Comparison of neurodevelopmental, educational and adult socioeconomic outcomes in offspring of women with and without epilepsy: A systematic review and meta-analysis

Background : Adequate pre-pregnancy counselling and education planning are essential to improve outcomes for offspring of women with epilepsy (OWWE). The current systematic review and meta-analysis aimed to compare outcomes for OWWE and offspring of women without epilepsy (OWWoE). Methods : We conducted a systematic review and meta-analysis. We searched MEDLINE, EMBASE, CINAHL, PsycINFO (database inception-1 st January 2023), OpenGrey, GoogleScholar, and hand-searched journals and reference lists of included studies to identify eligible studies. We placed no language restrictions and included observational studies concerning OWWE and OWWoE. We followed the PRIMSA checklist for abstracting data. The Newcastle-Ottawa Scale for risk of bias assessment was conducted independently by two authors with mediation by a third. We report pooled unadjusted odds ratios (OR) or mean differences (MD) with 95% confidence intervals (95CI) from random (I 2 >50%) or fixed (I 2 <50%) effects meta-analyses. Outcomes of interest included offspring autism, attention deficit/hyperactive disorder, intellectual disability, epilepsy, developmental disorder, intelligence, educational, and adulthood socioeconomic outcomes.


Highlight
• Anti-seizure medication burden increases risk of neurodevelopmental disorders and special educational needs in offspring of women with epilepsy.
• Offspring of women with epilepsy may have worse developmental and or educational outcomes than offspring of women without epilepsy, even if there was no ASM exposure during pregnancy but further work is needed.
• Women with epilepsy receive pre-pregnancy counselling soon after diagnosis and regularly during management.

Key Points
Anti-seizure medication burden increases risk of neurodevelopmental disorders and special educational needs in offspring of women with epilepsy.
Offspring of women with epilepsy may have worse developmental and or educational outcomes than offspring of women without epilepsy, even if there was no ASM exposure during pregnancy but further work is needed.
Women with epilepsy should receive pre-pregnancy counselling soon after diagnosis and regularly during management.

Introduction
Most studies on outcomes of offspring of women with epilepsy (OWWE), are focused on neurodevelopmental outcome, but there is a paucity of data on educational and adulthood socioeconomic outcomes (1)(2)(3)(4).A Cochrane review in 2014 (2) found that prenatal (during pregnancy) anti-seizure medication (ASM) exposure was associated with an increased risk of neurodevelopmental disorders, but the review was restricted to randomised controlled trials and cohort studies.Other observational study types aside from cohort studies, such as casecontrol and or cross-sectional studies, can provide valuable data (4) and some major studies on the topic have been published since the Cochrane review (1,3,(5)(6)(7)(8)(9).
Studies on outcomes of OWWE tend to focus solely on exposure to specific prenatal ASM, with some, such as valproate and topiramate, conferring higher risk [5][6][7][8].However, there is evidence there may still be an increased risk for adverse offspring outcomes even if there is no prenatal ASM exposure (1,10,11).A recent systematic review and meta-analysis of perinatal outcomes of women with epilepsy themselves, found that even when women with epilepsy were not on prenatal ASM, perinatal outcomes were still worse than women without epilepsy (4).These data lead to the hypothesis that adverse neurodevelopmental outcomes experienced by OWWE are not due only to prenatal ASM exposure but there is a role played by the epilepsy itself.We also hypothesise that prenatal ASM burden in women with epilepsy contributes to adverse outcomes of their offspring.Although there is a body of evidence that generally points towards outcomes being worse in the OWWE compared to that of offspring of women without epilepsy (OWWoE), to our knowledge there is no recent systematic review nor meta-analysis to quantify risk on the neurodevelopmental, educational, and socioeconomic outcomes of OWWE.
The current study aimed to investigate: 1. Neurodevelopmental, educational, and socioeconomic outcomes in offspring of women with compared to without epilepsy regardless of ASM exposure.

Effect of ASM burden on offspring outcomes through multiple comparisons each
addressing a different level of ASM exposure: a. women with epilepsy not on prenatal ASM versus those of women without epilepsy; b. women with epilepsy on prenatal ASM versus those of women with epilepsy not on prenatal ASM; c. women with epilepsy on prenatal ASM polytherapy compared to those who were on prenatal ASM monotherapy Assessing associations of individual ASMs or specific combinations was not an aim of the current study.

Methods
We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (12) (Supplemental Material -PRISMA checklist).

Search strategy and study selection
The review was registered on the International Register of Systematic Review Protocols (PROSPERO-www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42020221100).
The search strategy is provided in Supplemental Material, Appendix 1. Study selection and analysis methods, including assessment of risk of bias have been previously described (4).

Outcomes of interest
Neurodevelopmental outcomes: Autism Spectrum Disorder, Attention Deficit/Hyperactive Disorder, Intellectual Disabilities, Epilepsy, Developmental disorder, Intelligence.
Educational outcomes: Special education needs, Educational achievement.

Data analysis
Studies similar in study design, outcomes investigated, and compared OWWE and OWWoE were combined in Review Manager 5 (RevMan) (13).We applied the Mantel-Haenszel method to estimate pooled unadjusted odds ratio (OR) or mean difference (MD) with a 95% confidence interval (CI) for each outcome.The I 2 statistic was used to assess inter-study heterogeneity.We used a random-effects model for studies with I² >= 50% and a fixedeffects model for those < 50%, with sensitivity analyses to determine the impact of model choice on effect estimates.If a study pooled data from individual studies, individual studies were not included to prevent double counting.Where articles used the same or overlapping time points assessing an outcome on the same participant population, we included the article with the largest total sample and longest time period in the meta-analysis (Supplemental material, Appendix 2. eTable 1).Where studies did not directly compare OWWE and OWWoE, but there were raw data available, they were included in a meta-analysis.
Additional meta-analyses were performed to address the effect of ASM burden as detailed in our specific study aims.Subgroup analyses were used to investigate risk of bias on effect estimates.Unadjusted pooled risk differences (RD) with 95% CI were determined for all binary outcomes, using fixed or random effects meta-analysis according to the corresponding odds ratio (OR) analysis.When only one article was identified, unadjusted analysis was conducted.
*Consider, if feasible to do so, reporting the number of records identified from each database or register searched (rather than the total number across all databases/registers). **If automation tools were used, indicate how many records were excluded by a human and how many were excluded by automation tools.

Discussion
For ethical reasons, assessment of the outcomes of OWWE compared to the OWWoE and the role of ASM burden must rest on observational studies rather than randomised trials.
Therefore, the findings of this systematic review and meta-analysis of the neurodevelopmental, educational, and socioeconomic outcomes of the OWWE were restricted to observational studies and complement and supplement our recent systematic review and meta-analysis of the perinatal outcomes of women with epilepsy (4).Our main findings in the current study are (1) there is a paucity of information on the topic with some outcomes having no reported studies or only single studies, and in some instances, only a few studies with modest sample sizes thereby limiting confidence on study findings; (2) OWWE seem to have higher unadjusted odds of autism spectrum disorders, attention deficit/hyperactivity disorder, intellectual disability and special educational needs and have reduced mean intelligence and development quotients compared to OWWoE, but further work is needed adjusting for potential confounders; (3) generally, there is an impression that outcomes are worse with greater prenatal ASM burden.Individual ASMs were not investigated, and it is possible that ASMs with a known higher risk for poor outcomes, such as sodium valproate and or topiramate, could be inflating these negative outcomes.

OWWE compared to OWWoE
We acknowledge there may be questions surrounding our finding that OWWE have worse outcomes compared to those of women without epilepsy.This is because in our unadjusted pooled estimates we were unable to for potential epilepsy and social confounders such as epilepsy aetiology and prenatal smoking (31)(32)(33).Notwithstanding, the finding of increased odds of ASD and ID is mainly from data from a large comprehensive cohort study of combined health data on 4.5 million children from Nordic countries between 1996 and 2017 (1).In that study, increased odds remained even after adjustment for a large number of potential confounders (1).Similarly, the finding of increased odds of ADHD is mainly from data from one large Danish study (3)

ASM burden
The finding that OWWE who were not on prenatal ASM were still at increased odds of having ASD, ADHD, and ID suggests that prenatal ASM exposure may not be the sole factor associated with these worse outcomes; other factors need to be considered.We could not determine whether women with epilepsy had been taking ASM pre-pregnancy and if so, for what duration and or dose.This is relevant given evidence of a potential transgenerational association between some ASMs, such as sodium valproate, and neurodevelopmental disorders (1,34,35).Potential epilepsy confounding/effect-modifying factors could be a further alternative explanation but we were unable to account for these.For example, individuals with epilepsy are significantly more likely to have autism, ADHD and cognitive difficulties which on their own could put their children at higher risk for such conditions (36,37).Future studies trying to tease out the effect of ASM would benefit from intergroup comparisons amongst women without epilepsy on prenatal ASM (e.g.women on valproate for mood stabilisation rather than anti-seizure effects), women with epilepsy on ASM, and women without any underlying neurological condition and not on ASMs.Further, studies adjusting for potential confounders such as those mentioned above, epilepsy severity and maternal history of learning and or behavioural problems would clarify risk factors that could be targets for intervention.
Lack of analysis on specific prenatal ASM exposure is a major limitation of the current study.
It is the most likely explanation for the finding of no differences in IQs whilst in the same intergroup comparison there was an increased odds of intellectual disability.The two studies that contributed to IQ analyses had modest sample sizes (17,19), they were older studies and did not include valproate or topiramate which are two of the most well-known prenatal ASMs associated with worse outcomes.The increased odds of ID were largely based on the recent combined Nordic study mentioned above(1) that included analyses on a large number of ASMs including the "bad actors" valproate and topiramate.In the Nordic study, prenatal carbamazepine was not associated with offspring ID, which would be more in keeping with the normal IQ found in the studies analysed.Future work on the effect of individual and combination prenatal ASM, especially the newer ASMs, is urgently needed.

Strengths and Limitations
This review was based on an a priori protocol, broad search terms, and stringent inclusion and exclusion criteria.We conducted a quality assessment of the included articles and have provided up-to-date ORs and MDs with 95CIs, complementing a Cochrane review on the topic (2).We also provide novel data on outcomes of OWWE not on ASM in pregnancy and have identified substantial gaps in the literature.We undertook meta-analyses to provide objective and statistical estimates of effect, rather than to provide narrative synthesis which would have relied on subjective interpretation (38).Studies which met inclusion criteria but did not have full data available for meta-analysis or could not be combined in meta-analysis for other reason(s), for example, Meador et al. (2021) (39), were thus, excluded.Despite the comprehensive of the current study, there are other limitations besides those already mentioned.ASM use in many of the included studies is based on prescription redeeming and do not account for adherence; moreover, these studies do not always exclude other possible teratogenic medications.Given that the International League Against Epilepsy (ILAE) have stated that most women with epilepsy should be on ASMs (40), the frequency of those not on ASMs raises concern for misclassification, reporting error, or under treatment.Most of the studies included in analyses were published over a decade ago; prescriber practices, including potentially teratogenic drugs and folic acid, have since changed (41), and neurodevelopmental outcomes for OWWE today may be different (39).
It must be noted that women with epilepsy not exposed to ASMs most likely either do not have epilepsy or are undertreated based on the ILAE's 2014 "practical definition of epilepsy" (40).These possibilities could confound any conclusions.When considering exposure to polytherapy vs monotherapy there is the likelihood of differences in epilepsy severity and, therefore, that seizures in the polytherapy group are less likely to be controlled.Ongoing seizures during pregnancy would be expected to be associated with worse outcomes in the offspring.It is, therefore, not possible to determine if increased odds are driven by ASM burden or epilepsy and/or seizure severity.
A limited number of articles are included in some meta-analyses and limited sample sizes for some comparisons, which means that results for some comparisons are highly influenced by a single or small number of studies.

Conclusion
Clinicians and pregnant people with epilepsy who are considering or capable of having children should consider the findings of the current study and pay close attention to the UK national guideline (42) that pre-pregnancy counselling should be provided at the time of epilepsy diagnosis and regularly during epilepsy management.This includes preconception counselling on the risks of use during pregnancy to offspring.These current guidelines are understandably aimed at adult people with epilepsy capable of pregnancy, but it may be worth considering revising and the need for such discussions to children with epilepsy who would be capable of pregnancy.
PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses RD: Risk difference ASD: autism spectrum disorder ADHD: Attention-deficit/hyperactivity disorder SEN: Attending special education needs/school for those with a learning disability DQ: Developmental quotient FSIQ: Full-scale intelligence quotient VIQ: Verbal intelligence quotient NVIQ: Non-verbal intelligence quotient ID: Intellectual disability WPPSI: Wechsler Preschool and Primary Scale of Intelligence WISC: Wechsler Intelligence Scale for Children.ILAE: International League Against Epilepsy.

-3 and Table 2; Supplemental Material Appendix 4). Single-study analysis OWWE
using the WPPSI or WISC were consistently lower in those whose mothers were not on prenatal ASM but the differences were not statistically significant (Figures 2who were not on prenatal ASM were still at increased odds of childhood epilepsy (20) compared to those whose mothers did not have epilepsy.In one study, there was a nonstatistically significant higher point estimate of mean DQ scores on the Griffiths Mental Development Scale in OWWE who were not on prenatal ASM (14) (Figures2-3 and

Figures 2-3 and Table 2; Supplemental Material Appendix 5). Offspring of women with epilepsy on prenatal ASM polytherapy vs Offspring of women with epilepsy on prenatal ASM monotherapy
(14)le-study analysisOWWE who were on prenatal polytherapy had similar odds of childhood epilepsy(20), ID(26), and mean DQ scores on the Griffiths Mental Development Scale(14)as those whose mothers were on monotherapy (Figures2-3and Table2; Supplemental Material Appendix 5).

Table 2 . Pooled unadjusted effect estimates and p-values for all outcomes under each comparison from meta-analyses and single studies* OWWE vs OWWoE OWWE not on ASM vs OWWoE OWWE on ASM vs OWWE not on ASM OWWE polytherapy vs monotherapy
* Unadjusted OR/MD (95 CI) calculated from single-study raw data.Abbreviations: OWWE: Offspring of women with epilepsy; OWWoE: Offspring of women without epilepsy; ASM: Anti-seizure medication; ASD: Autism spectrum disorder/autism; ADHD: Attention deficit/hyperactivity disorder; ID: Intellectual disability; SEN: Special educational needs; FSIQ: Full-scale intelligence quotient; VIQ: Visual intelligence quotient; NVIQ: Non-verbal intelligence quotient; DQ: Development quotient; MDI: Mental development index; PDI: Psychomotor development index; OR: Odds ratio; MD: Mean difference; RD: Risk difference; 95CI: 95% confidence interval• with 10 million person-years of observation, and the increased odds remained after adjusting for potential confounders including maternal epilepsy, maternal age, maternal psychiatric history, maternal diabetes, sex of the child, birth year, and parity.Thus, although neither of these two high quality studies were able to adjust for epilepsy characteristics aside from whether any prenatal ASM was used, their data suggest there is a true increased odds of worse outcome in the OWWE compared to OWWoE.The limited available data outside the Nordic region highlights the need for further research.Our finding that OWWE had lower DQ on the MDI domain but not on the PDI domain of the Bayley Infant Development Scale points towards differences in cognitive versus motor development, rather than a global effect on development.Further research is needed to determine whether this is related to the epilepsy itself or to ASM exposure including to specific individual medications, combination therapy and or dosage.