4) or ZNS (N = 1) (Analysis 71.3).

71.3.2 Routine health record studies

No included studies reported data on this outcome.

71.4 Oro‐facial cleft/craniofacial malformations

71.4.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported cases of oro‐facial cleft/ craniofacial malformations in children exposed to OXC (N = 4) or ZNS (N = 1) (Analysis 71.4).

71.4.2 Routine health record studies

No included studies reported data on this outcome.

71.5 Skeletal/limb malformations

71.5.1 Cohort studies

We were unable to estimate a RR from one study due to there being no reported cases of skeletal/limb malformations in children exposed to OXC (N = 4) or ZNS (N = 1) (Analysis 71.5).

71.5.2 Routine health record studies

No included studies reported data on this outcome.

72 PRM versus TPM

72.1 All major malformations

72.1.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 6.00, 95% CI 0.30 to 118.36, I² = NA), with no difference in the number of major malformations in children exposed to PRM (N = 2) and children exposed to TPM (N = 53) (Analysis 72.1). The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI −0.44 to 0.41; I² = NA).

72.1.2 Routine health record studies

We were unable to estimate a RR from one study due to there being no reported cases of major malformations in children exposed to PRM (N = 3) or TPM (N = 1) (Analysis 72.1).

72.2 Neural tube malformations

72.2.1 Cohort studies

No included studies reported data on this outcome.

72.2.2 Routine health record studies

No included studies reported data on this outcome.

72.3 Cardiac malformations

72.3.1 Cohort studies

No included studies reported data on this outcome.

72.3.2 Routine health record studies

No included studies reported data on this outcome.

72.4 Oro‐facial cleft/craniofacial malformations

72.4.1 Cohort studies

No included studies reported data on this outcome.

72.4.2 Routine health record studies

No included studies reported data on this outcome.

72.5 Skeletal/limb malformations

72.5.1 Cohort studies

No included studies reported data on this outcome.

72.5.2 Routine health record studies

No included studies reported data on this outcome.

73 PRM versus VPA

73.1 All major malformations

73.1.1 Cohort studies

Pooled results from six cohort studies suggested no evidence of a difference in risk (RR 0.74, 95% CI 0.39 to 1.40; I² = 21%), with no difference in the number of major malformations in children exposed to PRM (N = 103) and children exposed to VPA (N = 491), (Analysis 73.1). The RD also suggested no difference in the level of risk (RD −0.04, 95% CI −0.13 to 0.04; I² = 1%).

73.1.2 Routine health record studies

Results from one study suggested no evidence of a difference in risk (RR 1.27, 95% CI 0.09 to 17.39, I² = NA), with no difference in the number of major malformations in children exposed to PRM (N = 3) and children exposed to VPA (N = 268) (Analysis 73.1). The RD also suggested no difference in the level of risk (RD ‐0.10, 95% CI −0.42 to 0.23; I² = NA).

73.2 Neural tube malformations

73.2.1 Cohort studies

Included studies did not reach the threshold for reporting of the meta‐analysis (Analysis 73.2). However, available data showed there were 0/39 cases of neural tube malformations in children exposed to PRM and 5/45 cases in children exposed to VPA, based on data from two studies (Milan Study 1999; Pardi 1982).

73.2.2 Routine health record studies

No included studies reported data on this outcome.

73.3 Cardiac malformations

73.3.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 73.3). However, available data showed there were 1/39 cases of cardiac malformations in children exposed to PRM and 0/45 cases in children exposed to VPA, based on data from two studies (Milan Study 1999; Pardi 1982).

73.3.2 Routine health record studies

No included studies reported data on this outcome.

73.4 Oro‐facial cleft/craniofacial malformations

73.4.1 Cohort studies

We were unable to estimate a RR from two studies due to there being no reported oro‐facial cleft/craniofacial malformations in children exposed to PRM (N = 39) or VPA (N = 45). (Analysis 73.4).

73.4.2 Routine health record studies

No included studies reported data on this outcome.

73.5 Skeletal/limb malformations

73.5.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 73.5). However, available data showed there were 0/39 cases of skeletal/limb malformations in children exposed to PRM and 1/45 cases in children exposed to VPA, based on data from two studies (Milan Study 1999; Pardi 1982).

73.5.2 Routine health record studies

No included studies reported data on this outcome.

74 LEV vs LAC

74.1 All major malformations

74.1.1 Cohort studies

We were unable to estimate a RR for one cohort study due to there being no major malformations observed in children exposed to LEV (N = 12) or LAC (N = 1) (Analysis 74.1).

74.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

75 CBZ versus LAC

75.1 All major malformations

75.1.1 Cohort studies

We could not estimate RR from one cohort study as there were no malformations observed in children exposed to CBZ (N = 7) and children exposed to LAC (N = 1) (Analysis 75.1).

75.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

76 OXC vs LAC

76.1 All major malformations

76.1.1 Cohort studies

We were unable to estimate the RR for one cohort study due to there being no major malformations observed in children exposed to OXC (N = 4) or LAC (N = 1) (Analysis 76.1).

76.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

77 PB versus LAC

77.1 All major malformations

77.1.1 Cohort studies

We were unable to estimate a RR from one cohort study due to there being no malformations observed in children exposed to PB (N = 2) and children exposed to LAC (N = 1) (Analysis 77.1).

77.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

78 LAC vs ZNS

78.1 All major malformations

78.1.1 Cohort studies

We were unable to estimate a RR from one cohort study due to there being no major malformations observed in children exposed to LAC (N = 1) or ZNS (N = 1) (Analysis 78.1).

78.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

79 GPB versus PGB

79.1 All major malformations

79.1.1 Cohort studies

We could not estimate RR from one cohort study as there were no malformations observed in children exposed to GPB (N = 14) and children exposed to PGB (N = 1) (Analysis 79.1).

79.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

80 GBP vs CZP

80.1 All major malformations

80.1.1 Cohort studies

No included studies reported data on this outcome.

80.1.2 Routine health record studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 80.1). However, available data showed there were 0/18 cases of major malformations in children exposed to GBP and 2/48 cases in children exposed to CZP from one study (Sweden Health Record Registers).

Due to limited numbers, we did not investigate specific malformation types.

81 VPA vs BNZ

81.1 All major malformations

81.1.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 81.1). However, available data showed there were 4/44 cases of major malformations in children exposed to VPA and 0/5 cases in children exposed to BNZ from two studies (Jimenez 2020; Melikova 2020).

81.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

82 LTG versus BNZ

82.1 All major malformations

82.1.1 Cohort studies

We were unable to estimate a RR from two cohort studies due to there being no malformations observed in children exposed to LTG (N = 26) and children exposed to BNZ (N = 5) (Analysis 82.1).

82.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

83 LEV versus BNZ

83.1 All major malformations

83.1.1 Cohort studies

We were unable to estimate a RR from two cohort studies due to there being no malformations observed in children exposed to LEV (N = 18) and children exposed to BNZ (N = 5) (Analysis 83.1).

83.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

84 CBZ vs BNZ

84.1 All major malformations

84.1.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 84.1). However, available data showed there were 1/43 cases of major malformations in children exposed to CBZ and 0/5 cases in children exposed to BNZ, based on data from two studies (Jimenez 2020; Melikova 2020).

84.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

85 OXC versus BNZ

85.1 All major malformations

85.1.1 Cohort studies

We were unable to estimate a RR from one cohort study due to there being no malformations observed in children exposed to OXC (N = 4) and children exposed to BNZ (N = 2) (Analysis 85.1).

85.1.2 Routine health record studies

No included studies reported data on this outcome.

86 PB vs BNZ

86.1 All major malformations

86.1.1 Cohort studies

We were unable to estimate a RR for one cohort study due to there being no major malformations observed in children exposed to PB (N = 2) or BNZ (N = 2) (Analysis 86.1).

86.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

87 LAC versus BNZ

87.1 All major malformations

87.1.1 Cohort studies

We were unable to estimate a RR from one cohort study due to there being no malformations observed in children exposed to OXC (N = 4) and children exposed to BNZ (N = 2) (Analysis 87.1).

87.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

88 ZNS vs BNZ

88.1 All major malformations

88.1.1 Cohort studies

We were unable to estimate a RR for one cohort study due to there being no major malformations observed in children exposed to ZNS (N = 1) or BNZ (N = 2) (Analysis 88.1).

88.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

89 CZP versus TPM

89.1 All major malformations

89.1.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.67, 95% CI 0.07 to 1.87, I² = NA), with no difference in the number of major malformations in children exposed to CZP (N = 26) and children exposed to TPM (N = 53) (Analysis 89.1). The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI −0.09 to 0.05; I² = NA).

89.1.2 Routine health record studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.37, 95% CI 0.03 to 15.83, I² = 0%), with no difference in the number of major malformations in children exposed to CZP (N = 26) and children exposed to TPM (N = 53) (Analysis 89.1). The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI −0.09 to 0.05; I² = 0%).

Due to limited numbers, we did not investigate specific malformation types.

90 CZP vs OXC

90.1 All major malformations

90.1.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 90.1). However, available data showed there were 0/26 cases of major malformations in children exposed to CZP and 1/19 cases in children exposed to OXC, based on data from one study (Australian Epilepsy and Pregnancy Register).

90.1.2 Routine health record studies

Pooled results from two routine health record studies suggested no evidence of a difference in risk (RR 0.81, 95% CI 0.13 to 5.06; I² = 0%), with no difference between children exposed to CZP (N = 161) and children exposed to OXC (N = 61) (Analysis 90.1). The RD also suggested no difference in the level of risk (RD ‐0.04, 95% CI −0.04 to 0.05; I² = 0%).

Due to limited numbers, we did not investigate specific malformation types.

91 CZP versus COZ

91.1 All major malformations

91.1.1 Cohort studies

We were unable to estimate a RR for one cohort study due to there being no major malformations observed in children exposed to CZP (N = 26) or COZ (N = 2) (Analysis 91.1).

91.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

92 CZP vs ESM

92.1 All major malformations

92.1.1 Cohort studies

We were unable to estimate the RR for one cohort study due to there being no reported major malformations observed in children exposed to CZP (N = 48) or ESM (N = 8) (Analysis 92.1).

92.1.2 Routine health record studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 92.1). However, available data showed there were 2/48 cases of major malformations in children exposed to CZP and 0/8 cases in children exposed to ESM, based on data from one study (Sweden Health Record Registers).

Due to limited numbers, we did not investigate specific malformation types.

93 CZP versus PRG

93.1 All major malformations

93.1.1 Cohort studies

We were unable to estimate a RR for one cohort study due to there being no major malformations observed in children exposed to CZP (N = 26) or children exposed to PRG (N = 1) (Analysis 93.1).

93.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

94 CZP vs PRM

94.1 All major malformations

94.1.1 Cohort studies

We were unable to estimate a RR for one cohort study due to there being no major malformations observed in children exposed to CZP (N = 26) or PRM (N = 2) (Analysis 94.1).

94.1.2 Routine health record studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 94.1). However, available data showed there were 2/48 cases of major malformations in children exposed to CZP and 0/3 cases in children exposed to PRM, based on data from one study (Sweden Health Record Registers).

Due to limited numbers, we did not investigate specific malformation types.

95 CZP versus VGB

95.1 All major malformations

95.1.1 Cohort studies

We were unable to estimate RR for one cohort study due to there being no major malformations observed in children exposed to CZP (N = 26) or VGB (N = 1) (Analysis 95.1).

95.1.2 Routine health record studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 95.1). However, available data showed there were 2/48 cases of major malformations in children exposed to CZP and 0/3 cases in children exposed to VGB, based on data from one study (Sweden Health Record Registers).

Due to limited numbers, we did not investigate specific malformation types.

96 TPM vs BNZ

96.1 All major malformations

96.1.1 Cohort studies

We were unable to estimate a RR for two cohort studies due to there being no major malformations observed in children exposed to TPM (N = 7) or BNZ (N = 5) (Analysis 96.1).

96.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

97 ESM versus VPA

97.1 All major malformations

97.1.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 0.56, 95% CI 0.04 to 8.03, I² = NA), with no difference in the number of major malformations in children exposed to ESM (N = 5) and children exposed to VPA (N = 290) (Analysis 97.1). The RD also suggested no difference in the level of risk (RD ‐0.15, 95% CI −0.37 to 0.08; I² = NA).

97.1.2 Routine health record studies

Results from one study suggested no evidence of a difference in risk (RR 0.56, 95% CI 0.04 to 8.84, I² = NA), with no difference in the number of major malformations in children exposed to ESM (N = 8) and children exposed to VPA (N = 268) (Analysis 97.1). The RD also suggested no difference in the level of risk (RD ‐0.10, 95% CI −0.25 to 0.06; I² = NA).

Due to limited numbers, we did not investigate specific malformation types.

98 ESM vs CBZ

98.1 All major malformations

98.1.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 1.39, 95% CI 0.10 to 20.37, I² = NA), with no difference in the number of major malformations in children exposed to ESM (N = 4) and children exposed to CBZ (N = 409) (Analysis 98.1). The RD also suggested no difference in the level of risk (RD ‐0.06, 95% CI −0.28 to 0.16; I² = NA).

98.1.2 Routine health record studies

Results from one study suggested no evidence of a difference in risk (RR 1.37, 95% CI 0.09 to 20.78, I² = NA), with no difference in the number of major malformations in children exposed to ESM (N = 8) and children exposed to CBZ (N = 703) (Analysis 98.1). The RD also suggested no difference in the level of risk (RD ‐0.04, 95% CI −0.19 to 0.11; I² = NA).

Due to limited numbers, we did not investigate specific malformation types.

99 ESM versus PRM

99.1 All major malformations

99.1.1 Cohort studies

We were unable to estimate a RR for one cohort study due to there being no major malformations observed in children exposed to ESM (N = 5) or PRM (N = 2) (Analysis 99.1).

99.1.2 Routine health record studies

We were unable to estimate a RR for one routine health record study due to there being no major malformations observed in children exposed to ESM (N = 8) or PRM (N = 3) (Analysis 99.1).

Due to limited numbers, we did not investigate specific malformation types.

100 ESM vs PB

100.1 All major malformations

100.1.1 Cohort studies

We were unable to estimate the RR for one cohort study due to there being no reported major malformations observed in children exposed to ESM (N = 5) or PB (N = 2) (Analysis 100.1).

100.1.2 Routine health record studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 100.1). However, available data showed there were 0/8 cases of major malformations in children exposed to ESM and 1/7 cases in children exposed to PB, based on data from one study (Sweden Health Record Registers).

Due to limited numbers, we did not investigate specific malformation types.

101 ESM versus PHT

101.1 All major malformations

101.1.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 101.1). However, available data showed there were 0/5 cases of major malformations in children exposed to ESM and 1/44 cases in children exposed to PHT, based on data from one study (Australian Epilepsy and Pregnancy Register).

101.1.2 Routine health record studies

Results from one study suggested no evidence of a difference in risk (RR 0.77, 95% CI 0.05 to 12.42, I² = NA), with no difference in the number of major malformations in children exposed to ESM (N = 8) and children exposed to PHT (N = 103) (Analysis 101.1). The RD also suggested no difference in the level of risk (RD ‐0.07, 95% CI −0.23 to 0.09; I² = NA).

Due to limited numbers, we did not investigate specific malformation types.

102 ESM vs OXC

102.1 All major malformations

102.1.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 102.1). However, available data showed there were 0/5 cases of major malformations in children exposed to ESM and 1/19 cases in children exposed to OXC, based on data from one study (Australian Epilepsy and Pregnancy Register).

102.1.2 Routine health record studies

We were unable to estimate the RR for one routine health record study due to there being no reported no major malformations observed in children exposed to ESM (N = 8) or OXC (N = 4) (Analysis 102.1).

Due to limited numbers, we did not investigate specific malformation types.

103 ESM versus VGB

103.1 All major malformations

103.1.1 Cohort studies

We were unable to estimate a RR for one cohort study due to there being no major malformations observed in children exposed to ESM (N = 5) or VBG (N = 1) (Analysis 103.1).

103.1.2 Routine health record studies

We were unable to estimate a RR for one routine health record study due to there being no major malformations observed in children exposed to ESM (N = 8) or VGB (N = 3) (Analysis 103.1).

Due to limited numbers, we did not investigate specific malformation types.

104 ESM vs LTG

104.1 All major malformations

104.1.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 1.65, 95% CI 0.11 to 24.30, I² = NA), with no difference in the number of major malformations in children exposed to ESM (N = 5) and children exposed to LTG (N = 406) (Analysis 104.1). The RD also suggested no difference in the level of risk (RD ‐0.05, 95% CI −0.27 to 0.17; I² = NA).

104.1.2 Routine health record studies

Results from one study suggested no evidence of a difference in risk (RR 1.12, 95% CI 0.07 to 19.24, I² = NA), with no difference in the number of major malformations in children exposed to ESM (N = 8) and children exposed to LTG (N = 90) (Analysis 104.1). The RD also suggested no difference in the level of risk (RD ‐0.04, 95% CI −0.20 to 0.11; I² = NA).

Due to limited numbers, we did not investigate specific malformation types.

105 ESM versus TPM

105.1 All major malformations

105.1.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 3.00, 95% CI 0.14 to 65.77, I² = NA), with no difference in the number of major malformations in children exposed to ESM (N = 8) and children exposed to TPM (N = 53) (Analysis 105.1) The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI −0.24 to 0.21; I² = NA).

105.1.2 Routine health record studies

We were unable to estimate the RR for one routine health record study due to there being no major malformations observed in children exposed to ESM (N = 8) or TPM (N = 1) (Analysis 105.1).

Due to limited numbers, we did not investigate specific malformation types.

106 ESM vs GBP

106.1 All major malformations

106.1.1 Cohort studies

No included studies reported data on this outcome.

106.1.2 Routine health record studies

We were unable to estimate the RR for one routine health record study due to there being no reported major malformations observed in children exposed to ESM (N = 8) or GBP (N = 18) (Analysis 106.1).

Due to limited numbers, we did not investigate specific malformation types.

107 VGB versus VPA

107.1 All major malformations

107.1.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 1.67, 95% CI 0.15 to 18.73, I² = NA), with no difference in the number of major malformations in children exposed to VGB (N = 1) and children exposed to VPA (N = 290) (Analysis 107.1). The RD also suggested no difference in the level of risk (RD ‐0.15, 95% CI −0.75 to 0.45; I² = NA).

107.1.2 Routine health record studies

Results from one study suggested no evidence of a difference in risk (RR 1.27, 95% CI 0.09 to 17.39, I² = NA), with no difference in the number of major malformations in children exposed to VGB (N = 3) and children exposed to VPA (N = 268) (Analysis 107.1). The RD also suggested no difference in the level of risk (RD ‐0.10, 95% CI −0.42 to 0.23; I² = NA).

Due to limited numbers, we did not investigate specific malformation types.

108 VGB vs CBZ

108.1 All major malformations

108.1.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 4.18, 95% CI 0.37 to 47.57, I² = NA), with no difference in the number of major malformations in children exposed to VGB (N = 1) and children exposed to CBZ (N = 409) (Analysis 108.1). The RD also suggested no difference in the level of risk (RD ‐0.06, 95% CI −0.66 to 0.54; I² = NA).

108.1.2 Routine health record studies

Results from one study suggested no evidence of a difference in risk (RR 3.09, 95% CI 0.23 to 42.31, I² = NA), with no difference in the number of major malformations in children exposed to VGB (N = 3) and children exposed to CBZ (N = 703) (Analysis 108.1). The RD also suggested no difference in the level of risk (RD ‐0.04, 95% CI −0.36 to 0.28; I² = NA).

Due to limited numbers, we did not investigate specific malformation types.

109 VGB versus PRM

109.1 All major malformations

109.1.1 Cohort studies

We were unable to estimate a RR for one cohort study due to there being no major malformations observed in children exposed to VGB (N = 1) or PRM (N = 2) (Analysis 109.1).

109.1.2 Routine health record studies

We were unable to estimate a RR for one routine health record study due to there being no major malformations observed in children exposed to VGB (N = 3) or PRM (N = 3) (Analysis 109.1).

Due to limited numbers, we did not investigate specific malformation types.

110 VGB vs PB

110.1 All major malformations

110.1.1 Cohort studies

No included studies reported data on this outcome.

110.1.2 Routine health record studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 110.1). However, available data showed there were 0/3 cases of major malformations in children exposed to VGB and 1/7 cases in children exposed to PB, based on data from one study (Sweden Health Record Registers).

Due to limited numbers, we did not investigate specific malformation types.

111 VGB versus PHT

111.1 All major malformations

111.1.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 111.1). However, available data showed there were 0/1 cases of major malformations in children exposed to VGB and 1/44 cases in children exposed to PHT, based on data from one study (Australian Epilepsy and Pregnancy Register).

111.1.2 Routine health record studies

Results from one study suggested no evidence of a difference in risk (RR 1.73, 95% CI 0.13 to 25.35, I² = NA), with no difference in the number of major malformations in children exposed to VGB (N = 3) and children exposed to PHT (N = 103) (Analysis 111.1). The RD also suggested no difference in the level of risk (RD ‐0.07, 95% CI −0.40 to 0.28; I² = NA).

Due to limited numbers, we did not investigate specific malformation types.

112 VGB vs OXC

112.1 All major malformations

112.1.1 Cohort studies

Included studies did not meet the threshold for reporting of the meta‐analysis (Analysis 112.1). However, available data showed there were 0/1 cases of major malformations in children exposed to VGB and 1/19 cases in children exposed to OXC, based on data from one study (Australian Epilepsy and Pregnancy Register).

112.1.2 Routine health record studies

We were unable to estimate the RR for one routine health record study due to there being no reported major malformations observed in children exposed to VGB (N = 3) or OXC (N = 4) (Analysis 112.1).

Due to limited numbers, we did not investigate specific malformation types.

113 VGB versus LTG

113.1 All major malformations

113.1.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 3.31, 95% CI 0.25 to 43.03, I² = NA), with no difference in the number of major malformations in children exposed to VGB (N = 2) and children exposed to LTG (N = 406) (Analysis 113.1). The RD also suggested no difference in the level of risk (RD ‐0.05, 95% CI −0.47 to 0.37; I² = NA).

113.1.2 Routine health record studies

Results from one study suggested no evidence of a difference in risk (RR 2.53, 95% CI 0.16 to 39.34, I² = NA), with no difference in the number of major malformations in children exposed to VGB (N = 3) and children exposed to LTG (N = 90) (Analysis 113.1). The RD also suggested no difference in the level of risk (RD ‐0.04, 95% CI −0.37 to 0.28; I² = NA).

Due to limited numbers, we did not investigate specific malformation types.

114 VGB vs TPM

114.1 All major malformations

114.1.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 9.00, 95% CI 0.51 to 159.15, I² = NA), with no difference in the number of major malformations in children exposed to VGB (N = 1) and children exposed to TPM (N = 53) (Analysis 114.1). The RD also suggested no difference in the level of risk (RD ‐0.02, 95% CI −0.62 to 0.58; I² = NA).

114.1.2 Routine health record studies

We were unable to estimate the RR for one routine health record study due to there being no reported major malformations observed in children exposed to VGB (N = 3) or TPM (N = 1) (Analysis 114.1).

Due to limited numbers, we did not investigate specific malformation types.

115 VGB vs GBP

115.1 All major malformations

115.1.1 Cohort studies

No included studies reported data on this outcome.

115.1.2 Routine health record studies

We were unable to estimate the RR for one routine health record study due to there being no reported no major malformations observed in children exposed to VGB (N = 3) or GBP (N = 18) (Analysis 115.1).

Due to limited numbers, we did not investigate specific malformation types.

116 CZP vs PB

116.1 All major malformations

116.1.1 Cohort studies

Included studies did not meet the threshold for inclusion in the meta‐analysis (Analysis 116.1). However, available data showed there were 0/27 cases of major malformations in children exposed to CZP and 1/6 cases in children exposed to PB, based on data from two studies (Australian Epilepsy and Pregnancy Register; D'Souza 1991).

116.1.2 Routine health record studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.26, 95% CI 0.06 to 1.12, I² = 0%), with no difference in the number of major malformations in children exposed to CZP (N = 161) and children exposed to PB (N = 34) (Analysis 116.1). The RD also suggested no difference in the level of risk (RD ‐0.07, 95% CI −0.16 to 0.03; I² = 0%).

Due to limited numbers, we did not investigate specific malformation types.

117 CZP vs PHT

117.1 All major malformations

117.1.1 Cohort studies

Pooled results from two studies suggested no evidence of a difference in risk (RR 0.71, 95% CI 0.10 to 5.11, I² = 0%), with no difference in the number of major malformations in children exposed to CZP (N = 27) and children exposed to PHT (N = 66) (Analysis 117.1). The RD also suggested no difference in the level of risk (RD ‐0.04, 95% CI −0.13 to 0.06; I² = 0%).

117.1.2 Routine health record studies

Results from one study suggested no evidence of a difference in risk (RR 0.61, 95% CI 0.13 to 2.48, I² = NA), with no difference in the number of major malformations in children exposed to CZP (N = 48) and children exposed to PHT (N = 103) (Analysis 117.1). The RD also suggested no difference in the level of risk (RD ‐0.03, 95% CI −0.10 to 0.05; I² = NA).

Due to limited numbers, we did not investigate specific malformation types.

118 ESM vs LEV

118.1 All major malformations

118.1.1 Cohort studies

Results from one study suggested no evidence of a difference in risk (RR 2.12, 95% CI 0.13 to 34.10, I² = NA), with no difference in the number of major malformations in children exposed to ESM (N = 5) and children exposed to LEV (N = 139) (Analysis 118.1) The RD also suggested no difference in the level of risk (RD ‐0.04, 95% CI −0.26 to 0.19; I² = NA).

118.1.2 Routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

119 ESM vs Controls

119.1 All major malformations

119.1.1 ESM versus no medication (in women without epilepsy): cohort studies

No included studies reported data on this outcome.

119.1.2 ESM versus no medication (in women with epilepsy): cohort studies

Results from one study suggested no evidence of a difference in risk (RR 2.68, 95% CI 0.17 to 43.16, I² = NA), with no difference in the number of major malformations in children exposed to ESM (N = 5) and control children (N = 176) (Analysis 119.1). The RD also suggested no difference in the level of risk (RD ‐0.03, 95% CI −0.25 to 0.19; I² = NA).

119.1.3 ESM versus no medication (in women without epilepsy): routine health record studies

No included studies reported data on this outcome.

119.1.4 ESM versus no medication (in women with epilepsy): routine health record studies

No included studies reported data on this outcome.

120 VGB vs Control

120.1 All major malformations

120.1.1 VGB versus no medication (in women without epilepsy): cohort studies

No included studies reported data on this outcome.

120.1.2 VGB versus no medication (in women with epilepsy): cohort studies

Results from one study suggested no evidence of a difference in risk (RR 8.05, 95% CI 0.64 to 101.76, I² = NA), with no difference in the number of major malformations in children exposed to VGB (N = 1) and control children (N = 176) (Analysis 120.1). The RD also suggested no difference in the level of risk (RD ‐0.03, 95% CI −0.63 to 0.57; I² = NA).

120.1.3 VGB versus no medication (in women without epilepsy): routine health record studies

No included studies reported data on this outcome.

120.1.4 VGB versus no medication (in women with epilepsy): routine health record studies

No included studies reported data on this outcome.

Due to limited numbers, we did not investigate specific malformation types.

Studies not included in the meta‐analysis and not narratively reported

The publications of EURAP 2018; Samren 1997 required narrative reporting due to their overlap with other research initiatives. Israeli Teratogen Service showed variability in its reporting and, therefore, required narrative reporting for certain outcomes. Further, studies using US Medicaid Registers also required narrative review due to the format of reporting of the monotherapy TPM and GBP pregnancies in women with epilepsy.

Discussion

We reported results from three study types, meta‐analysis of data from cohort studies, data from EURAP 2018 and others, and meta‐analysis of data from epidemiological health record studies. Each study design has its inherent methodological strengths and weaknesses. We undertook a stratified approach to evidence synthesis to ensure a sensitive approach to combining data and to allow for the development of evidence groupings; this will allow for replication of findings across different study types and will also allow for increased confidence in the evidence .

The meta‐analyses included 17,964 ASM‐exposed pregnancies from cohort studies and 7913 from routine health record studies; additional exposed pregnancies from the EURAP collaboration and other studies were reviewed narratively. Individual ASM prevalence of major malformation ranged from 2.0% to 9.8% for cohort study data and 3.6% to 9.7% for studies utilising routine health record data (Table 3, Figure 3). Table 1 and Table 2 along with Table 3, Table 4, Table 5, Table 6, Table 7, Table 8 and Table 9 provide a summary of the meta‐analysis results for all comparisons for risk of major malformation. This update has included the most recent study data, which strengthens the previously identified risk associations for older ASMs such as CBZ, PB and VPA, where the comparisons include large numbers of exposed pregnancies. For other ASMs, there are some differences in the results across the comparisons, but the better powered comparisons demonstrate an increased risk for PHT and for an overall major malformation risk with CBZ and for TPM.

Whilst CBZ, PB, PHT, TPM and VPA are associated with an increase in the risk of major malformation, the level of risk varies between the four ASMs. For example, CBZ showed a lower overall major malformation risk rate than PB and VPA, and a lower risk for specific major malformation types compared to some ASMs (e.g. PB or TPM). All ASMs, regardless of their own association with an increased risk, carried a lower risk than VPA‐exposure which had the highest prevalence from both cohort (9.8%) and routine health record data (9.7%). LTG remains the lowest risk ASM with adequate cohort size (N => 4700 pregnancies from cohort studies and N = 2502 from routine health record studies). LTG does appear to have a dose effect, as do CBZ, PB and VPA. The number of LEV‐exposed pregnancies remains more limited than LTG but, in direct comparison, there is currently no significant difference between these two ASMs, and LEV‐exposed children have a lower overall major malformation risk than CBZ, VPA and PHT. TPM exposure was associated with a higher overall major malformation risk in comparison to LTG, but not other ASMs; this finding may be due to relatively small numbers of participants in the TPM studies and only a few malformation events. TPM exposure, however, was associated with higher specific risks of oro‐facial, craniofacial, skeletal and limb malformations in comparison to LTG, LEV, CBZ, but not in comparison to the two control groups; but specific major malformation data were limited for the control cohorts.

There remains limited information for GBP, OXC, TPM and ZNS and other ‘newer’ medications. The evidence for each ASM monotherapy is summarised below.

Summary of main results

Carbamazepine (CBZ)

CBZ was the most frequently investigated ASM both in terms of the number of publications and the number of included pregnancies (over 8220). The pooled major malformation prevalence for CBZ was 4.7% (95% CI, 3.7 to 5.9) from cohort studies, 4.0% (95% CI 3.0 to 5.4) for routine health record studies and 2.9% (95% CI 2.9 to 5.4); these were relatively consistent with the 5.5% (95% CI 4.5 to 6.6) prevalence from EURAP 2018 (Figure 3).

In comparison to both children born to women without epilepsy and children born to women with untreated epilepsy, pooled data from the cohort studies found that children exposed to CBZ in utero had an increased risk of having a major malformation, with the difference in risk ranging from 1% to 2%, respectively. The findings from routine health record studies are more mixed. In the larger of the two comparisons, CBZ was significantly associated with a higher risk of a major malformation, which was consistent with the cohort study data.

In comparison to unexposed control children (both groups), there was no specific malformation which was increased above the background rate provided by the control children. Data were limited in terms of the specific malformation risk, mainly due to the absence of control data from some large pregnancy and epilepsy registry studies (e.g. North American Epilepsy and Pregnancy Register; UK and Ireland Epilepsy and Pregnancy Register) and reporting of this level of detail from the population datasets (e.g. Denmark Health Record Registers; UK Clinical Research Practice Database; UK Health Record THIN Register). This likely contributed to the non‐significant outcomes found for neural tube malformations, which has been an association found by others using different methodologies (Jentink 2010b).

Data from the cohort studies were more numerous for CBZ‐exposed pregnancies compared to other monotherapy ASMs. There was a higher risk of major malformation in comparison to children exposed to LEV or LTG in comparisons which included over 500 pregnancies in each arm, with the risk being 1% higher. However, comparisons at the specific malformation level were not significant; this was likely due to fewer data being available at this level of investigation. The increased risk observed in data from cohort studies was not replicated in studies using databases containing routine health records, however, patient/event numbers in the databases were smaller than the pooled experience from cohort studies.

Despite its associated higher risk, CBZ exposure was associated with a lower level of risk than VPA exposure, with a 5% difference in overall major malformation risk and a lower risk across neural tube, cardiac, oro‐facial and skeletal/limb malformations. Further, while CBZ is comparable to the increased risk from PB, PHT or TPM in data from cohort studies for overall major malformation, the specific malformation risk varied. The association of CBZ exposure with cardiac malformations was lower in comparison to PB or to PHT exposure and lower for skeletal/limb or oro‐facial/craniofacial malformation risk when compared to TPM exposure.

There was no significant increase in risk in comparison to OXC, GBP, PRM or CZP exposure from both cohort and health record studies, but data were more limited for these comparator ASMs currently and caution is required.

Dose is a key feature of teratogenic malformation risk (Brent 2004). Data from EURAP 2018 is the most reliable to investigate dose associations due to its large number of CBZ‐exposed pregnancies and its low risk of bias on the domain of outcome measurement. It demonstrates a dose‐related risk for CBZ with doses over 700 mg/d carrying a higher risk, although even their lower dose group had a high level of risk in comparison to children exposed to LTG (EURAP 2018).

Clonazepam (CZP)

Data relating to the use of monotherapy clonazepam during pregnancy and child major malformation risk is substantially limited with fewer than 150 pregnancies reported across both cohort and routine health record studies. The generated prevalence reported here was 2.1% (95% CI 0.2 to 17.3%) to 2.5% (95% CI 0.0 to 131.8) but the confidence intervals were very wide, due to the limited data. When compared to other ASMs, however, a lower risk was identified in comparison to VPA exposure, with a risk difference of 9%. Due to the current limited data available, no firm conclusion can be made currently whether CZP is associated with an increased risk in comparison to control children.

Gabapentin (GBP)

Experience with GBP exposure in pregnancy was also limited. Outcomes from only 210 reported pregnancies could be included in our meta‐analysis. The pooled prevalence of major malformation was 2.0% (95% CI 0.1 to 32.2) from cohort studies, with too few data being available from studies using routine health records to provide a pooled prevalence rate. We found no difference between the children exposed to GBP compared with either type of control group, but caution is warranted due to limited numbers. Data which could not be included in meta‐analysis, from the US Medicaid Registers, found that children exposed to GBP (N = 347) for the indication of maternal epilepsy were not at a greater risk of major malformation, which matched the wider finding from 3745 GBP‐exposed pregnancies (any indication). Whilst this study may offer reassurance, caution is required as the indications were not predominantly epilepsy and, therefore, there may be differences in typically prescribed doses; replication of this finding is also required in another adequately powered cohort.

We found no difference in overall major malformation rate or in the specific malformations investigated for the children exposed to GBP compared to CBZ, LTG, LEV, OXC, PHT, PB, TPM and ZNS, but there was a very limited number of GBP‐exposed children. In comparison to the children exposed to VPA, children exposed to GBP in utero had a significantly lower risk (8%) of having a malformation than children exposed to VPA, but data were too limited to investigate specific malformation differences.

Data for GBP dose and malformation rate were limited from cohort studies but data from the US Medicaid Registers failed to find an association between higher doses of GBP and a higher risk of malformation. However, the vast majority of women taking GBP were doing so for conditions other than epilepsy and, therefore, caution is warranted.

Lamotrigine (LTG)

The use of LTG has increased over the last decade in women of childbearing age (Ackers 2009; Man 2012; Wen 2015). The pooled prevalence of major malformation for LTG was 2.7% (95% CI 1.9 to 3.8%) from cohort studies, 3.5% (95% CI 2.5 to 4.9) from routine health record studies, and 2.9% (95% CI 2.3 to 3.7) from the EURAP 2018 collaboration. Most of the evidence indicated no difference in the overall major malformation rate between the children exposed to LTG and either type of control group. However, in comparison to children born to women without epilepsy, there was a significant difference from the pooled cohort data, with the risk being increased by 1% for the children exposed to LTG. Whilst levels of heterogeneity were not increased overall, one study had a much higher rate of malformation in the LTG group. Further, a larger LTG‐exposed group was available for comparison to the women with epilepsy who were not treated, which showed the LTG major malformation risk was not significantly different from the control children for overall malformation rate or for the specific malformation types investigated; this was consistent with the non‐significant finding from the routine health record studies in comparison to both control groups. Further, no increase in specific malformation types were identified, and data from the US Medicaid Registers failed to find an association with oral cleft malformations in over 2000 LTG‐exposed children in comparison to control children who were not exposed to ASMs.

In comparison to LEV, which has also seen a significant increase in use in women of childbearing age in a lot of countries (Meador 2009; Wen 2015), there were no significant differences for LTG in either the overall major malformation rate or the specific malformation types investigated. Children exposed to LTG also did not differ either in terms of overall major malformation rate or in terms of specific malformations compared with children exposed to OXC, CZP, GBP and ZNS; however, data were limited for all of these comparisons due to small numbers of OXC, CZP, GBP and ZNS‐exposed pregnancies.

The children exposed to LTG were at a significantly lower risk of overall major malformation compared with children exposed to CBZ, PB, PHT and TPM exposures (see results for specific risk levels) in cohort study data. Routine health record studies did not replicate this, however, were limited in terms of numbers of included LTG‐exposed pregnancies in comparison to PB, PHT and TPM and, therefore, were perceived as less reliable. Analyses of cohort study data showed that, at the specific malformation level, children exposed to LTG were at a lower risk of cardiac malformations in comparison to PB and PHT exposures and fewer skeletal malformations than children exposed to TPM and PHT. This latter finding was also observed in the large cohort from the US Medicaid Registers. Finally, children exposed to LTG had a three‐fold lower risk of overall major malformation when compared to the children exposed to VPA, with a risk difference showing that the significant reduction in risk was 6% for children exposed to LTG. Neural tube, cardiac, oro‐facial cleft, craniofacial, skeletal and limb malformations were all significantly lower for the LTG‐exposed children.

The large, well‐designed EURAP 2018 study has demonstrated a dose relationship between LTG treatment and major malformation risk, with exposures to LTG ≤ 325 mg/d associated with the lowest malformation prevalence. Other studies did not find a dose relationship, however, the EURAP 2018 collaboration is by far the largest and most standardised in their assessment of the malformation, comprising of participants from 42 countries (N = 2514). Therefore, higher doses of LTG may be associated with higher levels of risk and this should be considered when prescribing doses over 325 mg/d. It is possible that this dose association may account, at least in part, for the variation seen in the outcome in one of the control comparisons.

Levetiracetam (LEV)

The frequency of data and the number of included pregnancies exposed to LEV were more limited than for CBZ, LTG or VPA. There were 1242 LEV‐exposed pregnancies included from cohort data across all comparisons and just 248 from studies employing routine health record data currently. This delay is likely due in part to the time it takes for adequate numbers of women taking newer ASMs to accumulate, however, it undermines counselling for a now commonly established medication (Meador 2009; Wen 2015).

The pooled prevalence for major malformation occurrences following LEV‐exposure was 2.6% (95% CI 1.6 to 4.4) for cohort study data, 2.8% (95% CI 0.0 to 321.9) for routine health record study data (which had large confidence intervals) and 2.8% (95% CI 1.7 to 4.5) from EURAP 2018. There was no significant difference between the children exposed to LEV and control children in the meta‐analysis for overall major malformation rate; these comparisons also both contained > 1000 pregnancies in each arm. Data pertaining to specific malformation types in comparison to control children were limited, however, and it is not possible to draw conclusions until more data are available.

In comparison to other ASM treatments, children exposed to LEV were not significantly different from children exposed to LTG in terms of overall major malformation prevalence or the specific malformation types investigated. In addition, we found no significant difference between children exposed to LEV compared with those exposed to GBP, OXC, CZP, TPM ZNS or PRM, although data within these comparisons were somewhat limited. Children exposed to LEV had a lower overall MCM rate than the children exposed to CBZ, PB and PHT exposures, but there was no difference in terms of the specific malformation types investigated. While the overall major malformation risk was not significantly different for the children exposed to LEV versus those exposed to TPM, the children exposed to LEV were at lower risk of having an oro‐facial/craniofacial, skeletal or limb malformation in comparison to the TPM‐exposed children. Additionally, children exposed to LEV had a lower specific risk of developing a cardiac malformation in comparison to PB‐exposed children. Finally, the children exposed to LEV had a 7 to 8% lower risk of overall malformations compared with the children exposed to VPA, with a lower risk identified for all investigated specific types of malformations for the children exposed to LEV.

Investigation between dose of LEV and major malformation outcome was limited by numbers included within the individual studies (i.e. Denmark Health Record Registers; Kerala Epilepsy and Pregnancy Registry; North American Epilepsy and Pregnancy Register; UK and Ireland Epilepsy and Pregnancy Register), including the EURAP 2018 study. In 599 LEV‐exposed pregnancies, EURAP 2018 did not investigate lower versus higher‐dose LEV. Caution is required, therefore, regarding the malformation risk with above average doses of LEV, until more data are available.

Oxcarbazepine (OXC)

Data for pregnancy outcomes following exposure to OXC were limited to just under 400 pregnancies in cohort studies and 507 pregnancies from routine health record studies. The prevalence of major malformation was 2.8% (95% CI 1.1 to 6.6) versus 4.8% (95% CI 0.7 to 31.5) with routine health record studies containing more OXC‐exposed pregnancies across comparisons. The prevalence reported from EURAP 2018 for 333 OXC monotherapy‐exposed pregnancies was 3.0% (95% CI 1.4 to 5.4).

In comparison to control children, the pooled routine health record study data found no elevated risk for OXC exposure in comparison to controls; these results were similar in the cohort study data. Given the numbers of included OXC‐exposed pregnancies across both study types, further research is required for conclusions to be drawn in regard to OXC exposure and major malformation outcome. While limited comparisons to other ASMs could be made, where evidence was available, there was no significant difference between the overall major malformation rate or the specific malformations investigated compared with children exposed to CBZ, CZP, LEV, LTG, GBP, PHT, PB, TPM, PRM and ZNS. Children exposed to OXC were at a significantly lower risk of having an major malformation of any type compared with children exposed to VPA, with the risk difference being 4 to 6% depending on the study type. There were very limited data pertaining to specific malformation types, and caution is required.

Only EURAP 2018 reported dose and malformation rates for OXC‐exposed pregnancies. Whilst they did not compare lower versus higher OXC dose, they did report that certain dose levels of OXC were comparable to lower‐dose LTG. More studies of OXC‐exposed pregnancies are required, however, before it is determined whether a higher level of OXC dose carries a higher malformation risk.

Phenobarbital (PB)

Despite years of use, data from prospective studies investigating PB as monotherapy were surprisingly limited, with only 840 monotherapy‐exposed pregnancies across the different comparisons and study types. The data pooled from included studies generated a major malformation prevalence of 6.3% (95% CI 4.8 to 8.3) from cohort studies and 8.8% (95% CI 0.0 to 9722.4) from routine health record studies; the latter was limited in cohort size and the prevalence should be interpreted with caution. There was a prevalence of 6.5% (95% CI 4.2 to 9.9) from EURAP 2018.

The results regarding PB‐exposure in comparison to control children demonstrated variable results. We found a significantly increased risk of overall major malformation compared with children born to women without epilepsy, with a risk difference of 4%. However, we found no significant difference compared with children born to women without epilepsy. However, both comparisons included under 500 PB‐exposed pregnancies, which may account for the unstable pattern of the findings. Routine health record data studies included too few PB‐exposed pregnancies at this time to provide reliable estimates. Data pertaining to specific malformations were extremely limited or missing and likely contributed to the non‐significant differences found for PB in comparison to the control children. This is certainly the case for cardiac malformations, where rates of cardiac malformations are increased in comparison to numerous other ASMs exposures.

In comparison to other ASMs, children exposed to PB were not at a significantly increased rate of overall major malformation compared with children exposed to CBZ, CZP, GBP, PHT, OXC, TPM, PRM, LEV and ZNS exposures; but the comparison to CBZ exposure was the only one where the PB‐exposed group had over 500 pregnancies. PB exposure was significantly associated with an increased risk of oro‐facial clefts and craniofacial malformations when compared to LEV or LTG exposures. Children exposed to PB had a higher overall major malformation than the children exposed to LTG, but a lower risk compared with the children exposed to VPA, with the risk being 4% lower. Therefore, despite both PB and VPA being associated with an increased risk of being born with an major malformation, the risk associated with VPA is significantly higher, including for cardiac malformations.

The majority of studies did not investigate or report on a potential relationship between dose of PB and major malformation risk, due to limited included pregnancies. A dose‐mediated risk was also apparent for cardiac malformations, with the prevalence increasing from 1% to 8% for doses < 150 mg/d and those ≥ 150 mg/d, respectively (EURAP 2018). Samren 1997 also found a dose effect for PB. Given the size of the EURAP 2018 cohort and their standardised approach to reviewing, it is concluded that there is likely a strong dose relationship for PB.

Phenytoin (PHT)

The pooled prevalence of major malformation in the PHT‐exposed children was 5.4% (95% CI 3.6 to 8.1%) for cohort studies, 6.8% (95% CI 0.1 to 701.2%) for routine health record studies and 6.4% (95% CI 2.8 to 12.2%) for EURAP 2018. There were 1327 PHT‐exposed pregnancies included in the cohort studies, but just 103 children reported from routine health record studies. The children exposed to PHT were at a significantly increased risk in comparison with both types of control group, with the difference in risk being 3% in the cohort data. However, we found no association between PHT and specific major malformation types; although data were limited in these comparisons due to the limited control data reported in publications from the epilepsy and pregnancy registers.

In comparison to other ASMs, children exposed to PHT were not at an increased risk of overall major malformation compared with children exposed to CZP, CBZ, GBP, OXC, TPM, PRM, PB or ZNS; however, data comparing PHT with the 'newer' ASMs were limited and caution is needed in the interpretation of these non‐significant findings. In contrast, compared to studies with a greater number of included children, the children exposed to PHT were at an increased risk of overall major malformation compared with children exposed to LTG or LEV, with the risk difference indicating a 2% increase in major malformation; however, these RDs were not statistically significant. In contrast, the children exposed to PHT were significantly less likely to have a major malformation than the children exposed to VPA, with the difference in risk being 5% lower. Further, the children exposed to PHT were also at a lower risk than those exposed to VPA for their risk of neural tube, cardiac and skeletal/limb malformations.

In terms of specific malformations, children exposed to PHT were less likely than those exposed to PB to have a craniofacial malformation. There was a noted increase in cardiac, skeletal and limb malformations for the PHT‐exposed children compared with those exposed to LTG, which was one of the larger comparisons in terms of PHT‐exposed pregnancies. Finally, the rates of neural tube, cardiac, skeletal and limb malformations were significantly lower for the children exposed to PHT in comparison to the VPA‐exposed children.

The majority of studies did not report on whether the risk of being born with a major malformation was associated with dose of PHT; however, those that did investigate such an association did not show a consistent pattern (Kaaja 2003; Kaneko 1999; Motherisk Registry; North American Epilepsy and Pregnancy Register; Samren 1997), therefore, the conclusion around dose effects is uncertain.

Primidone (PRM)

This is an old ASM with limited utilisation currently. Evidence pertaining to PRM was extremely limited to 112 pregnancies and caution is warranted when interpreting results. Pooled data from included cohort studies gave a malformation prevalence of 7.9% (95% CI 2.6 to 21.5%). There were just 3 PRM‐exposed cases reported in the routine health record studies. The children exposed to PRM were at a higher risk of overall major malformation in comparison to the children born to women with an untreated epilepsy, which contained the larger number of PRM‐exposed pregnancies. A comparable major malformation risk was found for PRM in comparison to PHT, PB and VPA exposures, but the data were limited. There was either extremely limited or no data available to compare risks to other monotherapy ASMs.

Only one study of 19 PRM cases investigated the dose of PRM and outcome (Kaneko 1999). Therefore, it remains unknown whether there is an association between PRM dose and increased major malformation risk.

Topiramate (TPM)

Experience with TPM was limited to 510 exposed pregnancies in cohort studies (3.9%, 95% CI 2.3 to 6.5%). There were 49 cases from routine health record investigations which met the criteria for being included in the meta‐analyses (4.1%, 95% CI 0.0 to 27,060.0); therefore, caution is required when considering our results. The EURAP 2018 collaboration is also limited currently in its experience with TPM exposures with just 152 exposed pregnancies with a major malformation prevalence of 3.9% (95% CI 1.5 to 8.4%).

In pooled cohort data, in comparison to children born to women without epilepsy, children exposed to TPM had a higher rate of being born with an major malformation with the risk difference being 3%. We found no significant difference compared with the no medication control group, but this comparison had even fewer TPM cases. Pooled data were too limited here to allow for the investigation of specific malformation outcomes in comparison to control children. We found no significant difference in the rate of major malformation compared with children exposed to CBZ, CZP GBP, PHT, PB, PRM, OXC and ZNS. We found a significant increase in the rate of major malformation for the children exposed to TPM compared with the children exposed to LTG, with skeletal/limb and oro‐facial cleft/craniofacial being specifically increased.

Data from US Medicaid Registers provides the largest dataset regarding oro‐facial clefts and reported an association between topiramate and oral clefts from medical reimbursement databases (4.1 per 1000 live births). This is similar to a retrospective study which was not included in this review (Mines 2014), in a case‐control study (Margulis 2012), and in a previous meta‐analysis (Alsaad 2015) which were beyond the inclusion criteria of this review. This demonstrates the cohort sizes which are required to investigate very specific rare events, such as specific types of major malformation.

The overall major malformation risk was comparable to that for LEV or CBZ‐exposed children, but the LEV‐exposed children were at a lower risk of skeletal and limb malformations, as were the CBZ‐exposed children. The children exposed to TPM had a lower cardiac risk than the children exposed to PB, and they were less likely to have a malformation of any type compared with the children exposed to VPA, with the difference in risk being 7%.

Most studies were too limited to be able to provide reliable investigations into a dose association, however, Hernandez‐Diaz, using (US Medicaid Registers) data, found that the adjusted RRs for oro‐facial clefts at doses ≤ 100 mg/d and > 100 mg/d were 1.64 (95% CI 0.53 to 5.07) and 5.16 (95% CI 1.94 to 13.73) for lower and higher doses, respectively.

Valproate (VPA)

In utero exposure to VPA and its possible association with an increased teratological risk has been discussed in the literature since the 1980s, when the first case reports emerged documenting children with a specific constellation of malformations following exposure to VPA (Ardinger 1988; DiLiberti 1984). Larger cohorts such as EURAP 2018 and data from population‐based health records studies (e.g. Denmark Health Record Registers; Sweden Health Record Registers) as well as the pregnancy registries (Australian Epilepsy and Pregnancy Register; Kerala Epilepsy and Pregnancy Registry; North American Epilepsy and Pregnancy Register; UK and Ireland Epilepsy and Pregnancy Register) and observational studies (e.g. Meador 2006; Omtzigt 1992; Samren 1997) included here, have all provided evidence to confirm that VPA is a significant human teratogen which is associated with an increase is a variety of malformation types. Here, we reported on 3018 VPA‐exposed children from prospective cohort style studies and 1482 VPA‐exposed pregnancies from routine health record studies.

In the meta‐analyses reported here a consistent pattern emerged: children exposed to VPA were at an increased risk of both a higher overall major malformation risk and risk of specific malformations including neural tube, cardiac, oro‐facial cleft, craniofacial, skeletal and limb malformations. The prevalence of major malformation following exposure to VPA in the womb was 9.8% (95% CI 8.1 to 11.9) for cohort studies, with similar rates from routine health record studies (9.7%, 95% CI 7.1 to 13.4) and from EURAP 2018 (10.3, 95% CI 8.8 to 12.0%). Children exposed to VPA were at an increased risk of being born with a major malformation compared with both the children of women without epilepsy and the children of women with untreated epilepsy, with the risk difference being 7% and 6% compared with the respective control groups. Analysis of the risks associated with VPA treatment at the specific malformation level was limited by a lack of control data; however, children exposed to VPA remained at a significantly increased risk for neural tube, cardiac and skeletal malformations compared with control children.

In comparison to other ASMs, in the meta‐analyses reported here, children exposed to VPA were at an increased risk of major malformation compared with children exposed to CBZ, CZP, GBP, LEV, LTG, TPM, OXC, PB and PHT, with the ZNS group being non‐significant but too being small to make a reliable comparison. The increased risk associated with VPA exposure ranged from 4% to 9%, depending on the comparator ASM.

At the specific malformation level, children exposed to VPA were at an increased risk of neural tube malformation compared with the children exposed to CBZ, LEV, LTG and PHT. We did not note any increase in the specific malformation type analyses compared to children exposed to GBP, OXC, PB or TPM, but this is most likely due to limited data. However, we found an increased rate of cardiac malformation compared to CBZ, LEV, LTG, TPM, PHT and an equal cardiac risk in comparison to the increased risk for PB‐exposed children. Oro‐facial cleft and craniofacial malformations were also significantly more common in the children exposed to VPA compared with children exposed to CBZ, LEV and LTG. There was no difference in the rate of oro‐facial cleft or craniofacial malformations compared with TPM, PB or PHT, but these are found to carry their own risks of this malformation type (US Medicaid Registers). Finally, skeletal or limb malformations in children exposed to VPA compared with children exposed to CBZ, LEV, LTG or PHT were significantly higher. All specific malformation comparisons that the data compared with CZP, GBP, ZNS and OXC were too limited for conclusions to be made.

When weighing up the risks and benefits of VPA treatment, the effects of VPA on other developmental outcomes including the developing brain should also be considered when considering the level of risk posed by VPA. VPA exposure is now also recognised as a neurobehavioural teratogen, with implications for the future cognitive functioning of the exposed child (Bromley 2014), and an increased risk of neurodevelopmental disorders such as autistic spectrum disorders (Christensen 2013) and attention deficit hyperactivity disorder (Christensen 2019).

More than any other ASM, studies have reported dose associations with level of major malformation risk for VPA (Australian Epilepsy and Pregnancy Register; EURAP 2018; Fairgrieve 2000; Israeli Teratogen Service; Kaneko 1999; Lindhout 1992; Milan Study 1999; North American Epilepsy and Pregnancy Register; Samren 1997; UK and Ireland Epilepsy and Pregnancy Register). The largest data set with clear dose comparisons is the EURAP 2018 collaboration, which found that the prevalence of major congenital malformations increased from 6.3% at doses < 650 mg daily to 25.2% for doses ≥ 1450 mg daily. Interestingly, pregnancy registers have reported a decrease in the mean dose for new registrations (UK and Ireland Epilepsy and Pregnancy Register) and have noted that this is associated with a reduction in the number of observed cases of neural tube malformations and hypospadias (Australian Epilepsy and Pregnancy Register).

Zonisamide

Experience with ZNS exposure was limited to 130 cases described in four studies (Jimenez 2020; MONEAD 2020; North American Epilepsy and Pregnancy Register; UK and Ireland Epilepsy and Pregnancy Register), therefore, it is not possible to draw conclusions at this time. Further efforts are needed to develop experience with this medication in pregnancy, as it has been in use for a long period in certain parts of the world (Oommen 1999).

Other antiepileptic drugs

Either no, or very limited numbers, of pregnancies were found for other ASMs from the searches such as ethosuximide, sulthiame, perampanel, lacosamide or vigabatrin.

Overall completeness and applicability of evidence

Efforts were made to ensure that the evidence presented here was as complete as possible by including the two dominant research study designs for this area of research; cohort study designs and datasets which contain routinely collected health records. However, we did not include case‐control congenital anomaly registers. In these registers, children are enrolled when the presence or absence of a malformation is known and, therefore, we classified recruitment as retrospective (e.g. Jentink 2010a; Jentink 2010b). Further, the nature of this data meant that it could not be directly combined into meta‐analysis with the data from the prospective observational studies. Additionally, in order to make the results of this review applicable to the treatment of women with epilepsy, included studies were required to include 70% or greater proportions of women taking ASMs for the treatment of epilepsy. This, however, will have reduced the sample size and may not be necessary. Whilst Christensen and colleagues (Denmark Health Record Registers) in 2021 found no difference in risk estimates in the children of women with epilepsy in comparison to the children born to women with other indications, Hernandez Diaz and colleagues, using US Medicaid Registers, derived data found that, in the context of TPM, the indication did alter the outcome reported. Further investigations are required to answer whether limiting this review to a high proportion of women with epilepsy is required.

Efforts were made to ensure that the most up‐to‐date information from the longitudinal research initiatives was utilised, which meant that we often had to take outcomes for different ASMs from a number of different papers, or that authors investigated malformation types separately over different papers, or published updates for certain ASMs only. The largest challenge in terms of the completeness of the evidence came from some studies not reporting specific monotherapy outcomes or reporting monotherapy and polytherapy outcomes for a particular ASM together (e.g. Richmond 2004; Sabers 2004). However, this appeared to be a more frequent finding with older studies and there was a noticeable trend regarding separate reporting for each ASM for monotherapy exposures.

The final challenge to the completeness of the data was in regard to the risk of specific types of malformations, due in a large part to the failure of included studies to publish specific malformation outcomes for all included groups. Whilst this is undoubtedly due to publication space, providing such information is critical for understanding the risks associated with specific malformation types. As demonstrated, in the case of PB or TPM, an ASM exposure may be associated with specific malformations, so reporting only an overall malformation figure may mask important associations. Further, unclear reporting and differences in the defining of certain malformation types or groups meant that we could not investigate hypospadias or gastrointestinal malformations, which have been linked to certain ASM exposures (EURAP 2018; Sweden Health Record Registers).

A few points of heterogeneity were found between included studies, which may limit the completeness of the evidence. Studies varied in how they dealt with the inclusion of foetal deaths or interruptions of pregnancy (with and without malformations) and in whether they counted genetic causes of malformation in their overall prevalence. At the outset of this review, we decided to use the author‐defined major malformation rate, as the review authors would be unlikely to have all the data required to determine information about reported major malformations. Considering this, however, we cannot confirm that all the studies applied the same criteria for classifying a major malformation. Further, there were differences between studies in the time at which the outcome was reported. For example, the UK and Ireland Epilepsy and Pregnancy Register has a major malformation reporting time before three months of age, whilst others included malformation presence at birth (e.g. Bozhinov 2009). Data from the EURAP 2018 collaboration and by Christensen and colleagues (2021) using Denmark Health Record Registers demonstrates that the reviewing of major malformation outcome at 12 months of age leads to an increased detection and, therefore, higher prevalence. Thus, data reported from some studies may in fact be an underestimation of the prevalence of major malformation if the assessment of the child occurs prior to 12 months of age.

Finally, major malformation risk is not the only outcome of importance in pregnancy exposures. Beyond the scope of this review, small‐for‐gestational‐age, prematurity, minor congential malformations as well as longer‐term child health and neurodevelopmental outcomes can be altered, with life impacting consequences (Bromley 2014; Clayton‐Smith 2019; Dean 2000) and, therefore, require consideration when understanding the total impact of an ASM exposure on the developing child. Minor malformations, for example, are an important part of the diagnostic criteria for foetal anticonvulsant syndromes, in particular (Clayton‐Smith 2019; Dean 2000) and their presence may lead to a more detailed physical examination to check for more severe physical symptoms of exposure or neurodevelopmental impairment. Neurodevelopmental impairments are also a more commonly occurring outcome in the general population and, therefore, will occur more frequently in the ASM‐exposed populations and can have a significant impact on quality of life (Bromley 2014; Clayton‐Smith 2019).

Strengths of this review update include the creation and advance publication of the review protocol, the clear inclusion criteria, extensive searches, the acquisition of unpublished data, the inclusion of articles not written in English, meta‐analysis for all possible comparisons, the consideration of specific as well as overall major malformation risk, the balance of both systematic reviewing and content expertise and the assessment of risk of bias and quality in the non‐randomised evidence. Further, we improved the quality of the meta‐analyses by stratifying by type of control group and importantly study design. The results across the different study types were summarised in meta‐analysis separately due to the potential overlap in the cases (e.g. a national epilepsy and pregnancy register may contain the same children with a malformation as a population dataset which utilised routine health records for that same region or population). Further, at the start of this review, there were concerns about likely heterogeneity coming from different measurement approaches, periods of follow‐up and different patterns of maternal indications. However, in comparisons with larger numbers of included exposed pregnancies, the prevalences were similar (Table 3, Figure 3). We therefore take the view that cohort studies and studies utilising population level health records offer complimentary evidence which can be viewed as replicating the results of each other, to ensure evidence consistency across the total available data.

Under the Cochrane guidelines, this review will continue to be updated every two years, or following the publication of a significant amount of new data, to ensure it remains up‐to‐date which adds further strength.

Quality of the evidence

The methodological quality for each individual study is displayed in the Risk of bias in included studies and in Figure 2. Randomised controlled trials are thought to be unethical in this area due to the permanence of potential adverse effects for the foetus. Gold standard evidence for this area would, therefore, comprise data coming from a recruitment approach with low selection bias, prospective follow‐up, blinded outcome assessment to a standardised protocol and statistical methods to limit the influence of confounding or mediating variables. Obtaining all of these features in a single study is difficult and different study designs have a different set of strengths and weaknesses.

The RoB ratings provided by an adaption of ROBINS‐I, for example, showed that the certain routine health record studies scored at a lower risk of bias than the cohort studies for risk of selection biases, yet the routine health record studies were at higher risk for outcome measurement which was completed in a non‐standardised manner by clinicians who were not blinded to the ASM exposure of the child. To balance these strengths and weaknesses which are inherent within these study designs, a complimentary set of pharmacovigilance approaches are required in order to have an accurate understanding of the data pertaining to possible risk associated with ASM exposures.

It should be considered that ROBINS‐I is not optimised for pregnancy pharmacovigilance studies where the person taking the medication (mother) is not the person in which the outcome in is being assessed (child) and it was challenging to adopt the signalling questions and ratings to function for this review. Further, the recommended GRADE framework for rating the certainty of evidence was not used, as it would produce differential ratings depending on whether there were differences between the medications or not. For reviews of pregnancy pharmacovigilance data, bespoke risk of bias and certainty of evidence tools are required.

In conclusion, our risk of bias review indicates that, across the included studies, there are a number of important biases assessed as high risk which should be taken into account when interpreting the results. The biases, however, were thought to be balanced across the ASMs investigated and, therefore, it is not felt that the findings were due solely to these biases.

Potential biases in the review process

Review authors RB and JCS were authors on three included studies (Mawer 2010; Meador 2006; UK and Ireland Epilepsy and Pregnancy Register) and author JC on one (Denmark Health Record Registers). This potential bias was reduced by delegating data extraction and risk of bias assessments to other review authors. The ROBINS‐I adaptation, all analyses and interpretation were provided to all authors for review and input.

Agreements and disagreements with other studies or reviews

Despite many review articles in this area, there are few systematic reviews where meta‐analysis has been conducted and, where they have been completed, there are variations in study methodology (i.e. inclusion criteria). For example, the reviews by Veroniki 2017 and Meador 2008, included both prospective and retrospective studies, studies using population‐based electronic healthcare records, and data from case‐control studies. Whilst such a wide inclusion criteria led to increased numbers of included pregnancies within the meta‐analysis, the comparability of data from these different methodological types is unclear. Charlton 2008, for example, had demonstrated different rates of malformations from the UK Clinical Research Practice Database in comparison to the UK and Ireland Epilepsy and Pregnancy Register. Further, combining data from population studies using healthcare records with national epilepsy and pregnancy registers may lead to cases being represented twice; which, for rare outcomes, could alter the analyses significantly. We took a more cautious approach and did not combine data from cohort studies with data from studies that used population‐level routine health records. Whilst our findings were comparable to the more recent Veroniki 2017 review in regard to VPA, PTM, PB, PHT, and CBZ, we did not have enough data to investigate their finding that ethosuximide is associated with an increased risk of major malformation. Overall, our approach of reviewing and undertaking meta‐analysis separately for primary and secondary data sources provides internal comparison and validation of the results which, we feel, is a strength.

Further consistent findings were reported by Jentink 2010b who found the prevalence of malformation following CBZ to be 3.3% based on 2680 CBZ children from eight studies. In contrast to our review, however, Jentink 2010b found a significant association between CBZ exposure and spina bifida. However, as in our review, Jentink 2010a found that eight studies (N = 1565 pregnancies) showed a prevalence rate of 7.5% (95% CI 6.3 to 9.0) in those exposed to VPA, and noted an increase in terms of specific malformations. The data reported here pertaining to LEV is consistent with a previous systematic review (Chaudhry 2014), which also included the three prospective studies reported here (Australian Epilepsy and Pregnancy Register; North American Epilepsy and Pregnancy Register; UK and Ireland Epilepsy and Pregnancy Register) as well as studies utilising other methodologies and reported a prevalence rate of 2.2% (27/1213, 95% CI 1.53 to 3.22).

This updated meta‐analysis did not consistently replicate the reported association between TPM exposure and oral clefts, but we did narratively review data from the large US Medicaid Registers study, which reported an association. In a previously completed meta‐analysis, Alsaad 2015 had wider inclusion criteria which included 3420 patients taking TPM (mixed aetiologies and study design types) and 1,204,981 controls and reported a significant odds ratio (OR 6.26, 95% CI: 3.13 to 12.51). As noted throughout this discussion, data were limited pertaining to the newer ASMs and by the reporting of specific malformations in included studies, therefore, it is possible that the limited data that contributed to this meta‐analysis do not consistently uphold this association across all comparisons.

Authors' conclusions

Implications for practiceThere is consistent evidence, across different study designs, that prenatal exposure to VPA increases the risk of having a child with a major malformation with the risk including neural tube, cardiac, skeletal, limb, oro‐facial cleft and craniofacial malformations. Whilst the prevalence of major malformation is 9.8%, this outcome is only one of a constellation of symptoms associated with VPA exposure in utero (Clayton‐Smith 2019; Dean 2000; Yerby 1992) and which constitute the condition, foetal valproate spectrum disorder (ICD 11 LD2F.03) (Clayton‐Smith 2019). The impact of VPA on the developing foetus is clearly dose‐related (EURAP 2018) and this should be considered when counselling regarding the risks associated with in utero exposure to VPA. The evidence reported here therefore supports regulatory limitations on VPA’s use, unless clinically necessary, to treat maternal epilepsy (NICE 2022) and where clear counselling has been given to the patient. There are other ASMs, however, which also require careful patient counselling and these include CBZ, PB, TPM and PHT.The increased data included in this review update did not alter the previous findings which suggested no increased risk of major malformation for children exposed to either LTG or LEV in utero compared with either control group across the different study types. There is more limited information on LEV exposure and specific malformation outcomes, however. For all other ASMs, the data are limited, and more data are required before conclusions can be drawn for either an overall major malformation risk or for specific malformation types. Further, it is now clear that the dose of ASM is a key component to major malformation risk for non‐VPA ASMS also. CBZ, PB and even LTG have demonstrated such an association when cohorts are adequately powered. For other ASMs, including LEV, the data are limited at present to inform reliably on malformation risks at higher doses. The EURAP 2018 collaboration has the largest dataset stratified by dose of ASM currently. This lack of limited data for specific doses should be openly discussed with women planning a pregnancy or who are in the childbearing years and an absence of data should not imply a lack of risk.Given the variance in major malformation risk associated with individual ASM treatments and at different doses, preconceptual counselling should be tailored to the individual patient. Although traditional counselling has been that 90% of children born to women with epilepsy have healthy children, this oversimplifies a complex set of data. The ASM type, but also dose and considerations regarding specific malformation types, should also be central to counselling. It is also important to note that major malformation risk is just one aspect and that minor malformations and longer‐term child health and neurodevelopmental outcome risks should also feature in counselling.Finally, every effort should be made by clinicians to inform women about local initiatives collecting data on ASM use in pregnancy and child malformation outcomes to improve the availability of evidence on which to base treatment decisions. Epilepsy and pregnancy registers have made a large contribution to the available dataset, but this is only possible with the support of referring clinicians and the women who participate.

Implications for researchImplications for research and pharmacovigilanceThere is an obvious delay between the approval of a medication for use and obtaining comprehensive evidence regarding the potential major malformation risk. Some delay is inevitable, however, a longer delay than necessary will limit evidence‐based decision‐making regarding optimising the treatment of maternal epilepsy whilst limiting potential foetal risk. A failure to document the first few years’ worth of pregnancies to women on newer medications delays knowledge acquisition and new ASMs use in women of childbearing age may be unnecessarily avoided for longer than required. There are numerous medications approved for the treatment of epilepsy around the world, yet we see many without data at this time. The emergence of population level datasets using routine health record databases will likely have a positive impact on this latency, due to their automatic inclusion of large populations (Denmark Health Record Registers). Whilst low in participation selection bias, utilising routine healthcare data has reduced measurement sensitivity though, and disease pregnancy and epilepsy registers or clinical studies which employ blinded, standardised review of the malformation outcome offer a more sensitive approach to outcome measurement. The pharmacovigilance strategy for the ASMs, therefore, should actively include different study designs which balance each other's methodological areas of strength and weakness to form a reliable and comprehensive evidence base.The RoB ratings highlight the issue that within‐study methodological improvements are required. Few studies, for example, report on how the major malformation was assessed and determined to be major or minor and whether this was done blinded to the ASM history, despite this being the primary study outcome. Therefore, an easily adopted improvement for research is to encourage the use of blinded, standardised assessments of the physical outcomes and use standardised classification approaches, such as those used by clinical geneticists, including the Human Phenotype Ontology (HPO) (http:// human‐phenotype‐ontology.github.io/about.html) to allow for more accurate comparison across studies.Whilst research methodologies have become more refined over the years, for example, by reporting individual ASM types, rather than a single monotherapy group or recognising the importance of ASM dose, there are still several limitations in the approach to data reporting. The provision of an overall major malformation risk figure, for example, is unlikely to be reliable, as demonstrated for PB and TPM, and future data collection and analysis should implement automatic reporting at the specific malformation level, including this as supplementary information. To improve the data at this finer level, initiatives will require large cohorts and, therefore, there should be a movement towards standardised protocols and procedural alignment across research initiatives to allow for large enough datasets regarding specific malformation types for specific ASMs and specific doses.Further investigations are also required into the factors which may modify the major malformation risk. This includes further consideration regarding folate supplementation and regarding the optimal dose for women with epilepsy. As cohorts increase in size, more nuanced investigations into dose associations are required by specific malformation types and future work should also consider any family risk factors. Observations have shown that some women who take ASMs, even at a very low dose, appear to be at higher risk of having a child with an ASM‐associated malformation. Further research focusing on identification of genomic variants which might modify how different women metabolise ASMs is crucial so that those who may be at higher risk of having a child with a major malformation, even when taking a lower dose of a specific ASM, can be identified and ASM treatment selected accordingly. Whilst this has proven difficult in the past, whole exome/genome sequencing, with careful selection of individuals for testing, is likely to make this more achievable (Ku 2011).Finally, longitudinal work which also investigates the longer‐term health outcomes of children with ASM exposures should be undertaken to understand the true impact. Where possible, research initiatives which recruit pregnant women with epilepsy for the purpose of investigating major malformation outcomes should also seek to, where possible, utilise these populations to understand child health and neurodevelopmental outcomes.

What's new

History

Protocol first published: Issue 11, 2012
Review first published: Issue 11, 2016

Acknowledgements

AcknowledgementsThe authors would like to thank Rachael Kelly, Barbara Kreilkamp, Sarah Nevitt, Silviya Balabanova, William Nevitt, Gabriela Czanner, Sandra Peternel, Esther van Zuuren, Krzysztof Lach, Mariangela Pannebianco, Asaf Achiron, Kiyomi Shinohara, Masi Perez, Antione Wolff, Liqing Yao, Stewart Rust and Dora Lozsadi for their help with the translation of articles. The authors would also like to thank Juliet Hounsome, Andrew McKay, Alexandra Rigby for their input into the earlier version of this review and to the authors who provided additional clarification or unpublished data.The authors, and the Cochrane Epilepsy Group, are grateful to the following external peer reviewers for their time and comments. John Craig, Leslie ChoiThe Cochrane Epilepsy Group supported the authors in the development of this review. The following people conducted the editorial process for this review.Managing Editor (provided editorial guidance to authors, edited the update, conducted editorial policy checks): Rachael KellyInformation Specialist (managed and updated searches): Graham ChanCopy Editor (copy‐editing and production): Anne Lethaby

Competing interests

Declarations of interestRB's institution has received consultancy fees from UCB Pharma on one occasion due to work undertaken by RB.NA has been sponsored to attend educational meetings and conferences in epilepsy over the last five years by UCB Pharma, GSK and Boehringer Ingelheim, and has participated in regional advisory Board meetings for Eisai on their product eslicarbazepine and zonisamide.AM leads the National Audit of Seizure Management in Hospitals (NASH), which is funded via a grant from UCB Pharma paid to the University of Liverpool. He has also given lectures at educational events sponsored by Sanofi and GSK, with honoraria paid to University of Liverpool. Professor Tony Marson is Theme Leader for Managing Complex Needs at NIHR CLAHRC NWC and an NIHR Senior Investigator.No other conflicts of interest were declared.

References

1. Al BunyanM, Abo-TalibZ. Outcome of pregnancies in epileptic women: a study in Saudi Arabia. Seizure1999;8(1):26-9. [PMID: 10091844]10.1053/seiz.1998.0239 doi:10.1053/seiz.1998.0239 PubMed
2. AlSheikhMH. Prevalence of epilepsy in Saudi pregnant women and possible effects of anti-epileptic drugs on pregnancy outcomes. Neurosciences2020;25(1):32-7. [PMID: 31982897]10.17712/nsj.2020.1.20190077PMC8015625 doi:10.17712/nsj.2020.1.20190077 PubMed
3. JazayeriD, GrahamJ, HitchcockA, O'BrienTJ, VajdaFJ. Outcomes of pregnancies in women taking antiepileptic drugs for non-epilepsy indications. Seizure2018;56:111-4. [PMID: 29471258]10.1016/j.seizure.2018.02.009 doi:10.1016/j.seizure.2018.02.009 PubMed
4. VajdaF, LanderC, O'BrienT, HitchcockA, GrahamJ, SolinasC, et al. Australian pregnancy registry of women taking antiepileptic drugs. Epilepsia2004;45(11):1466. [PMID: 15509252]10.1111/j.0013-9580.2004.451103.x doi:10.1111/j.0013-9580.2004.451103.x PubMed
5. VajdaF, O'BrienT, GrahamJ, LanderC, EadieM. Dose dependence of fetal malformations associated with valproate. Neurology2013;81(11):999-1003. [PMID: 23911758]10.1212/WNL.0b013e3182a43e81 doi:10.1212/WNL.0b013e3182a43e81 PubMed
6. VajdaFJ, EadieMJ. Maternal valproate dosage and foetal malformations. Acta Neurologica Scandinavica2005;112(3):137-43. [PMID: 16097954]10.1111/j.1600-0404.2005.00458.x doi:10.1111/j.1600-0404.2005.00458.x PubMed
7. VajdaFJ, GrahamJ, LanderCM, O'BrienTJ, EadieM. Teratogenicity of the newer antiepileptic drugs - the Australian experience. Journal of Clinical Neuroscience2012;19(1):57-9. [PMID: 22104350]10.1016/j.jocn.2011.08.003 doi:10.1016/j.jocn.2011.08.003 PubMed
8. VajdaFJ, GrahamJE, HitchcockAA, LanderCM, O'BrienTJ, EadieMJ. Antiepileptic drugs and foetal malformation: analysis of 20 years of data in a pregnancy register. Seizure2019;65:6-11. [PMID: 30593875]10.1016/j.seizure.2018.12.006 doi:10.1016/j.seizure.2018.12.006 PubMed
9. VajdaFJ, GrahamJE, HitchcockAA, O'BrienTJ, LanderCM, EadieMJ. Is lamotrigine a significant human teratogen? Observations from the Australian Pregnancy Register. Seizure2010;19(9):558-61. [PMID: 20739196]10.1016/j.seizure.2010.07.019 doi:10.1016/j.seizure.2010.07.019 PubMed
10. VajdaFJ, HitchcockA, GrahamJ, O'BrienT, LanderC, EadieM. The Australian Register of Antiepileptic Drugs in Pregnancy: the first 1002 pregnancies. Australian and New Zealand Journal of Obstetrics and Gynaecology2007;47(6):468-74. [PMID: 17991111]10.1111/j.1479-828X.2007.00781.x doi:10.1111/j.1479-828X.2007.00781.x PubMed
11. VajdaFJ, LanderCM, HitchcockA, GrahamJ, SolinasC, O'BrienT, et al. Changing Australian prescribing patterns for antiepileptic drugs in pregnancy and their possible consequences. Journal of Clinical Neuroscience2007;14(7):611-7. [PMID: 17400456]10.1016/j.jocn.2006.04.025 doi:10.1016/j.jocn.2006.04.025 PubMed
12. VajdaFJ, O'BrienTJ, GrahamJ, LanderCM, EadieMJ. Associations between particular types of fetal malformation and antiepileptic drug exposure in utero. Acta Neurologica Scandinavica2013;128(4):228-34. [PMID: 23461556]10.1111/ane.12115 doi:10.1111/ane.12115 PubMed
13. VajdaFJ, O'BrienTJ, GrahamJ, LanderCM, EadieMJ. Is carbamazepine a human teratogen?Journal of Clinical Neuroscience2016;23:34-7. [PMID: 26521756]10.1016/j.jocn.2015.07.011 doi:10.1016/j.jocn.2015.07.011 PubMed
14. VajdaFJ, O'BrienTJ, GrahamJ, LanderCM, EadieMJ. Prediction of the hazard of foetal malformation in pregnant women with epilepsy. Epilepsy Research2014;108(6):1013-7. [PMID: 24880523]10.1016/j.eplepsyres.2014.04.005 doi:10.1016/j.eplepsyres.2014.04.005 PubMed
15. VajdaFJ, O'BrienTJ, GrahamJE, HitchcockAA, LanderCM, EadieMJ. Antiepileptic drugs, foetal malformations and spontaneous abortions. Acta Neurologica Scandinavica2017;135(3):360-5. [PMID: 27573510]10.1111/ane.12672 doi:10.1111/ane.12672 PubMed
16. VajdaFJ, O'BrienTJ, GrahamJE, HitchcockAA, LanderCM, EadieMJ. Preexisting illness, fetal malformation, and seizure control rates in pregnant women with epilepsy. Epilepsy & Behavior2020;103(Pt A):106481. [PMID: 31711866]10.1016/j.yebeh.2019.106481 doi:10.1016/j.yebeh.2019.106481 PubMed
17. VajdaFJ, O'BrienTJ, GrahamJE, HitchcockAA, LanderCM, EadieMJ. Pregnancy after valproate withdrawal - fetal malformations and seizure control. Epilepsia2020;61(5):944-50. [PMID: 32314363]10.1111/epi.16505 doi:10.1111/epi.16505 PubMed
18. VajdaFJ, O'BrienTJ, GrahamJE, HitchcockAA, LanderCM, EadieMJ. Valproate-associated foetal malformations - rates of occurrence, risks in attempted avoidance. Acta Neurologica Scandinavica2019;139(1):42-8. [PMID: 30109700]10.1111/ane.13005 doi:10.1111/ane.13005 PubMed
19. VajdaFJ, O'BrienTJ, HitchcockA, GrahamJ, CookM, LanderC, et al. Critical relationship between sodium valproate dose and human teratogenicity: results of the Australian register of anti-epileptic drugs in pregnancy. Journal of Clinical Neuroscience2004;11(8):854-8. [PMID: 15519862]10.1016/j.jocn.2004.05.003 doi:10.1016/j.jocn.2004.05.003 PubMed
20. VajdaFJ, O'BrienTJ, HitchcockA, GrahamJ, LanderC. The Australian Registry of Anti-Epileptic Drugs in Pregnancy: experience after 30 months. Journal of Clinical Neuroscience2003;10(5):543-9. [PMID: 12948456]10.1016/s0967-5868(03)00158-9 doi:10.1016/s0967-5868(03)00158-9 PubMed
21. VajdaFJ, O'BrienTJ, LanderCM, GrahamJ, EadieMJ. Antiepileptic drug combinations not involving valproate and the risk of fetal malformations. Epilepsia2016;57(7):1048-52. [PMID: 27265509]10.1111/epi.13415 doi:10.1111/epi.13415 PubMed
22. VajdaFJ, O'BrienTJ, LanderCM, GrahamJ, RotenA, EadieMJ. Teratogenesis in repeated pregnancies in antiepileptic drug-treated women. Epilepsia2013;54(1):181-6. [PMID: 22882134]10.1111/j.1528-1167.2012.03625.x doi:10.1111/j.1528-1167.2012.03625.x PubMed
23. BagS, BehariM, AhujaGK, KarmarkarMG. Pregnancy and epilepsy. Journal of Neurology1989;236(5):311-3. [PMID: 2760652]10.1007/BF00314466 doi:10.1007/BF00314466 PubMed
24. BarqawiR. Evaluation of antiepileptic drugs in pregnancy in a Jordanian army hospital. Eastern Mediterranean Health Journal2005;11(4):601-5. [PMID: 16700374] PubMed
25. CassinaM, DilaghiA, Di GianantonioE, CesariE, De SantisM, MannaioniG, et al. Pregnancy outcome in women exposed to antiepileptic drugs: teratogenic role of maternal epilepsy and its pharmacologic treatment. Reproductive Toxicology2013;39:50-7. [PMID: 23591043]10.1016/j.reprotox.2013.04.002 doi:10.1016/j.reprotox.2013.04.002 PubMed
26. D'SouzaSW, RobertsonIG, DonnaiD, MawerG. Fetal phenytoin exposure, hypoplastic nails, and jitteriness. Archives of Disease in Childhood1991;65(3):320-4. [PMID: 2025009]10.1136/adc.66.3.320PMC1792889 doi:10.1136/adc.66.3.320 PubMed
27. ĐelmišJ, DražančićA, TkalčevićT, IvaniševićM. Epilepsy in pregnancy [Epilepsija i trudnoća]. Jugoslavenska Ginekologija i Perinatologija1991;31(1-2):23-6. [PMID: 1875716] PubMed
28. ChristensenJ, TrabjergBB, SunY, GilhusNE, Bjork M-H, TomsonT, et al. Prenatal exposure to valproate and risk of congenital malformations - could we have known earlier? A population-based cohort study. Epilepsia2021;62(12):2981-93. [PMID: 34585373]10.1111/epi.17085 doi:10.1111/epi.17085 PubMed
29. Mølgaard-NielsenD, HviidA. Newer-generation antiepileptic drugs and the risk of major birth defects. JAMA2011;305(19):1996-2002. [PMID: 21586715]10.1001/jama.2011.624 doi:10.1001/jama.2011.624 PubMed
30. EroğluE, GökçilZ, BekS, UlaşUH, OdabaşiZ. Pregnancy and teratogenicity of antiepileptic drugs. Acta Neurologica Belgica2008;108(2):53-7. [PMID: 18795597] PubMed
31. Huber-MollemaY, Van ItersonL, OortFJ, LindhoutD, RodenburgR. Neurocognition after prenatal levetiracetam, lamotrigine, carbamazepine or valproate exposure. Journal of Neurology2020;267(6):1724-36. [PMID: 32112258]10.1007/s00415-020-09764-wPMC7293688 doi:10.1007/s00415-020-09764-w PubMed
32. TomsonT, BattinoD, BonizzoniE, CraigJ, LindhoutD, PeruccaE, et al. Comparative risk of major congenital malformations with eight different antiepileptic drugs: a prospective cohort study of the EURAP registry. Lancet Neurology2018;17(6):530-8. [PMID: 29680205]10.1016/S1474-4422(18)30107-8 doi:10.1016/S1474-4422(18)30107-8 PubMed
33. TomsonT, BattinoD, BonizzoniE, CraigJ, LindhoutD, PeruccaE, et al. Declining malformation rates with changed antiepileptic drug prescribing: an observational study. Neurology2019;93(9):e831-40. [PMID: 31391249]10.1212/WNL.0000000000008001 doi:10.1212/WNL.0000000000008001 PubMed
34. TomsonT, BattinoD, BonizzoniE, CraigJ, LindhoutD, PeruccaE, et al. Dose-dependent teratogenicity of valproate in mono- and polytherapy: an observational study. Neurology2015;85(10):866-72. [PMID: 26085607]10.1212/WNL.0000000000001772 doi:10.1212/WNL.0000000000001772 PubMed
35. TomsonT, BattinoD, BonizzoniE, CraigJ, LindhoutD, SabersA, et al. Dose-dependent risk of malformations with antiepileptic drugs: an analysis of data from the EURAP epilepsy and pregnancy registry. Lancet Neurology2011;10(7):609-17. [PMID: 21652013]10.1016/S1474-4422(11)70107-7 doi:10.1016/S1474-4422(11)70107-7 PubMed
36. TomsonT, BattinoD, BonizzoniE, CraigJJ, LindhoutD, PeruccaE, et al. Antiepileptic drugs and intrauterine death: a prospective observational study from EURAP. Neurology2015;85(7):580-8. [PMID: 26187231]10.1212/WNL.0000000000001840 doi:10.1212/WNL.0000000000001840 PubMed
37. FairgrieveSD, JacksonM, JonasP, WalshawD, WhiteK, MontgomeryTL, et al. Population based, prospective study of the care of women with epilepsy in pregnancy. British Medical Journal2000;321(7262):674-5. [PMID: 10987772]10.1136/bmj.321.7262.674PMC27482 doi:10.1136/bmj.321.7262.674 PubMed
38. ArtamaM, AuvinenA, RaudaskoskiT, IsojärviI, IsojärviJ. Antiepileptic drug use of women with epilepsy and congenital malformations in offspring. Neurology2005;64(11):1874-8. [PMID: 15955936]10.1212/01.WNL.0000163771.96962.1F doi:10.1212/01.WNL.0000163771.96962.1F PubMed
39. ArtamaM, RitvanenA, GisslerM, IsojärviJ, AuvinenA. Congenital structural anomalies in offspring of women with epilepsy - a population-based cohort study in Finland. International Journal of Epidemiology2006;35(2):280-7. [PMID: 16280367]10.1093/ije/dyi234 doi:10.1093/ije/dyi234 PubMed
40. FröscherW, HerrmannR, NiesenM, BülauP, PeninH, HildenbrandG. The course of pregnancy and teratogenicity of antiepileptic agents in 66 patients with epilepsy [Untersuchungen zum schwangerschaftsverlauf und zur teratogenität der antiepileptika bei 66 epilepsie-patientinnen]. Schweizer Archiv fur Neurologie und Psychiatrie1991;142(5):389-407. [PMID: 1721236] PubMed
41. Garza-MoralesS, Ibarra-PuigJM, Poblano-LunaA, Gilda Mayén-MolinaD, Córdova-LópezS. Epilepsy and pregnancy: prospective study of 100 cases [Epilepsia y embarazo. Estudio prospectivo 100 casos]. Ginecología y Obstetricia de México1996;64:449-54. [PMID: 8974948] PubMed
42. HosnyH, ElkattanM, ZakiMA, RamzyGM, MagdyR, Abo Al-AzayemS. Risk factors of fetal deaths and major birth defects in newborns of women with epilepsy: an Egyptian prospective study. Epilepsy & Behavior2021;123:108251. [PMID: 34411949]10.1016/j.yebeh.2021.108251 doi:10.1016/j.yebeh.2021.108251 PubMed
43. Diav-CitrinO, ShechtmanS, ArnonJ, OrnoyA. Is carbamazepine teratogenic? A prospective controlled study of 210 pregnancies. Neurology2001;57(2):321-4. [PMID: 11468320]10.1212/wnl.57.2.321 doi:10.1212/wnl.57.2.321 PubMed
44. Diav-CitrinO, ShechtmanS, Bar-OzB, CantrellD, ArnonJ, OrnoyA. Pregnancy outcome after in utero exposure to valproate: evidence of dose relationship in teratogenic effect. CNS Drugs2008;22(4):325-34. [PMID: 18336060]10.2165/00023210-200822040-00004 doi:10.2165/00023210-200822040-00004 PubMed
45. Diav-CitrinO, ShechtmanS, ZviN, Finkel-PekarskyV, OrnoyA. Is it safe to use lamotrigine during pregnancy? A prospective comparative observational study. Birth Defects Research2017;109(15):1196-1203. [PMID: 28657171]10.1002/bdr2.1058 doi:10.1002/bdr2.1058 PubMed
46. OrnoyA, CohenE. Outcome of children born to epileptic mothers treated with carbamazepine during pregnancy. Archives of Disease in Childhood1996;75(6):517-20. [PMID: 9014606]10.1136/adc.75.6.517PMC1511819 doi:10.1136/adc.75.6.517 PubMed
47. OrnoyA, ZviN, ArnonJ, WajnbergR, ShechtmanS, Diav-CitrinO. The outcome of pregnancy following topiramate treatment: a study on 52 pregnancies. Reproductive Toxicology2008;25(3):388-9. [PMID: 18424066]10.1016/j.reprotox.2008.03.001 doi:10.1016/j.reprotox.2008.03.001 PubMed
48. PutignanoD, ClavennaA, CampiR, CaneviniMP, VignoliA, BattinoD, et al. Perinatal outcome and healthcare resource utilization in the first year of life after antiepileptic exposure during pregnancy. Epilepsy & Behavior2019;92:14-7. [PMID: 30599457]10.1016/j.yebeh.2018.09.033 doi:10.1016/j.yebeh.2018.09.033 PubMed
49. JiménezM, Grau-LópezL, CiuransJ, García-EsperónC, FumanalA, BarambioS, ChíesE, CodinaM, BecerraJL. Epilepsy and pregnancy. Factors associated with epileptic seizures during pregnancy. Neurologia (Engl Ed) 2020 [Epub ahead of print];38(2):106-113. [DOI: 10.1016/j.nrl.2020.04.024] [PMID: 36162697] doi:10.1016/j.nrl.2020.04.024 PubMed
50. KaajaE, KaajaR, HiilesmaaV. Major malformations in offspring of women with epilepsy. Neurology2003;60(4):575-9. [PMID: 12601095]10.1212/01.wnl.0000044157.28073.dc doi:10.1212/01.wnl.0000044157.28073.dc PubMed
51. FukushimaY, NakamuraY, OgawaY, SaitoY, KanR, KumashiroH, et al. Teratogenicity of antiepileptic drugs: is the prevention possible?Japanese Journal of Psychiatry and Neurology1991;45(2):478-81. [PMID: 1762253]10.1111/j.1440-1819.1991.tb02526.x doi:10.1111/j.1440-1819.1991.tb02526.x PubMed
52. KanekoS, BattinoD, AndermannE, WadaK, KanR, TakedaA, et al. Congenital malformations due to antiepileptic drugs. Epilepsy Research1999;33(2-3):141-58. [PMID: 10094426]10.1016/s0920-1211(98)00084-9 doi:10.1016/s0920-1211(98)00084-9 PubMed
53. KanekoS, OtaniK, FukushimaY, OgawaY, NomuraY, OnoT, et al. Teratogenicity of antiepileptic drugs: analysis of possible risk factors. Epilepsia1988;29(4):459-67. [PMID: 3134192]10.1111/j.1528-1157.1988.tb03746.x doi:10.1111/j.1528-1157.1988.tb03746.x PubMed
54. KanekoS, OtaniK, KondoT, FukushimaY, KanR, TakedaA, et al. Teratogenicity of antiepileptic drugs and drug specific malformations. Japanese Journal of Psychiatry and Neurology1993;47(2):306-8. [PMID: 8271575]10.1111/j.1440-1819.1993.tb02084.x doi:10.1111/j.1440-1819.1993.tb02084.x PubMed
55. KanekoS, OtaniK, KondoT, FukushimaY, NakamuraY, OgawaY, et al. Malformation in infants of mothers with epilepsy receiving antiepileptic drugs. Neurology1992;42(4 Suppl 5):68-74. [PMID: 1574179] PubMed
56. KaurTP, SahuL, RathoreAM, BhasinS. Obstetric outcomes in pregnant women with seizure disorder: a hospital-based, longitudinal study. Turkish Journal of Obstetrics & Gynecology2020;17(3):161-9. [PMID: 33072419]10.4274/tjod.galenos.2020.87300PMC7538818 doi:10.4274/tjod.galenos.2020.87300 PubMed
57. KellyTE, EdwardsP, ReinM, MillerJQ, DreifussFE. Teratogenicity of anticonvulsant drugs II: a prospective study. American Journal of Medical Genetics1984;19(3):435-43. [PMID: 6507489]10.1002/ajmg.1320190303 doi:10.1002/ajmg.1320190303 PubMed
58. BegumS, SarmaSP, ThomasSV. Malformation in index pregnancy in women with epilepsy is not followed by recurrence in subsequent pregnancy. Epilepsia2013;54(12):e163-7. [PMID: 24138300]10.1111/epi.12411 doi:10.1111/epi.12411 PubMed
59. KeniRR, JoseM, ReshmaAS, BaishyaJ, Sankara SarmaP, ThomasSV. Anti-epileptic drug and folic acid usage during pregnancy, seizure and malformation outcomes: changes over two decades in the Kerala Registry of Epilepsy and Pregnancy. Epilepsy Research2020;159:106250. [PMID: 31855827]10.1016/j.eplepsyres.2019.106250 doi:10.1016/j.eplepsyres.2019.106250 PubMed
60. KeniRR, JoseM, SarmaPS, ThomasSV, Kerala Registry of Epilepsy and Pregnancy Study Group. Teratogenicity of antiepileptic dual therapy: dose-dependent, drug-specific, or both?Neurology2018;90(9):e790-6. [PMID: 29429975]10.1212/WNL.0000000000005031 doi:10.1212/WNL.0000000000005031 PubMed
61. SeshachalaBB, JoseM, LathikakumariAM, MuraliS, KumarAS, ThomasSV. Valproate usage in pregnancy: an audit from the Kerala Registry of Epilepsy and Pregnancy. Epilepsia2021;62(5):1141-7. [PMID: 33782943]10.1111/epi.16882 doi:10.1111/epi.16882 PubMed
62. ThomasSV, AjaykumarB, SindhuK, FrancisE, NamboodiriN, SivasankaranS, et al. Cardiac malformations are increased in infants of mothers with epilepsy. Pediatric Cardiology2008;29(3):604-8. [PMID: 18188637]10.1007/s00246-007-9161-4 doi:10.1007/s00246-007-9161-4 PubMed
63. ThomasSV, IndraniL, DeviGC, JacobS, BeegumJ, JacobPP, et al. Pregnancy in women with epilepsy: preliminary results of Kerala registry of epilepsy and pregnancy. Neurology India2001;49(1):60-6. [PMID: 11303244] PubMed
64. ThomasSV, JeemonP, PillaiR, JoseM, LalithakumariAM, MuraliS, et al. Malformation risk of new anti-epileptic drugs in women with epilepsy; observational data from the Kerala Registry of Epilepsy and Pregnancy (KREP). Seizure2021;93:127-32. [PMID: 34740142]10.1016/j.seizure.2021.10.015 doi:10.1016/j.seizure.2021.10.015 PubMed
65. ThomasSV, JoseM, DivakaranS, Sankara SarmaP. Malformation risk of antiepileptic drug exposure during pregnancy in women with epilepsy: results from a pregnancy registry in South India. Epilepsia2017;58(2):274-81. [PMID: 28084641]10.1111/epi.13632 doi:10.1111/epi.13632 PubMed
66. Jäger-RomanE, DeichlA, JakobS, HartmannAM, KochS, RatingD, et al. Fetal growth, major malformations, and minor anomalies in infants born to women receiving valproic acid. Journal of Pediatrics1986;108(6):997-1004. [PMID: 3086531]10.1016/s0022-3476(86)80949-0 doi:10.1016/s0022-3476(86)80949-0 PubMed
67. KochS, Gopfert-GeyerI, Jager-RomanE, JakobS, HuthH, HartmannA, et al. Anti-epileptic agents during pregnancy: a prospective study on the course of pregnancy, malformations and child development [Antiepileptika wahrend der schwangerschaft. Eine prospektive studie uber schwangerschaftsverlauf, fehlbildungen und kindliche entwicklung]. Deutsche Medizinische Wochenschrift1983;108(7):250-7. [PMID: 6402356]10.1055/s-2008-1069536 doi:10.1055/s-2008-1069536 PubMed
68. KochS, LöscheG, Jager-RomänE, JakobS, RatingD, DeichlA, et al. Major and minor birth malformations and antiepileptic drugs. Neurology1992;42(4 Suppl 5):83-8. [PMID: 1574183] PubMed
69. KuhnzW, Jäger-RomanE, RatingD, DeichlA, KunzeJ, HelgeH, et al. Carbamazepine and carbamazepine-10, -11 epoxide during pregnancy and postnatal period in epileptic mother and their nursed infants: pharmacokinetics and clinical effects. Pediatric Pharmacology1983;3(3-4):199-208. [PMID: 6677873] PubMed
70. KuhnzW, KochS, JakobS, HartmannA, HelgeH, NauH. Ethosuximide in epileptic women during pregnancy and lactation period: placental transfer, serum concentrations in nursed infants and clinical status. British Journal of Clinical Pharmacology1984;18(5):671-7. [PMID: 6508976]10.1111/j.1365-2125.1984.tb02528.xPMC1463560 doi:10.1111/j.1365-2125.1984.tb02528.x PubMed
71. RatingD, Jager-RomanE, KochS, DeichlA, HartmannH, JakobS, et al. Major malformations and minor anomalies in the offspring of epileptic parents: the role of antiepileptic drugs. In: NauH, Scott WJ Jr, editors(s). Pharmacokinetics in Teratogenesis. Vol. 1. CRC Press, 1987:205-23. [ISBN: 978-0849368738]
72. RatingD, NauH, Jäger-RomanE, Göpfert-GeyerI, KochS, Beck-MannagettaG, et al. Teratogenic and pharmacokinetic studies of primidone during pregnancy and in the offspring of epileptic women. Acta Paediatrica Scandinavia1982;71(2):301-11. [PMID: 7136638]10.1111/j.1651-2227.1982.tb09418.x doi:10.1111/j.1651-2227.1982.tb09418.x PubMed
73. LindhoutD, HöppenerRJ, MeinardiH. Teratogenicity of antiepileptic drug combinations with special emphasis on epoxidation (of carbamazepine). Epilepsia1984;25(1):77-83. [PMID: 6692794]10.1111/j.1528-1157.1984.tb04158.x doi:10.1111/j.1528-1157.1984.tb04158.x PubMed
74. LindhoutD, MeinardiH, MeijerJW, NauH. Antiepileptic drugs and teratogenesis in two consecutive cohorts: changes in prescription policy paralleled by changes in pattern of malformations. Neurology1992;42(4 Suppl 5):94-110. [PMID: 1574185] PubMed
75. Martinez FerriM, Pena MayorP, Perez Lopez-FraileI, Escartin SiquierA, Martin MoroM, Forcadas BerdusanM, et al. Comparative study of antiepileptic drug use during pregnancy over a period of 12 years in Spain. Efficacy of the newer antiepileptic drugs lamotrigine, levetiracetam, and oxcarbazepine. Neurologia2018;33(2):78-84. [PMID: 27452623]10.1016/j.nrl.2016.05.004 doi:10.1016/j.nrl.2016.05.004 PubMed
76. Martinez FerriM, Pena MayorP, Perez Loppez-FraileI, Castro VilanovaMD, Escartin SiquierA, Martin MoroM, et al. Malformations and fetal death in the Spanish antiepileptic drug and pregnancy registry: results at 6 years [Malformaciones y muerte fetal en el registro espanol de farmacos antiepilepticos y embarazo: resultados a los 6 anos]. Neurologia2009;24(6):360-5. [PMID: 19798601] PubMed
77. MawerG, BriggsM, BakerGA, BromleyR, CoyleH, EatockJ, et al. Pregnancy with epilepsy: obstetric and neonatal outcome of a controlled study. Seizure2010;19(2):112-9. [PMID: 20036166]10.1016/j.seizure.2009.11.008PMC2823982 doi:10.1016/j.seizure.2009.11.008 PubMed
78. MeadorKJ, BakerGA, FinnellRH, KalayjianLA, LiporaceJD, LoringDW, et al. In utero antiepileptic drug exposure: fetal death and malformations. Neurology2006;67(3):407-12. [PMID: 16894099]10.1212/01.wnl.0000227919.81208.b2PMC1986655 doi:10.1212/01.wnl.0000227919.81208.b2 PubMed
79. MeischenguiserR, D'GianoCH, FerraroSM. Oxcarbazepine in pregnancy: clinical experience in Argentina. Epilepsy & Behaviour2004;5(2):163-7. [PMID: 15123016]10.1016/j.yebeh.2003.11.020 doi:10.1016/j.yebeh.2003.11.020 PubMed
80. MelikovaS, BagirovaH, MagalovS. The impact of maternal epilepsy on delivery and neonatal outcomes. Child's Nervous System2020;36(4):775-82. [PMID: 31786631]10.1007/s00381-019-04435-2 doi:10.1007/s00381-019-04435-2 PubMed
81. BattinoD, BinelliS, CaccamoML, CaneviniMP, CangerR, ComoML, et al. Malformations in offspring of 305 epileptic women: a prospective study. Acta Neurologica Scandinavica1992;85(3):204-7. [PMID: 1575005]10.1111/j.1600-0404.1992.tb04029.x doi:10.1111/j.1600-0404.1992.tb04029.x PubMed
82. BattinoD, KanekoS, AndermannE, AvanziniG, CaneviniMP, CangerR, et al. Intrauterine growth in the offspring of epileptic women: a prospective multicentre study. Epilepsy Research1999;36(1):53-60. [PMID: 10463850]10.1016/s0920-1211(99)00020-0 doi:10.1016/s0920-1211(99)00020-0 PubMed
83. CangerR, BattinoD, CaneviniMP, FumarolaC, GuidolinL, VignoliA, et al. Malformations in offspring of women with epilepsy: a prospective study. Epilepsia1999;40(9):1231-6. [PMID: 10487185]10.1111/j.1528-1157.1999.tb00851.x doi:10.1111/j.1528-1157.1999.tb00851.x PubMed
84. MiskovS, Gjergja JuraskiR, MikulaI, BasicS, Bosnjak PasicM, KosecV, et al. The Croatian model of integrative prospective management of epilepsy and pregnancy. Acta Clinica Croatica2016;55(4):535-48. [PMID: 29116720]10.20471/acc.2016.55.04.02 doi:10.20471/acc.2016.55.04.02 PubMed
85. MeadorKJ, PennellPB, MayRC, Van MarterL, McElrathTF, BrownC, et al. Fetal loss and malformations in the MONEAD study of pregnant women with epilepsy. Neurology2020;94(14):e1502-11. [PMID: 31806691]10.1212/WNL.0000000000008687PMC7251524 doi:10.1212/WNL.0000000000008687 PubMed
86. Van MarterLJ, PennellPB, BrownC, HartmanAL, MayRC, McElrathT, et al. Neonatal outcomes in the MONEAD study of pregnant women with epilepsy. Journal of Pediatrics2021;7:100073. [DOI: 10.1016/j.ympdx.2021.100073]PMC1021063637234096 doi:10.1016/j.ympdx.2021.100073 PubMed
87. DanskyLV, AndermannE, RosenblattD, SherwinAL, AndermannF. Anticonvulsants, folate levels, and pregnancy outcome: a prospective study. Annals of Neurology1987;21(2):176-82. [PMID: 3827226]10.1002/ana.410210210 doi:10.1002/ana.410210210 PubMed
88. OguniM, DanskyL, AndermannE, SherwinA, AndermannF. Improved pregnancy outcome in epileptic women in the last decade: relationship to maternal anticonvulsant therapy. Brain & Development1992;14(6):371-80. [PMID: 1492649]10.1016/s0387-7604(12)80343-3 doi:10.1016/s0387-7604(12)80343-3 PubMed
89. GladstoneDJ, BologaM, MaguireC, PastuszakA, KorenG. Course of pregnancy and fetal outcome following maternal exposure to carbamazepine and phenytoin: a prospective study. Reproductive Toxicology1992;6(3):257-61. [PMID: 1591483]10.1016/0890-6238(92)90181-r doi:10.1016/0890-6238(92)90181-r PubMed
90. NulmanI, ScolnikD, ChitayatD, FarkasLD, KorenG. Findings in children exposed in utero to phenytoin and carbamazepine monotherapy: independent effects of epilepsy and medications. American Journal of Medical Genetics1997;68(1):18-24. [PMID: 8986270] PubMed
91. BokhariA, CoullBA, HolmesLB. Effect of prenatal exposure to anticonvulsant drugs on dermal ridge patterns of fingers. Teratology2002;66(1):19-23. [PMID: 12115776]10.1002/tera.10044 doi:10.1002/tera.10044 PubMed
92. BromfieldEB, DworetzkyBA, WyszynskiDF, SmithCR, BaldwinEJ, HolmesLB. Valproate teratogenicity and epilepsy syndrome. Epilepsia2008;49(12):2122-4. [PMID: 18557775]10.1111/j.1528-1167.2008.01696.x doi:10.1111/j.1528-1167.2008.01696.x PubMed
93. Hernandez-DiazS, SmithCR, ShenA, MittendorfR, HauserWA, YerbyM, et al. Comparative safety of antiepileptic drugs during pregnancy. Neurology2012;78(21):1692-9. [PMID: 22551726]10.1212/WNL.0b013e3182574f39 doi:10.1212/WNL.0b013e3182574f39 PubMed
94. Hernández-DíazS, MittendorfR, SmithCR, HauserWA, YerbyM, HolmesLB, et al. Association between topiramate and zonisamide use during pregnancy and low birth weight. Obstetrics and Gynecology2014;123(1):21-8. [PMID: 24463659]10.1097/AOG.0000000000000018 doi:10.1097/AOG.0000000000000018 PubMed
95. HolmesLB, BaldwinEJ, SmithCR, HabeckerE, GlassmanL, WongSL, et al. Increased frequency of isolated cleft palate in infants exposed to lamotrigine during pregnancy. Neurology2008;70(22 Pt 2):2152-8. [PMID: 18448870]10.1212/01.wnl.0000304343.45104.d6 doi:10.1212/01.wnl.0000304343.45104.d6 PubMed
96. HolmesLB, CoullBA, DorfmanJ, RosenbergerPB. The correlation of deficits in IQ with midface and digit hypoplasia in children exposed in utero to anticonvulsant drugs. Journal of Pediatrics2005;146(1):118-22. [PMID: 15644835]10.1016/j.jpeds.2004.08.048 doi:10.1016/j.jpeds.2004.08.048 PubMed
97. HolmesLB, MittendorfR, ShenA, SmithCR, Hernandez-DiazS. Fetal effects of anticonvulsant polytherapies: different risks from different drug combinations. Archives of Neurology2011;68(10):1275-81. [PMID: 21670385]10.1001/archneurol.2011.133 doi:10.1001/archneurol.2011.133 PubMed
98. HolmesLB, RosenbergerPB, HarveyEA, KhoshbinS, RyanL. Intelligence and physical features of children of women with epilepsy. Teratology2000;61(3):196-202. [PMID: 10661909]10.1002/(SICI)1096-9926(200003)61:3<196::AID-TERA7>3.0.CO;2-T doi:10.1002/(SICI)1096-9926(200003)61:3<196::AID-TERA7>3.0.CO;2-T PubMed
99. HolmesLB, WyszynskiDF. North American Antiepileptic Drug Pregnancy Registry. Epilepsia2004;45(11):1465. [PMID: 15509251]10.1111/j.0013-9580.2004.451102.x doi:10.1111/j.0013-9580.2004.451102.x PubMed
100. WyszynskiDF, NambisanM, SurveT, AlsdorfRM, SmithCR, HolmesLB, et al. Increased rate of major malformations in offspring exposed to valproate during pregnancy. Neurology2005;64(6):961-5. [PMID: 15781808]10.1212/01.WNL.0000154516.43630.C5 doi:10.1212/01.WNL.0000154516.43630.C5 PubMed
101. BorthenI, EideMG, VeibyG, DaltveitAK, GilhusNE. Complications during pregnancy in women with epilepsy: population-based cohort study. British Journal of Obstetrics and Gynecology2009;116(13):1736–42. [PMID: 19781049]10.1111/j.1471-0528.2009.02354.x doi:10.1111/j.1471-0528.2009.02354.x PubMed
102. VeibyG, DaltveitAK, EngelsenBA, GilhusNE. Fetal growth restriction and birth defects with newer and older antiepileptic drugs during pregnancy. Journal of Neurology2014;261(13):579-88. [PMID: 24449062]10.1007/s00415-013-7239-x doi:10.1007/s00415-013-7239-x PubMed
103. VeibyG, DaltveitAK, EngelsenBA, GilhusNE. Pregnancy, delivery, and outcome for the child in maternal epilepsy. Epilepsia2009;50(9):2130-9. [PMID: 19490036]10.1111/j.1528-1167.2009.02147.x doi:10.1111/j.1528-1167.2009.02147.x PubMed
104. OmtzigtJG, LosFJ, GrobbeeDE, PijpersL, JahodaMG, BrandenburgH, et al. The risk of spina bifida aperta after first-trimester exposure to valproate in a prenatal cohort. Neurology1992;42(4 Suppl 5):119-25. [PMID: 1574165] PubMed
105. OmtzigtJG, LosFJ, HagenaarsAM, StewartPA, SachsES, LindhoutD. Prenatal diagnosis of spina bifida aperta after first-trimester valproate exposure. Prenatal Diagnosis1992;12(11):893-7. [PMID: 1283633]10.1002/pd.1970121107 doi:10.1002/pd.1970121107 PubMed
106. PardiG, ComoML, De GiambattistaM, OldriniA, PifarottiG. Epilepsy and pregnancy: obstetrical aspects of a prospective multidisciplinary study [Epilessia e gravidanza: aspetti ostetrici di uno studio prospettico multidisciplinare]. Annali di Ostetricia, Ginecologia, Medicina Perinatale1982;103(4):254-63. [PMID: 7181355] PubMed
107. SamrénEB, Van DuijnCM, KochS, HiilesmaaVK, KlepelH, BardyAH, et al. Maternal use of antiepileptic drugs and the risk of major congenital malformations: a joint European prospective study of human teratogenesis associated with maternal epilepsy. Epilepsia1997;38(9):981-90. [PMID: 9579936]10.1111/j.1528-1157.1997.tb01480.x doi:10.1111/j.1528-1157.1997.tb01480.x PubMed
108. Steegers-TheunissenRP, RenierWO, BormGF, ThomasCM, MerkusHM, Op de CoulDA, et al. Factors influencing the risk of abnormal pregnancy outcome in epileptic women: a multi-centre prospective study. Epilepsy Research1994;18(3):261-9. [PMID: 7805647]10.1016/0920-1211(94)90046-9 doi:10.1016/0920-1211(94)90046-9 PubMed
109. KällénB. A register study of maternal epilepsy and delivery outcome with special reference to drug use. Acta Neurologica Scandinavica1986;73(3):253-9. [PMID: 3716762]10.1111/j.1600-0404.1986.tb03271.x doi:10.1111/j.1600-0404.1986.tb03271.x PubMed
110. WideK, WinbladhB, KällénB. Major malformations in infants exposed to antiepileptic drugs in utero, with emphasis on carbamazepine and valproic acid: a nation-wide, population-based register study. Acta Paediatrica2004;93(2):174-6. [PMID: 15046269]10.1080/08035250310021118 doi:10.1080/08035250310021118 PubMed
111. RegestaG, TanganelliP. The risk of malformations and developmental disturbances in children exposed to antiepileptic drugs: a prospective controlled study. Bollettino Lega Italiana contro l'Epilessia1996;95/96:351-4.
112. TanganelliP, RegestaG. Epilepsy, pregnancy, and major birth anomalies: an Italian prospective, controlled study. Neurology1992;42(4 Suppl 5):89-93. [PMID: 1574184] PubMed
113. CampbellE, DevenneyE, MorrowJ, RussellA, SmithsonWH, ParsonsL, et al. Recurrence risk of congenital malformations in infants exposed to antiepileptic drugs in utero. Epilepsia2013;54(1):165-71. [PMID: 23167802]10.1111/epi.12001 doi:10.1111/epi.12001 PubMed
114. CampbellE, KennedyF, RussellA, SmithsonWH, ParsonsL, MorrisonPJ, et al. Malformation risks of antiepileptic drug monotherapies in pregnancy: updated results from the UK and Ireland Epilepsy and Pregnancy Registers. Journal of Neurology, Neurosurgery & Psychiatry2014;85(9):1029-34. [PMID: 24444855]10.1136/jnnp-2013-306318 doi:10.1136/jnnp-2013-306318 PubMed
115. HuntS, CraigJ, RussellA, GuthrieE, ParsonsL, RobertsonI, et al. Levetiracetam in pregnancy: preliminary experience from the UK Epilepsy and Pregnancy Register. Neurology2006;67(10):1876-9. [PMID: 17130430]10.1212/01.wnl.0000244491.48937.55 doi:10.1212/01.wnl.0000244491.48937.55 PubMed
116. HuntS, RussellA, SmithsonWH, ParsonsL, RobertsonI, WaddellR, et al. Topiramate in pregnancy. Neurology2008;71(4):272-6. [PMID: 18645165]10.1212/01.wnl.0000318293.28278.33 doi:10.1212/01.wnl.0000318293.28278.33 PubMed
117. KinneyMO, MorrowJ, PattersonCC, CampbellE, RussellA, SmithsonHW, et al. Changing antiepilepsy drug-prescribing trends in women with epilepsy in the UK and Ireland and the impact on major congenital malformations. Journal of Neurology, Neurosurgery, and Psychiatry2018;89(12):1320-3. [PMID: 29661925]10.1136/jnnp-2017-317368 doi:10.1136/jnnp-2017-317368 PubMed
118. MawhinneyE, CampbellJ, CraigJ, RussellA, SmithsonW, ParsonsL, et al. Valproate and the risk for congenital malformations: is formulation and dosage regime important?Seizure2012;21(3):215-8. [PMID: 22364656]10.1016/j.seizure.2012.01.005 doi:10.1016/j.seizure.2012.01.005 PubMed
119. MawhinneyE, CraigJ, MorrowJ, RussellA, SmithsonWH, ParsonsL, et al. Levetiracetam in pregnancy: results from the UK and Ireland epilepsy and pregnancy registers. Neurology2013;80(4):400-5. [PMID: 23303847]10.1212/WNL.0b013e31827f0874 doi:10.1212/WNL.0b013e31827f0874 PubMed
120. McCluskeyG, KinneyMO, RussellA, SmithsonWH, ParsonsL, MorrisonPJ, et al. Zonisamide safety in pregnancy: data from the UK and Ireland epilepsy and pregnancy register. Seizure2021;91:311-5. [PMID: 34273670]10.1016/j.seizure.2021.07.002 doi:10.1016/j.seizure.2021.07.002 PubMed
121. MorrowJ, RussellA, GuthrieE, ParsonsL, RobertsonI, WaddellR, et al. Malformation risks of antiepileptic drugs in pregnancy: a prospective study from the UK Epilepsy and Pregnancy Register. Journal of Neurology, Neurosurgery, and Psychiatry2006;77(2):193-8. [PMID: 16157661]10.1136/jnnp.2005.074203PMC2077578 doi:10.1136/jnnp.2005.074203 PubMed
122. CharltonRA, WeilJG, CunningtonMC, De VriesCS. Identifying major congenital malformations in the UK General Practice Research Database (GPRD): a study reporting on the sensitivity and added value of photocopied medical records and free text in the GPRD. Drug Safety2010;33(9):741-50. [PMID: 20701407]10.2165/11536820-000000000-00000 doi:10.2165/11536820-000000000-00000 PubMed
123. CharltonRA, WeilJG, CunningtonMC, RayS, De VriesCS. Comparing the General Practice Research Database and the UK Epilepsy and Pregnancy Register as tools for postmarketing teratogen surveillance: anticonvulsants and the risk of major congenital malformations. Drug Safety2011;34(2):157-71. [PMID: 21247222]10.2165/11584970-000000000-00000 doi:10.2165/11584970-000000000-00000 PubMed
124. BanL, FlemingKM, DoyleP, SmeethL, HubbardRB, FiaschiL, et al. Congenital anomalies in children of mothers taking antiepileptic drugs with and without periconceptional high dose folic acid use: a population-based cohort study. PLOS One2015;10(7):e0131130. [PMID: 26147467]10.1371/journal.pone.0131130PMC4492893 doi:10.1371/journal.pone.0131130 PubMed
125. PetersenI, Collings S-L, McCreaRL, NazarethI, OsbornDP, CowenPJ, et al. Antiepileptic drugs prescribed in pregnancy and prevalence of major congenital malformations: comparative prevalence studies. Clinical Epidemiology2017;9:95-103. [PMID: 28243149]10.2147/CLEP.S118336PMC5317245 doi:10.2147/CLEP.S118336 PubMed
126. Hernandez-DiazS, HuybrechtsKF, DesaiRJ, CohenJM, MogunH, PennellPB, et al. Topiramate use early in pregnancy and the risk of oral clefts: a pregnancy cohort study. Neurology2018;90(4):e342-51. [PMID: 29282333]10.1212/WNL.0000000000004857PMC5798655 doi:10.1212/WNL.0000000000004857 PubMed
127. PatornoE, Hernandez-DiazS, HuybrechtsKF, DesaiRJ, CohenJM, MogunH, et al. Gabapentin in pregnancy and the risk of adverse neonatal and maternal outcomes: a population-based cohort study nested in the US Medicaid Analytic eXtract dataset. PLOS Medicine2020;17(9):e1003322. [PMID: 32870921]10.1371/journal.pmed.1003322PMC7462308 doi:10.1371/journal.pmed.1003322 PubMed
128. WatersCH, BelaiY, GottPS, ShenP, De GiorgioCM. Outcomes of pregnancy associated with antiepileptic drugs. Archives of Neurology1994;51(3):250-3. [PMID: 8129635]10.1001/archneur.1994.00540150044014 doi:10.1001/archneur.1994.00540150044014 PubMed
129. AnnegersJF, ElvebackLR, HauserWA, KurlandLT. Do anticonvulsants have a teratogenic effect?Archives of Neurology1974;31(6):364-73. [PMID: 4441320]10.1001/archneur.1974.00490420030002 doi:10.1001/archneur.1974.00490420030002 PubMed
130. AnnegersJF, ElvebackLR, HauserWA, KurlandLT. Epilepsy anticonvulsants and malformations. Birth Defects original article series1975;11(5):157-60. [PMID: 1218208] PubMed
131. Arteaga-VazquezJ, Luna-MunozL, MutchinickOM. Congenital malformations in the offspring of epileptic mothers with and without anticonvulsant treatment [Malformaciones congenitas en hijos de madres epilepticas con y sin tratmiento con anticonvulsivantes]. Salud Publica Mexico2012;54(6):579-86. [PMID: 23318894]10.1590/s0036-36342012000600006 doi:10.1590/s0036-36342012000600006 PubMed
132. ArulmozhiT, DhanarajM, RangarajR, VengatesanA. Physical growth and psychomotor development of infants exposed to antiepileptic drugs in utero. Neurology India2006;54(1):42-6; discussion 47. [PMID: 16679642]10.4103/0028-3886.24701 doi:10.4103/0028-3886.24701 PubMed
133. BärmigH. Epilepsy and pregnancy [Epilepsie und schwangerschaft]. Geburtshilfe und Frauenheilkunde1973;33(3):203-4. [PMID: 4692245] PubMed
134. BorthenI, EideMG, VeibyG, DaltveitAK, GilhusNE. Complications during pregnancy in women with epilepsy: population-based cohort study. BJOG2009;116(13):1736-42. [PMID: 19781049]10.1111/j.1471-0528.2009.02354.x doi:10.1111/j.1471-0528.2009.02354.x PubMed
135. BozhinovP, BozhinovaC, MarkovaC. [Fetal malformations in women with epilepsy]. Akusherstvo i Ginekologiia2009;48(1):16-21. [PMID: 19496458] PubMed
136. BozhinovaS, BozhinovP. The course of pregnancy and labor in patients with epilepsy [Protichane na bremennostta i razhdaneto pri bremenni s epilepsiia]. Akusherstvo i Ginekologiia1998;37(4):12-4. [PMID: 10360042] PubMed
137. Canún-SerranoS, Zafra de la RosaG, Landeros-VelázquezG, Givaudan-MorenoM. Anticonvulsants and pregnancy [Anticonvulsivos y embarazo]. Boletín Médico del Hospital Infantil de México1986;43(4):219-27. [PMID: 3707704] PubMed
138. Castilla-PuentesR, FordL, ManeraL, Kwarta RF Jr, AscherS, LiQ. Topiramate monotherapy use in women with and without epilepsy: pregnancy and neonatal outcomes. Epilepsy Research2014;108(4):717-24. [PMID: 24598456]10.1016/j.eplepsyres.2014.01.021 doi:10.1016/j.eplepsyres.2014.01.021 PubMed
139. Díaz-RomeroRM, Garza-MoralesS, Mayén-MolinaDG, Ibarra-PuigJ, Avila-RosasH. Facial anthropometric measurements in offspring of epileptic mothers. Archives of Medical Research1999;30(3):186–9. [PMID: 10427868]10.1016/s0188-0128(99)00007-x doi:10.1016/s0188-0128(99)00007-x PubMed
140. DobosM, SchulerD, MarosfiS, BogáthyB. Congenital developmental anomalies in the offspring of epileptic mothers [Veleszületett fejlödési rendellenességek vizsgálata epilepsziás anyák utódaiban]. Orvosi Hetilap1985;126(37):2267-72. [PMID: 4047631] PubMed
141. DravetC, JulianC, LegrasC, MagauddaA, GuerriniR, GentonP, et al. Epilepsy, antiepileptic drugs, and malformations in children of women with epilepsy: a French prospective cohort study. Neurology1992;42(4 Suppl 5):75-82. [PMID: 1574181] PubMed
142. ElshoveJ, Van EckJH. Congenital abnormalities, cleft lip and cleft palate in particular, in children of epileptic mothers [Aangeboren misvormingen, met name gespleten lip met of zonder gespleten verhemelte, bij kinderen van moeders met epilepsie]. Nederlands Tijdschrift voor Geneeskunde1971;115(33):1371-5. [PMID: 5565182] PubMed
143. ThangaratinamS, MarlinN, NewtonS, WeckesserA, BagaryM, GreenhillL, et al. AntiEpileptic drug Monitoring in PREgnancy (EMPiRE): a double-blind randomised trial on effectiveness and acceptability of monitoring strategies. Southampton (UK): NIHR Journals Library (Health Technology Assessment, No. 22.23), 2018. [DOI: 10.3310/hta22230] [PMID: 29737274]PMC5960819 doi:10.3310/hta22230 PubMed
144. GailyE, GranströmML, HiilesmaaV, BardyA. Minor anomalies in offspring of epileptic mothers. Journal of Pediatrics1988;112(4):520-9. [PMID: 3351676]10.1016/s0022-3476(88)80162-8 doi:10.1016/s0022-3476(88)80162-8 PubMed
145. GailyE. Distal phalangeal hypoplasia in children with prenatal phenytoin exposure: results of a controlled anthropometric study. American Journal of Medical Genetics1990;35(4):574-8. [PMID: 2333888]10.1002/ajmg.1320350425 doi:10.1002/ajmg.1320350425 PubMed
146. GailyEK, GranströmML, HiilesmaaVK, BardyAH. Head circumference in children of epileptic mothers: contributions of drug exposure and genetic background. Epilepsy Research1990;5(3):217-22. [PMID: 2384077]10.1016/0920-1211(90)90041-s doi:10.1016/0920-1211(90)90041-s PubMed
147. HiilesmaaVK, BardyA, TeramoK. Obstetric outcome in women with epilepsy. American Journal of Obstetrics and Gynecology1985;152(5):499-504. [PMID: 4014343]10.1016/0002-9378(85)90615-5 doi:10.1016/0002-9378(85)90615-5 PubMed
148. HiilesmaaVK, TeramoK, GranströmML, BardyAH. Serum folate concentrations during pregnancy in women with epilepsy: relation to antiepileptic drug concentrations, number of seizures, and fetal outcome. British Medical Journal1983;287(6392):577-9. [PMID: 6411231]10.1136/bmj.287.6392.577PMC1549018 doi:10.1136/bmj.287.6392.577 PubMed
149. FujiiH, GoelA, BernardN, PistelliA, YatesLM, StephensS, et al. Pregnancy outcomes following gabapentin use: results of a prospective comparative cohort study. Neurology2013;80(17):1565-70. [PMID: 23553472]10.1212/WNL.0b013e31828f18c1PMC3662323 doi:10.1212/WNL.0b013e31828f18c1 PubMed
150. GalappatthyP, LiyanageCK, LucasMN, JayasekaraDTLM, AbhayaratnaSA, WeeraratneC, et al. Obstetric outcomes and effects on babies born to women treated for epilepsy during pregnancy in a resource limited setting: a comparative cohort study. BMC Pregnancy & Childbirth2018;18(1):230. [PMID: 29898689]10.1186/s12884-018-1857-3PMC6000926 doi:10.1186/s12884-018-1857-3 PubMed
151. GoujardJ, HuelG, Rumeau-RouquetteC. Antiepileptics and congenital malformations [Antiepileptiques et malformations congenitales]. Journal de Gynécologie, Obstétrique et Biologie de la Reproduction1974;3(6):831-42. [PMID: 4470757] PubMed
152. HillRM, VerniaudWM, HorningMG, McCulleyLB, MorganNF. Infants exposed in utero to antiepileptic drugs: a prospective study. American Journal of Diseases of Children1974;127(5):645-53. [PMID: 4825584]10.1001/archpedi.1974.02110240031002 doi:10.1001/archpedi.1974.02110240031002 PubMed
153. HolmesLB, HarveyEA, BrownKS, HayesAM, KhoshbinS. Anticonvulsant teratogenesis: 1. A study design for newborn infants. Teratology1994;49(3):202-7. [PMID: 8059427]10.1002/tera.1420490316 doi:10.1002/tera.1420490316 PubMed
154. JacobsenPE, HenriksenTB, HaubekD, OstergaardJR. Developmental enamel defects in children prenatally exposed to anti-epileptic drugs. PLOS One2013;8(3):e58213. [PMID: 23520494]10.1371/journal.pone.0058213PMC3592922 doi:10.1371/journal.pone.0058213 PubMed
155. JacobsenPE, HenriksenTB, HaubekD, OstergaardJR. Prenatal exposure to antiepileptic drugs and dental agenesis. PLOS One2014;9(1):e84420. [PMID: 24416231]10.1371/journal.pone.0084420PMC3885552 doi:10.1371/journal.pone.0084420 PubMed
156. JedrzejczakJ, Majkowska-ZwolinskaB. Clinical predictors for breastfeeding initiation among women with epilepsy. Seizure2022;96:59-65. [PMID: 35123234]10.1016/j.seizure.2022.01.013 doi:10.1016/j.seizure.2022.01.013 PubMed
157. JonesKL, LacroRV, JohnsonKA, AdamsJ. Pattern of malformations in the children of women treated with carbamazepine during pregnancy. New England Journal of Medicine1989;320(25):1661-6. [PMID: 2725616]10.1056/NEJM198906223202505 doi:10.1056/NEJM198906223202505 PubMed
158. KnightAH, RhindEG. Epilepsy and pregnancy: a study of 153 pregnancies in 59 patients. Epilepsia1975;16(1):99-110. [PMID: 804404]10.1111/j.1528-1157.1975.tb04726.x doi:10.1111/j.1528-1157.1975.tb04726.x PubMed
159. CunningtonM, FerberS, QuarteyG, International Lamotrigine Pregnancy Registry Scientific Advisory Committee. Effect of dose on the frequency of major birth defects following fetal exposure to lamotrigine monotherapy in an international observational study. Epilepsia2007;48(6):1207-10. [PMID: 17381445]10.1111/j.1528-1167.2007.01021.x doi:10.1111/j.1528-1167.2007.01021.x PubMed
160. CunningtonMC. The International Lamotrigine Pregnancy Registry update for the Epilepsy Foundation. Epilepsia2004;45(11):1468. [PMID: 15509254]10.1111/j.0013-9580.2004.451105.x doi:10.1111/j.0013-9580.2004.451105.x PubMed
161. Lamotrigine Pregnancy Registry. Final report: 1 September 1992–31 March 2010. pregnancyregistry.gsk.com/documents/lam_spring_2010_final_report.pdf 2010 (accessed prior to 4 Feb 2023).
162. TennisP, EldridgeRR, International Lamotrigine Pregnancy Registry Scientific Advisory Committee. Preliminary results on pregnancy outcomes in women using lamotrigine. Epilepsia2002;43(10):1161-7. [PMID: 12366730]10.1046/j.1528-1157.2002.45901.x doi:10.1046/j.1528-1157.2002.45901.x PubMed
163. LaskowskaM, Leszczyńska-GorzelakB, OleszczukJ. Evaluation of antiepileptic therapy during pregnancy [Ocena terapii przeciwpadaczkowej w okresie ciazy]. Ginekologia Polska2002;73(1):35-42. [PMID: 12001761] PubMed
164. MiskovS, Gjergja-JuraskiR, Cvitanovic-SojatL, BakulicTI, FucicA, Bosnjak-PasicM, et al. Prospective surveillance of Croatian pregnant women on lamotrigine monotherapy - aspects of pre-pregnancy counseling and drug monitoring. Acta Clinica Croatica2009;48(3):271-81. [PMID: 20055248] PubMed
165. MonsonRR, RosenbergL, HartzSC, ShapiroS, HeinonenOP, SloneD. Diphenylhydantoin and selected congenital malformations. New England Journal of Medicine1973;289(20):1049-52. [PMID: 4742220]10.1056/NEJM197311152892001 doi:10.1056/NEJM197311152892001 PubMed
166. MontourisG. Gabapentin exposure in human pregnancy: results from the Gabapentin Pregnancy Registry. Epilepsy & Behavior2003;4(3):310-7. [PMID: 12791334]10.1016/s1525-5050(03)00110-0 doi:10.1016/s1525-5050(03)00110-0 PubMed
167. MostacciB, BisulliF, PoluzziE, CocchiG, PiccinniC, CurtiA, et al. Emilia-Romagna Study on Pregnancy and Exposure to Antiepileptic drugs (ESPEA): a population-based study on prescription patterns, pregnancy outcomes and fetal health. Journal of Neurology, Neurosurgery & Psychiatry2018;89(9):983-8. [PMID: 29549194]10.1136/jnnp-2017-317833PMC6109238 doi:10.1136/jnnp-2017-317833 PubMed
168. MostacciB, PiccinniC, BisulliF, PoluzziE, NaldiI, AccettaG, et al. Prevalence of antiepileptic drugs exposure in pregnant women in the Emilia Romagna region (Italy): results from the ESPEA (Emilia Romagna Study on Pregnancy and Exposure to Antiepileptic Drugs). Epilepsia2014;55(Suppl 2):131-2, Abstract no: p400.
169. MurasakiO, YoshitakeK, TachikiH, NakaneY, KanekoS. Reexamination of the teratological effect of antiepileptic drugs. Japanese Journal of Psychiatry and Neurology1988;42(3):592-3. [PMID: 3241485]10.1111/j.1440-1819.1988.tb01370.x doi:10.1111/j.1440-1819.1988.tb01370.x PubMed
170. NakaneY, OkumaT, TakahashiR, SatoY, WadaT, SatoT, et al. Multi-institutional study on the teratogenicity and fetal toxicity of antiepileptic drugs: a report of a collaborative study group in Japan. Epilepsia1980;21(6):663-80. [PMID: 7439133]10.1111/j.1528-1157.1980.tb04320.x doi:10.1111/j.1528-1157.1980.tb04320.x PubMed
171. NakaneY. Congenital malformation among infants of epileptic mothers treated during pregnancy: the report of a collaborative study group in Japan. Folia Psychiatrica et Neurologica Japonica1979;33(3):363-9. [PMID: 520954]10.1111/j.1440-1819.1979.tb00771.x doi:10.1111/j.1440-1819.1979.tb00771.x PubMed
172. PearseSB, Garcia RodriguezLA, HartwellC, RussellG. A pregnancy register of patients receiving carbamazepine in the UK. Pharmacoepidemiology and Drug Safety1992;1(6):321-5. [DOI: 10.1002/pds.2630010603] doi:10.1002/pds.2630010603
173. RichmondJR, KrishnamoorthyP, AndermannE, BenjaminA. Epilepsy and pregnancy: an obstetric perspective. American Journal of Obstetrics and Gynecology2004;190(2):371-9. [PMID: 14981376]10.1016/j.ajog.2003.09.020 doi:10.1016/j.ajog.2003.09.020 PubMed
174. RobertE, RobertJM, LaprasC. Is valproic acid teratogenic? [L'acide valproïque est-il tératogène?]. Revue Neurologique1983;139(6-7):445-7. [PMID: 6412342] PubMed
175. SabersA, DamM, A-Rogvi-HansenB, BoasJ, SideniusP, Laue FriisM, et al. Epilepsy and pregnancy: lamotrigine as main drug used. Acta Neurologica Scandinavica2004;109(1):9-13. 10.1034/j.1600-0404.2003.00200.x14653845 doi:10.1034/j.1600-0404.2003.00200.x PubMed
176. ScheuerleAE, HolmesLB, AlbanoJD, BadalamentiV, BattinoD, CovingtonD, et al. Levetiracetam Pregnancy Registry: final results and a review of the impact of registry methodology and definitions on the prevalence of major congenital malformations. Birth Defects Research2019;111(13):872-87. [PMID: 31124321]10.1002/bdr2.1526 doi:10.1002/bdr2.1526 PubMed
177. ShapiroS, HartzSC, SiskindV, MitchellAA, SloneD, RosenbergL, et al. Anticonvulsants and parental epilepsy in the development of birth defects. Lancet1976;1(7954):272-5. [PMID: 55587]10.1016/s0140-6736(76)91403-3 doi:10.1016/s0140-6736(76)91403-3 PubMed
178. Starveld-ZimmermanA, Van der KolkW, ElshoveJ, MeinardiH. Teratogenicity of antiepileptic drugs. Clinical Neurology and Neurosurgery1975;77(2):81-95. [PMID: 1132213]10.1016/s0303-8467(74)80001-6 doi:10.1016/s0303-8467(74)80001-6 PubMed
179. TennisP, ChanKA, CurkendallSM, Li D-K, MinesD, PetersonC, et al. Topiramate use during pregnancy and major congenital malformations in multiple populations. Birth Defects Research2015;103(4):269-75. [PMID: 25776342]10.1002/bdra.23357 doi:10.1002/bdra.23357 PubMed
180. TorresLC, FélixR, CanúnS, MazónJJ. Epilepsy and pregnancy: risks and benefits of anticonvulsant treatments [Epilepsia y embarazo. Riegos y beneficios del tratamiento anticonvulsivo]. Ginecologia y Obstetricia de Mexico1995;63:282-6. [PMID: 7665113] PubMed
181. WideK, WinbladhB, TomsonT, Sars-ZimmerK, BerggrenE. Psychomotor development and minor anomalies in children exposed to antiepileptic drugs in utero: a prospective population-based study. Developmental Medicine and Child Neurology2000;42(2):87-92. [PMID: 10698324]10.1017/s0012162200000177 doi:10.1017/s0012162200000177 PubMed
182. YehCC, LussierEC, SunYT, LanTY, YuHY, ChangTY. Antiepileptic drug use among women from the Taiwanese Registry of Epilepsy and Pregnancy: obstetric complications and fetal malformation outcomes. PLOS One2017;12(12):e0189497. [PMID: 29253023]10.1371/journal.pone.0189497PMC5734752 doi:10.1371/journal.pone.0189497 PubMed
183. YerbyMS, LeavittA, EricksonDM, McCormickKB, LoewensonRB, SellsCJ, et al. Antiepileptics and the development of congenital anomalies. Neurology1992;42(4 Suppl 5):132-40. [PMID: 1574169] PubMed
184. BabicN, JovicM. Postnatal concerns in children born to women with juvenile myoclonic epilepsy. Epilepsia2014;55(Suppl 2):128, Abstract no: p389.
185. KaabiW, El AidliS, KastalliS, LakhouaG, ZaiemA, SrairiS, et al. Pregnancy outcomes in women using antiepileptic drugs. Drug Safety2013;36(9):844, Abstract no: ISP3556-47.
186. KutluG, ErdalA, AydoganS, GomceliYB, InanLE. Follow up and treatment of women with epilepsy during pregnancy. Epilepsia2013;54(Suppl 3):127-8, Abstract no: P399.
187. Lazzaroni FossatiF, De ToniT, MagnaniM, RepettoE, CalviA, Di SienaG. Intrauterine exposure to drugs: analysis of a sample of newborn infants pretreated with anticonvulsants [Esposizione in utero a farmaci. Analisi di un campione di neonati pretrattati con anticonvulsivanti]. Minerva Pediatrica1986;38(3-4):75-81. [PMID: 3702846] PubMed
188. MidiI, BulutB, OzbekD, OzdenHO, AganK. Antiepileptic drug usage and the effects of them on the foetus in epileptic pregnant woman. Epilepsia2014;55(Suppl 2):132, Abstract no: p401.
189. MidiI, CetinkayaDO, OzdenHO, AganK. Pregnant women with epilepsy: 43 patients results in 1 year period. Epilepsia2013;54(Suppl 3):131, Abstract no: P410.
190. ShvartzmanP, OrenB, KeinanA, AdarH. [Congenital malformations and anticonvulsant therapy in pregnancy]. Harefuah1986;110(8):377-80. [PMID: 3744149] PubMed
191. VlasovP, PetrukhinV, KarlovV, KrasnopolskiV, MelnikovA, TsivtsivadzeE. Antiepileptic drug therapy during pregnancy and obstetric outcomes in Moscow region: comparing of 1998 and 2013 years. Epilepsia2014;55(Suppl 2):130, Abstract no: p396.
192. AckersR, BesagFM, WadeA, MurrayML, WongIC. Changing trends in antiepileptic drug prescribing in girls of child-bearing potential. Archives of Disease in Childhood2009;94(6):443–7. [PMID: 19307197]10.1136/adc.2008.144386 doi:10.1136/adc.2008.144386 PubMed
193. AlsaadAM, ChaudhrySA, KorenG. First trimester exposure to topiramate and the risk of oral clefts in the offspring: a systematic review and meta-analysis. Reproductive Toxicology2015;53:45-50. [PMID: 25797654]10.1016/j.reprotox.2015.03.003 doi:10.1016/j.reprotox.2015.03.003 PubMed
194. ArdingerHH, AtkinJF, BlackstonRD, ElsasLJ, ClarrenSK, LivingstoneS, et al. Verification of the fetal valproate syndrome phenotype. American Journal of Medical Genetics1988;29(1):171-85. [PMID: 3125743]10.1002/ajmg.1320290123 doi:10.1002/ajmg.1320290123 PubMed
195. BrentRL. Environmental causes of human congenital malformations: the pediatrician's role in dealing with these complex clinical problems caused by a multiplicity of environmental and genetic factors. Pediatrics2004;113(4 Suppl):957-68. [PMID: 15060188] PubMed
196. BromleyR, WestonJ, AdabN, GreenhalghJ, SannitiA, McKayAJ, et al. Treatment for epilepsy in pregnancy: neurodevelopmental outcomes in the child. Cochrane Database of Systematic Reviews2014, Issue 10. Art. No: CD010236. [DOI: 10.1002/14651858.CD010236.pub2]PMC739002025354543 doi:10.1002/14651858.CD010236.pub2 PubMed
197. CharltonRA, CunningtonMC, De VriesCS, WeilJG. Data resources for investigating drug exposure during pregnancy and associated outcomes: the General Practice Research Database (GPRD) as an alternative to pregnancy registries. Drug Safety2008;31(1):39-51. [PMID: 18095745]10.2165/00002018-200831010-00004 doi:10.2165/00002018-200831010-00004 PubMed
198. ChaudhrySA, JongG, KorenG. The fetal safety of levetiracetam: a systematic review. Reproductive Toxicology2014;46:40-5. [PMID: 24602560]10.1016/j.reprotox.2014.02.004 doi:10.1016/j.reprotox.2014.02.004 PubMed
199. ChristensenJ, GrønborgTK, SørensenMJ, SchendelD, ParnerET, PedersenLH, et al. Prenatal valproate exposure and risk of autism spectrum disorders and childhood autism. JAMA2013;309(16):1696-703. [PMID: 23613074]10.1001/jama.2013.2270PMC4511955 doi:10.1001/jama.2013.2270 PubMed
200. ChristensenJ, PedersenL, SunY, DreierJW, BrikellI, DalsgaardS. Association of prenatal exposure to valproate and other antiepileptic drugs with risk for attention-deficit/hyperactivity disorder in offspring. JAMA Network Open2019;2(1):e186606. [PMID: 30646190]10.1001/jamanetworkopen.2018.6606PMC6324310 doi:10.1001/jamanetworkopen.2018.6606 PubMed
201. Clayton-SmithJ, BromleyR, DeanJ, JournelH, OdentS, WoodA, et al. Diagnosis and management of individuals with Fetal Valproate Spectrum Disorder; a consensus statement from the European Reference Network for Congenital Malformations and Intellectual Disability. Orphanet Journal of Rare Diseases2019;14(1):180. [PMID: 31324220]10.1186/s13023-019-1064-yPMC6642533 doi:10.1186/s13023-019-1064-y PubMed
202. DeanJC, MooreSJ, TurnpennyPD. Developing diagnostic criteria for the fetal anticonvulsant syndromes. Seizure2000;9(3):233–4. [PMID: 10775521]10.1053/seiz.2000.0392 doi:10.1053/seiz.2000.0392 PubMed
203. DiLibertiJH, FarndonPA, DennisNR, CurryCJ. The fetal valproate syndrome. American Journal of Medical Genetics1984;19(3):473-81. [PMID: 6439041]10.1002/ajmg.1320190308 doi:10.1002/ajmg.1320190308 PubMed
204. EUROCAT European Surveillance of Congenital Anomalies. EUROCAT guide 1.3 and reference documents: instructions for the registration and surveillance of congenital anomalies. eu-rd-platform.jrc.ec.europa.eu/sites/default/files/EUROCAT-Guide-1.3.pdf (accessed August 2015).
205. FiestKM, SauroKM, WiebeS, PattenSB, KwonCS, DykemanJ, et al. Prevalence and incidence of epilepsy: a systematic review and meta-analysis of international studies. Neurology2017;88(3):296-303. [PMID: 27986877]10.1212/WNL.0000000000003509PMC5272794 doi:10.1212/WNL.0000000000003509 PubMed
206. GuyattGH, OxmanAD, VistG, KunzR, Falck-YtterY, Alonso-CoelloP, et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ2008;336(7650):924-6. [PMID: 18436948]10.1136/bmj.39489.470347.ADPMC2335261 doi:10.1136/bmj.39489.470347.AD PubMed
207. HigginsJPT, Green S (editors). Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration, 2011. Available from training.cochrane.org/handbook/archive/v5.1/.
208. JentinkJ, LoaneMA, DolkH, BarisicI, GarneE, MorrisJK, et al. Valproic acid monotherapy in pregnancy and major congenital malformations. New England Journal of Medicine2010;362(23):2185-93. [PMID: 20558369]10.1056/NEJMoa0907328 doi:10.1056/NEJMoa0907328 PubMed
209. JentinkJ, DolkH, LoaneMA, MorrisJK, WellesleyD, GarneE, et al. Intrauterine exposure to carbamazepine and specific congenital malformations: systematic review and case-control study. BMJ2010;341:c6581. [PMID: 21127116]10.1136/bmj.c6581PMC2996546 doi:10.1136/bmj.c6581 PubMed
210. KuCS, NaidooN, WuM, SoongR. Studying the epigenome using next generation sequencing. Journal of Medical Genetics2011;48(11):721-30. [PMID: 21825079]10.1136/jmedgenet-2011-100242 doi:10.1136/jmedgenet-2011-100242 PubMed
211. ManSL, PetersenI, ThompsonM, NazarethI. Antiepileptic drugs during pregnancy in primary care: a UK population based study. PLOS One2012;7:e52339. [PMID: 23272239]10.1371/journal.pone.0052339PMC3525559 doi:10.1371/journal.pone.0052339 PubMed
212. MargulisAV, MitchellAA, GilboaSM, WerlerMM, MittlemanMA, GlynnRJ, et al, National Birth Defects Prevention Study. Use of topiramate in pregnancy and risk of oral clefts. American Journal of Obstetrics and Gynecology2012;207(5):405.e1-7. [PMID: 22917484]10.1016/j.ajog.2012.07.008PMC3484193 doi:10.1016/j.ajog.2012.07.008 PubMed
213. MarsonAG, Al-KharusiAM, AlwaidhM, AppletonR, BakerGA, ChadwickDW, et al, SANAD Study group. The SANAD study of effectiveness of valproate, lamotrigine, or topiramate for generalised and unclassifiable epilepsy: an unblinded randomised controlled trial. Lancet2007;369(9566):1016-26. [PMID: 17382828]10.1016/S0140-6736(07)60461-9PMC2039891 doi:10.1016/S0140-6736(07)60461-9 PubMed
214. MeadorK, ReynoldsMW, CreanS, FahrbachK, ProbstC. Pregnancy outcomes in women with epilepsy: a systematic review and meta-analysis of published pregnancy registries and cohorts. Epilepsy Research2008;81(1):1-13. [PMID: 18565732]10.1016/j.eplepsyres.2008.04.022PMC2660205 doi:10.1016/j.eplepsyres.2008.04.022 PubMed
215. MeadorKJ, PenovichP, BakerGA, PennellPB, BromfieldE, PackA, et al, NEAD Study Group. Antiepileptic drug use in women of childbearing age. Epilepsy and Behaviour2009;15(3):339-43. [PMID: 19410654]10.1016/j.yebeh.2009.04.026PMC2741411 doi:10.1016/j.yebeh.2009.04.026 PubMed
216. MinesD, TennisP, CurkendallSM, LiDK, PetersonC, AndrewsEB, et al. Topiramate use in pregnancy and the birth prevalence of oral clefts. Pharmacoepidemiology and Drug Safety2014;23(10):1017–25. [PMID: 24692316]10.1002/pds.3612 doi:10.1002/pds.3612 PubMed
217. National Institute for Clinical Excellence (NICE). Epilepsies in children, young people and adults. NICE guideline [NG217]. www.nice.org.uk/guidance/ng217 (accessed prior to 4 Feb 2023).
218. OommenKJ, MathewsS. Zonisamide: a new antiepileptic drug. Clinical Neuropharmacology1999;22(4):192-200. [PMID: 10442247] PubMed
219. SterneJA, HernánMA, ReevesBC, SavovićJ, BerkmanND, ViswanathanM, et al. ROBINS-I: a tool for assessing risk of bias in non-randomised studies of interventions. BMJ2016;355:i4919. [PMID: 27733354]10.1136/bmj.i4919PMC5062054 doi:10.1136/bmj.i4919 PubMed
220. TetroN, MoushaevS, ShmuelM, EyalS. Antiseizure medications and fetal nutrients: effects on choline transporters in a human placental cell line. Epilepsia2021;62(6):1451-9. [PMID: 33890297]10.1111/epi.16905 doi:10.1111/epi.16905 PubMed
221. TomsonT, BattinoD, BonizzoniE, CraigJ, LindhoutD, SabersA, et al, EURAP study group. Dose-dependent risk of malformations with antiepileptic drugs: an analysis of data from the EURAP epilepsy and pregnancy registry. Lancet Neurology2011;10(7):609-17. [PMID: 21652013]10.1016/S1474-4422(11)70107-7 doi:10.1016/S1474-4422(11)70107-7 PubMed
222. TomsonT, MarsonA, BoonP, CaneviniMP, CovanisA, GailyE, et al. Valproate in the treatment of epilepsy in girls and women of childbearing potential. Epilepsia2015;56(7):1006-19. [PMID: 25851171]10.1111/epi.13021 doi:10.1111/epi.13021 PubMed
223. VeronikiAA, CogoE, RiosP, StrausSE, FinkelsteinY, KealeyR, et al. Comparative safety of anti-epileptic drugs during pregnancy: a systematic review and network meta-analysis of congenital malformations and prenatal outcomes. BMC Medicine2017;15(1):95. [PMID: 28472982]10.1186/s12916-017-0845-1PMC5418725 doi:10.1186/s12916-017-0845-1 PubMed
224. WenX, MeadorKJ, HartzemaA. Antiepileptic drug use by pregnant women enrolled in Florida Medicaid. Neurology2015;84(9):944-50. [PMID: 25653296]10.1212/WNL.0000000000001304PMC4351665 doi:10.1212/WNL.0000000000001304 PubMed
225. AdabN, TudurSC, VintenJ, WilliamsonP, WinterbottomJ. Common antiepileptic drugs in pregnancy in women with epilepsy. Cochrane Database of Systematic Reviews2004, Issue 3. Art. No: CD004848. [DOI: 10.1002/14651858.CD004848]15266543 doi:10.1002/14651858.CD004848 PubMed
226. PulmanJ, BromleyR, AdabN, GreenhalghJ, McKayAJ, Tudur SmithC, et al. Treatment for epilepsy in pregnancy: congenital malformation outcomes in the child. Cochrane Database of Systematic Reviews2012, Issue 11. Art. No: CD010224. [DOI: 10.1002/14651858.CD010224]PMC646505527819746 doi:10.1002/14651858.CD010224 PubMed
227. WestonJ, BromleyR, JacksonCF, AdabN, Clayton-SmithJ, GreenhalghJ, et al. Monotherapy treatment of epilepsy in pregnancy: congenital malformation outcomes in the child. Cochrane Database of Systematic Reviews2016, Issue 11. Art. No: CD010224. [DOI: 10.1002/14651858.CD010224.pub2]PMC646505527819746 doi:10.1002/14651858.CD010224.pub2 PubMed