AntiEpileptic drug Monitoring in PREgnancy (EMPiRE): a double-blind randomised trial on effectiveness and acceptability of monitoring strategies

Primary objective

To determine, in pregnant women with epilepsy on AEDs who experience a 25% decrease from baseline in serum AED levels, if a strategy of additional therapeutic drug monitoring compared with clinical features monitoring alone to determine the optimal dose of AED reduces the risk of seizures.

Secondary objectives

1. To determine if there is a relationship between the level of reduction in serum AED levels and seizures, by comparing women in a non-randomised cohort with stable levels with those in randomised cohorts with a decrease in levels.

2. To evaluate the effect of the two monitoring strategies on maternal and fetal outcomes in women with a decrease in serum AED levels.

3. To assess the effect of therapeutic drug monitoring versus clinical features monitoring on QoL in pregnant women with epilepsy on AEDs.

4. To identify any differences in total AED dose exposure between therapeutic drug monitoring and clinical features monitoring strategies.

5. To assess the adverse effects of AEDs in all women exposed to the drugs.

6. To gain insight into the way pregnant women with epilepsy rationalise and make sense of the management of AEDs in the context of their lives through a qualitative study.

7. To evaluate the cost-effectiveness of the two strategies.

Inclusion criteria

• Viable pregnancy of < 24 weeks’ gestation.

• Confirmed diagnosis of epilepsy including primary, localised or unclassified.

• Lamotrigine monotherapy/polytherapy (with carbamazepine, phenytoin or levetiracetam) or carbamazepine monotherapy or phenytoin monotherapy or levetiracetam monotherapy.

• Capable of understanding the information provided.

Exclusion criteria

• Aged < 16 years.

• Documented status epilepticus in the last year or non-epileptic seizures in the last 2 years.

• Non-lamotrigine polytherapy or sodium valproate monotherapy or polytherapy.

• Participation in any blinded, placebo-controlled trials of investigational medicinal products in pregnancy.

• Significant learning disability.

• Unable to complete seizure diaries or recall frequency of seizures accurately.

• History of alcohol or substance abuse or dependence in the last 2 years.

• Expressed an intention not to take AEDs in pregnancy.

Therapeutic drug monitoring group

The monthly levels of serum AED were communicated to the responsible clinicians. Clinicians managed women based on knowledge of serum AED levels in addition to clinical factors. The management involved discussion with the patient of potential risks of reduced serum levels, and the risks and benefits of an increase in AED dose to mother and baby. Women were provided treatment options including more frequent monitoring, increase in dosage of the AED immediately or delayed increase pending early testing.

Clinical features monitoring group

The clinician and mother were not informed of the serum AED levels, unless requested as part of an unblinding procedure. The decision to change the dose of AED was made by a responsible clinician based on clinical features alone.

Primary outcome

The primary outcome was seizure deterioration, which was defined as time to first seizure, including first and subsequent seizures after randomisation, over the whole period of monitoring including 6 weeks post delivery.

Secondary outcomes

Assumption checks and sensitivity analysis

Extreme values were checked as part of the data cleaning procedure. Any remaining outliers were considered to be true data values and therefore analysed as reported. However, using box plots we identified one participant with extremely large numbers of seizures. We assessed the robustness of the secondary analysis of seizure rates by excluding this value and interpreted the results accordingly. For survival models, the proportional hazards assumption was checked using Schoenfeld residuals, log–log plots and through inclusion of time-dependent effects; subgroup effects were presented to investigate violations and compared using the Wald test for treatment–covariate interactions. We investigated whether or not any treatment effect differed by seizure type by only considering tonic–clonic seizures as outcomes.

Missing values

Withdrawals and those lost to follow-up were included in the analysis up to the last point that data are available. If the number of seizures was unknown for a date or a date range, we contacted the mothers by telephone or in person to obtain missing details. When this was not possible, records were reviewed by two independent neurologists (DM and AK), who commented on the likelihood of seizure and type of seizure in the missing slot. When the neurologists were not able to provide this opinion, or there was a discrepancy in their opinion, the opinion of a third neurologist (Shanika Samarasekera) was sought. When all neurologists were unable to provide estimation on likelihood of seizure, the average seizure rate for tonic–clonic seizures and the rate for non-tonic–clonic seizures over the period of the participant’s completed diary were applied.

When the seizure type was missing or no other data for the seizure type were available, the average rate over any seizure type was used. If multiple seizures occurred during a time frame, they were equally spaced out over the time frame. If the number of seizures was larger than the number of days, the seizures were equally spaced out over each day in the time frame. A sensitivity analysis was conducted for the analysis of time to first seizure to investigate if an interval censoring approach showed a different result. When there were missing data on seizure occurrence, participants were censored at the first date on which seizure occurrence was known. A sensitivity analysis will be performed for the analysis of multiple events ignoring any dates or date ranges in which seizure occurrence is unknown.

Sensitivity analysis for interval analysis for time to first seizure

This sensitivity analysis was planned a priori but not conducted. The reasons are as follows.

The exact date of the first seizure was uncertain in eight women. Three women had substantially more seizures than the number of days in the period of uncertainty. For these women we assumed daily seizures, as the actual number of seizures occurring on a single day is irrelevant for the primary analysis.

Three women reported more than one seizure occurring during a period of 3–6 weeks. Interval censoring approaches standardly available in the statistical packages allow only one event to occur during the period of uncertainty. Allowing for multiple events during the period of uncertainty would require the application of multistate models, which would probably introduce other issues, such as convergence problems. Two women had a single seizure during 1 month. Here, the interval censoring approach standardly available could have been applied. However, as a result of the small number and the issues arising for other women (as described in the previous paragraph), it was decided not to perform this analysis.

Oversight of the trial

The management of our study included an element of expert advice that was entirely independent from the investigators and their host institution(s). The trial was overseen by a 15-member TSC, which included three independent members and a consumer representative from Epilepsy Action. There were three independent members in the five-member Data Monitoring Committee (DMC). The terms of reference and charter for the DMC were determined at the outset, taking into account issues relevant to the monitoring of this study. 22

Health economics analysis

The original sample size for the study was 660 randomised women. In 2014, given the slow rate of recruitment of the trial, the funder, after discussion with the DMC and TSC, decided not to extend the recruitment period of the trial, prior to achievement of the planned sample size. The economic analysis was integral to the initial study design, but given that the planned sample size was not recruited, it was postponed pending results to see if it was justified. Given that the study ultimately found no evidence to support that regular monitoring of AED drug levels in pregnancy offers additional benefit in seizure control than management based only on clinical features, any justification for an economic evaluation has not materialised.

Proportional hazards assumption checks

For time to first seizure, and time to multiple seizures, we did not find any differences between the subgroups based on seizure status 3 months before pregnancy, type of AED intake at baseline and the type of seizure (see Table 6).

Sensitivity analysis

We undertook a sensitivity analysis by including only women with tonic–clonic seizures. For the analysis of time to first tonic–clonic seizure, 257 women provided a total analysis time of 31,572 days from randomisation to first seizure or censoring. Table 7 shows that the risk of time to first seizure was lower in the therapeutic drug monitoring group than in the clinical features monitoring group, but this difference was not statistically significant (HR 0.80, 95% CI 0.43 to 1.5). Figure 6 shows the results of the Cox regression of time to first tonic–clonic seizure and the corresponding Kaplan-Meier curve.

FIGURE 6.

Survival graph for time to first event analysis on tonic–clonic seizure only. CFM, clinical features monitoring; TDM, therapeutic drug monitoring.

Figure 7 shows the Kaplan–Meier curve of the multiple failure time analysis of time to first tonic–clonic seizure and subsequent seizures.

FIGURE 7.

Survival graphs for time to first tonic–clonic seizure and time to subsequent tonic–clonic seizures. Time to (a) event 1; (b) event 2; (c) event 3; and (d) event 4. CFM, clinical features monitoring; TCS, tonic–clonic seizure; TDM, therapeutic drug monitoring; trt, treatment allocation.

The analysis of time to first and subsequent tonic–clonic seizures showed a decrease with therapeutic drug monitoring compared with clinical features monitoring, which was not significant (Table 8; HR 0.62, 95% CI 0.28 to 1.4).

Maternal outcomes

Fetal outcomes

There were no neonatal deaths in any of the randomised women and two stillbirths in the clinical features monitoring group. The odds of major congenital malformations, small for gestational age fetuses and admission to the neonatal unit did not differ between the two groups (Table 11). We did not observe any differences in birthweight, head circumference, Apgar scores at 1 and 5 minutes, and cord arterial and venous pH of infants born to mothers exposed to therapeutic drug monitoring or clinical features monitoring strategies. The cord blood levels of the AEDs were available for babies born to women taking lamotrigine (n = 131), carbamazepine (n = 26) and levetiracetam (n = 66). We observed a significant increase in the cord blood levels of lamotrigine [mean difference (MD) 0.55 mg/l, 95% CI 0.11 to 1.0 mg/l) and levetiracetam (MD 7.8 mg/l, 95% CI 0.86 to 14.8 mg/l) in infants born to mothers managed in the therapeutic drug monitoring group compared with in the clinical features monitoring group. There were no differences in cord blood levels of carbamazepine (MD –0.47 mg/l, 95% CI 1.5 to 0.60 mg/l) between the two groups. We quantify the effect of an increase in AED dose on fetal outcomes in Appendix 3. An increase in exposure to AED dose by 1 mg significantly increased the cord blood levels of lamotrigine, by 0.007 mg/l (see Appendix 3), levetiracetam, by 0.008 mg/l, and carbamazepine, by 0.003 mg/l, in women on AED monotherapy. The cord blood levels of lamotrigine and levetiracetam were increased by 0.009 and 0.008 mg/l, respectively, for every 1-mg increase in dose of AED in women on polytherapy. The cord blood venous pH was significantly reduced by –0.0002 per 1-unit increase in dose of carbamazepine, but there were no effects on other fetal outcomes with increasing doses of AED (see Appendix 3).

Maternal quality of life

Table 12 compares the QoL measurements in mothers exposed to the two AED monitoring strategies. There were no differences in the EQ-5D postnatal scores between the two groups (MD 0.002, 95% CI –0.05 to 0.05). The scores for the QOLIE-31 and the overall health score were similar in both groups (see Table 12).

Effect of a reduction in serum antiepileptic drug levels on maternal seizures

Table 13 compares the seizure status between women in the non-randomised group with stable serum AED levels and women in the clinical features monitoring and therapeutic drug monitoring groups with a decrease in serum AED levels of > 25%. There were no significant differences between the groups in seizure status, which was adjusted for baseline seizures in the 3 months prior to pregnancy.

Table 14 shows that there were no differences in the average number of seizures per week and the average number of days with seizures per week, analysed using Poisson models. We removed an extreme outlier who had an average of 256 seizures per week.

Background

To date, research on epilepsy and pregnancy has been largely investigated using quantitative methods, and this is reflected in evidence-based reviews covering the area. 24,25 Expert reviews26,27 and guidelines28 on the management of epilepsy in pregnancy focus on aspects of care important to health professionals. However, there is a stark absence of research concerning the priorities and perspectives of patients themselves.

A review of qualitative literature29 in this area was conducted in 2013 and found only one study30 that directly investigated women’s experiences of epilepsy during pregnancy. This ‘exploratory qualitative’ study, carried out by Thompson et al. ,30 investigated the experiences of women living with epilepsy of health-care services at key phases of reproduction, including contraception, pre-conceptual care, pregnancy, birth and breastfeeding, and parenting and child safety. Women reported mixed experiences of health care during these stages: some felt that they had received good care, but others were given inadequate information and offered advice from practitioners only after an event, and, thus, they could not take appropriate preventative action. Thompson et al. 30 argue that the management of a chronic illness and reproductive health involves work of a ‘moral dimension’. For example, in relation to their pregnancy, the concern with the effects of AEDs on their unborn babies created a conflict for women between being a ‘good mother’ and being a ‘good patient’. Thompson et al. ’s study30 provides a much needed contribution to understandings of how epilepsy influences women’s experiences of the various stages of pregnancy and reproduction. However, as it is an exploratory study with a small sample size of 15 women, findings remain limited in scope.

The 2013 literature review29 included studies exploring not only women’s experiences of pregnancy, but also their experiences of reproductive health while managing epilepsy. This expansion of the review resulted in 16 additional publications, which were limited in their generalisability because of the small sample sizes and/or poor quality of data. Since the publication of this review, one additional study has been published in this area: Qiang et al. ’s 201631 small qualitative study on the support networks of 12 pregnant women living with epilepsy. There is, therefore, a dearth of high-quality research on the experiences of pregnant women living with epilepsy.

Study aim

To investigate the perspectives and experiences of pregnant women living with, and managing, epilepsy.

Objectives

To gain insight into the way pregnant women with epilepsy rationalise and make sense of the management of AEDs in the context of their lives by addressing the following research questions:

• How do women experience living with epilepsy before becoming pregnant?

• What do women perceive as issues of concern for them and their baby in terms of epilepsy management during pregnancy and childbirth?

• How do women construct and make sense of the risks, and benefits, for themselves and for their baby in terms of medication?

• How do women perceive maternal responsibility in the context of having epilepsy?

• What and who influences women’s decision-making in the management of their condition during pregnancy?

• How do women view their experience of pregnancy and childbirth and the management of their medication during this time?

Sample

Theoretical sampling was employed to purposely include women from different geographical regions, with a diversity of sociocultural backgrounds, and who had varied histories with epilepsy and had experienced a range of neurological symptoms. A total of 32 women participated in interviews, of whom 21 had enrolled in the RCT and 11 had declined the trial but agreed to take part in the qualitative study. Recruitment and sampling continued until data saturation was reached and no further analytical categories emerged from ongoing analysis of interview data. 34 Saturation was determined independently by the EMPiRE trial qualitative lead, Elaine Denny, and research fellow, Annalise Weckesser. Women were first approached face to face by research nurses and midwives and given informed consent forms for the qualitative study. Annalise Weckesser then telephoned women who had agreed to take part in the qualitative study and who had signed informed consent forms.

Method

The aim of this research was to gain insight into the way women make sense of living with epilepsy during pregnancy and, thus, a qualitative approach was appropriate. All women were requested to take part in two or three interviews, which were audio-recorded with their permission and transcribed verbatim. Annalise Weckesser or Elaine Denny interviewed the women twice. Annalise Weckesser and Elaine Denny are both women with experience conducting qualitative research. The interviewers did not establish relationships with participants prior to the commencement of the study. Participants knew that Annalise Weckesser and Elaine Denny were non-clinical members of the EMPiRE trial research team, and that both have research interests in gender, reproductive health and chronic illness.

The first interviews took place when women were pregnant and had either entered the trial or refused to enter the trial. The second interviews took place approximately 6 weeks after participants had given birth. First interviews lasted approximately 1 hour and second interviews lasted approximately half an hour. Eight women did not participate in follow-up postnatal interviews: one returned to her country of origin, two withdrew from the RCT and five were unable to be contacted.

Interviews were conducted at places and times convenient to participants. The majority of women were interviewed in their own homes; however, some preferred to be interviewed at hospitals after their antenatal clinic appointments and some interviews took place over the telephone. Most women were interviewed on their own; however, some women asked for their partner (n = 4) or mother (n = 2) to be present to help them remember details of their seizures and medication. Most postnatal interviews were conducted over the telephone as this was most convenient for women with the time constraints of caring for their newborn. In appreciation for their time and participation, women were given a £20 gift voucher following completion of their first interview.

First interviews took place on the women’s entry to or refusal of the randomised trial, and these interviews focused on the six research questions (see Objectives). Additional interviews were originally proposed with women in the qualitative study who had experienced a reduction in serum AED level or who experienced a seizure during pregnancy to explore if these events altered patients’ perspectives on epilepsy and pregnancy and raised new concerns. However, this was not possible as research nurses and midwives did not inform Annalise Weckesser and Elaine Denny when a patient had a seizure during pregnancy. However, in postnatal interviews, Annalise Weckesser and Elaine Denny learned that some participants did have seizures during their pregnancy and we were able to capture these experiences retrospectively. Postnatal interviews concentrated on women’s reflections on the research questions on the presumption that pregnancy experiences can only be fully reflected on once the outcome of the pregnancy is known. Short field notes were taken immediately after the first and second interviews to make note of, and describe, where interviews took place.

Elaine Denny and Annalise Weckesser developed interview guides to ensure data collection on relevant topics (see Appendix 4), but participants were also free to raise issues of importance to them. The interview guide was developed based on themes identified in a review of qualitative literature on the experiences of pregnancy and reproductive health of women living with epilepsy, which was published by Elaine Denny and Annalise Weckesser in 2013. 29 Basic demographic data including current age, parity and years living with epilepsy were collected from participants at the beginning of the first interview.

Ethical considerations

As epilepsy is considered to be a stigmatising condition,35 the researchers avoided stereotyping and discriminatory use of language. Each woman’s guidance was sought at the beginning of interviews concerning acceptable use of terminology. The researchers complied with the British Sociological Association’s statement of ethical practice. 36 Pseudonyms have been used to protect the anonymity of participants.

Analysis

A narrative analysis was adopted as this method has much to contribute to studies of chronic illness. As Riessman notes, ‘[t]elling narratives is a major way that individuals make sense of disruptive events [such as illness] in their lives’. 37 Within this narrative mode of analysis, a thematic approach was undertaken – a method that allowed for the identification of common themes across cases while enabling individual women’s stories to remain intact. 38 To ensure rigour in the analysis process and to establish trustworthiness in the findings, Elaine Denny, Annalise Weckesser and a member of the EMPiRE trial team read all interview transcriptions. Annalise Weckesser took the lead in developing the analysis to increase internal consistency, but all members agreed on coding frames and analytical themes for internal validity. Annalise Weckesser created a coding frame for categorisation of data using NVivo 10 (QSR International, Warrington, UK). Analytical themes and concepts were developed and explored using the constant comparison method. 39 An additional strand of narrative analysis was also conducted, allowing for the integrity of each woman’s interview to be maintained. 38 This analysis of narratives allowed for an understanding of the inter-relatedness of a person’s life story that can be lost and fragmented in the constant comparison method. 38 These two methods of analysis provided insights into how women experience pregnancy and epilepsy, and how they make sense of these events within the context of their lives.

Sample results

Participants came from urban areas, including London, Birmingham, Cardiff and Liverpool, as well as more rural areas such as Shrewsbury, Gwent and Worcestershire. Table 16 provides the sociodemographic details of participating women. At the time women were first interviewed, over half were becoming mothers for the first time (n = 18) and the rest had at least one child. The youngest participant was aged 19 years and the oldest was aged 42 years (mean age 31 years). More than half of participants were married (n = 18) and others lived with partners (n = 10), lived separate from partners (n = 2) or were single (n = 2). Women worked in professional occupations (n = 17) and in retail (n = 2), or were unemployed and/or full-time mothers (n = 12); one was a full-time student. The majority of participants had been born in the UK and self-identified as white British (n = 21); others identified as British Asian (n = 4) and British black Caribbean (n = 1). A number of women had immigrated to the UK, including three participants who identified as white European and one each as Chinese, black African and white American. One NHS Mandarin interpreter was required to provide an interview translation.

Participants had varied histories with epilepsy. One mother was diagnosed only within the past year and, at the other end of the spectrum, a participant had lived with the condition for 29 years. On average, women had lived with their condition for 11 years. Women also experienced a wide range of neurological symptoms. Participants self-identified their seizure types, ranging from tonic–clonic seizures (which constitute the more popular images of epileptic convulsive seizures with a person losing consciousness, their muscles stiffening and jerking), myoclonic seizures (involving the brief, shock-like jerking of muscles) and absence seizures (which are absences in awareness and women often described these experiences like ‘déjà vu’ or an ‘aura’). Some women experienced more than one type of seizure at different stages in their life, and the frequency of seizures also differed between women and among individual women over time.

Interview findings

The following findings are based on ante- and postnatal interviews. For the purposes of this chapter, findings are presented thematically rather than as narrative case studies to facilitate the reporting of findings related to the qualitative study’s research objectives of understanding:

• women’s experiences of living with epilepsy before becoming pregnant

• women’s concerns in relation to epilepsy management in pregnancy

• women’s strategies for balancing risks and benefits to themselves and their babies in relation to medication

• women’s perceptions of maternal responsibility in the context of having epilepsy

• the influences on women’s decision-making in the management of epilepsy and pregnancy

• women’s postnatal reflections on the experience of pregnancy and childbirth, and the management of epilepsy

• women’s reasons for declining trial participation (for participants who declined participation in the randomised trial).

Although reporting findings through this separation of strands of experiences brings clarity, it must be noted that it creates a false distinction; in reality, people’s experiences, feelings and actions are interlinked and, thus, cannot be easily reduced to simple, segregated categories.

Experiences of living with epilepsy before becoming pregnant

Women’s histories of living with epilepsy before becoming pregnant are highly diverse. This diversity is reflected in the spectrum of seizure types and frequencies experienced by participants, and the number of years they have lived with the condition (see Table 16). Although some women had been diagnosed in childhood, others did not receive a diagnosis until more recently and/or after a first pregnancy.

With regard to how the management of their condition impacted on day-to-day life, women’s responses ranged and were shaped by this diversity of seizure types and frequencies. For some, the fear of having a seizure was a daily occurrence:

[I] feel quite nervous and self-conscious all the time because I don’t know when I’m going to have my next seizure.

Cecilia

For some, epilepsy impacted on their work and chosen career paths. For example, one participant reported losing her job in a factory after her diagnosis as she was not allowed to work near the machinery in case she should have a seizure. Another participant had been training as a beautician but was told she could not continue with the course after disclosing her diagnosis, as she would not be allowed to use some of the electrolysis machines.

The majority of participants, however, reported that on a day-to-day basis their condition did not impact on them greatly. Some made modifications to their lifestyle (ensuring that they get enough sleep, refraining from excessive alcohol consumption, not bathing alone, etc.) but saw these as minor adjustments. Riva is one such woman; despite these adjustments, she states that she leads a ‘normal life’:

[Prior to becoming pregnant] I could get up every day, I would have my medication, I could go to work and have a normal life and go back home. And you know, it wasn’t something that would impact me greatly . . . I had a gap of several years between my seizures. So for me the seizures, like it didn’t feel like it had a particularly detrimental impact on my life.

Other participants had such infrequent seizures, some having ever experienced only one seizure, that they reported not feeling they had ‘real epilepsy’ and that they were ‘lucky’ as they believed they faced less hardship and stigma than those with less controlled and more ‘severe’ seizures.

In addition to the diversity of epilepsy experiences, some participants faced additional pregnancy and/or concurrent health concerns that took primacy over their epilepsy. Some women reported fertility issues, challenges having a baby in their forties, undergoing in vitro fertilisation treatments and/or having past experiences of miscarriage. Becoming pregnant was reported, by some, to be more of a challenge and concern than managing their epilepsy:

I thought, ‘OK, I just need to manage my medication and then I’ll get pregnant’ . . . I think more than anything to do with my medication that was the biggest shock for me, that actually it isn’t that easy to get pregnant. Like it’s easy to manage what dosage you take and to keep tabs on what you’re taking, making sure you go and see the specialist and [your epilepsy is] managed. But I think the biggest shock to me was just the process of actually getting pregnant in the first place.

Sonia

Other participants reported additional health concerns that impacted on their day-to-day life, including Tourette syndrome, congenital talipes equinovarus, overactive thyroid and high blood pressure, as well as other related health issues that arose during pregnancy, such as pre-eclampsia and gestational diabetes. One participant felt that managing her Tourette’s affected her more than her epilepsy:

The epilepsy I don’t notice, because if I have a fit, it’s always at night time. So I’ve never had one in the day, I’ve always been fine. Because in the day, I kind of control the Tourette’s.

Tanya

Women’s experiences of epilepsy were also influenced by their different sociocultural and religious backgrounds. One participant believed that her epilepsy had been caused by a curse and attended evangelical faith healing sessions. Another participant, Amina (aged 31 years), after being diagnosed with the condition reported becoming a more ‘devout’ practising Muslim, signified by adopting a hijab. Amina believes that her faith helps her manage her condition; however, she continues to take her AEDs:

I’ve got my religion, but I’ve also got the doctors. This medication’s there for a reason. It’s helped me not have a seizure all this time so I’ll just continue.

Women’s experiences of living with epilepsy prior to becoming pregnant are highly diverse and are shaped by their particular seizure type(s) and history, whether or not they have additional health and fertility concerns and their sociocultural backgrounds.

Concerns in relation to epilepsy management and pregnancy

As discussed in the previous section, for many participants the everyday management of epilepsy prior to becoming pregnant had become routine and normalised. However, pregnancy often becomes a stage at which women have to reflect on their condition and how it impacts on their health and that of their baby. This is illustrated by the case of Philomena (aged 31 years), who states that before becoming pregnant she ‘wasn’t thinking about [her epilepsy] from one day to the next . . .’. After first being diagnosed, she initially took on ‘all the good habits’. She continues:

I never used to drink . . . [I] only took a bath when people were there. So you kind of start with really good practices . . . [A]nd then as the years go by you just stop thinking about it completely and you just take your tablets every night and you’re fine . . . But then as soon as you became pregnant you need to get back into good habits . . . [W]hen you have the baby you need to change [them] on the floor, you shouldn’t do this, you shouldn’t do that, and you were just a bit like, ‘Oh yeah, I completely forgot!’.

Thus, pregnancy is a time that raises many concerns for women living with epilepsy, concerns about things that some may have previously taken for granted.

Participants reported that their primary concern was to give birth to a healthy baby, with no abnormalities:

I think really you worry about everything, you could worry about anything, but I think the main thing is I just want to have a happy and healthy baby at the end of it.

Simone

At the moment I get concerned about whether [the baby’s] going to be normal or not. Otherwise, I haven’t got any concerns.

Mary

The first question I’m asked is, ‘Do you want a girl or a boy?’. And I just say, ‘I want a normal baby’. A healthy, normal baby . . . I don’t care whether it’s a girl or a boy . . . I hope it’s a healthy, fat baby.

Samina

Women also expressed concerns about the effects of their AEDs on their unborn babies. Although participants often reported feeling some reassurance from health practitioners who advised them that the medication they were on were newer AEDs believed to be safer during pregnancy, some still had concerns about possible teratogenic effects, including spina bifida and learning disabilities. This is illustrated in the following extract:

The doctors are like ‘Oh this is a great drug, it’s much better than the one you were on before’. And I’m thinking, ‘Yes, but the one I was on before it’s been around 20, 30 years’. So you know that there’s defects, but you know at what levels those defects occur and how likely it is to happen. Whereas you put me on this new medication, which actually has been around maybe 4 or 5 years so you have a little bit of experience . . . Even though [the doctors’] experience and knowledge is quite valid, but at the same time they don’t have to live with the consequences of it.

Riva

Some women were concerned that they could pass their condition on to their babies, despite the knowledge that genetic inheritance of epilepsy is very rare:

The other thing that you worry about is sometimes it’s like I hope [epilepsy’s] not something I can pass on to my baby because I really wouldn’t want that . . . Even though they say it’s not inherited you just don’t know because they say some forms can be.

Fatimah

Women also expressed concerns about having a seizure during their pregnancy, labour and/or in the early postnatal period. During pregnancy, women were especially concerned about having a seizure in their final trimester, when the baby is seen to be more ‘fully formed’. As Li Min stated:

[I’m] worried that [I’m] going to have a seizure towards the end, when the baby’s like mature . . . [I’m] really scared that it’s going to happen and it’s going to affect the baby and me as well.

Participants worried that a seizure could cause a decrease in the amount of oxygen that gets to the fetus. Tiredness and stress during pregnancy and from the strains of giving birth were also seen as possible seizure triggers. For example, one participant, Laura, feared that she was more likely to have a seizure in her current pregnancy than in her former, as she was now experiencing more sleeplessness as a result of taking care of her firstborn, who was still a baby:

I think I’m probably more concerned this time round because I have it in the back of my mind that I’ll probably have more seizures during this pregnancy because of having [my son] and the sort of stress of looking after him really, as well as the tiredness . . . [T]he end of the day comes and you think ‘I don’t think I’ve even brushed my teeth today to start off with’ . . . So I’m more concerned that I’m going to be more tired, which will lead to me having more seizures, which will lead to me being more tired.

Finally, women were very concerned about having a seizure because this could result in losing driving privileges until they have been seizure free for 1 year:

The main hope is not to have any déjà vu. Because even the slightest déjà vu I need to inform [my medical team] . . . So that is a worry, but mainly for the selfish reasons of driving.

Laura

[If I have a seizure] then I wouldn’t be able to drive, which I think would be a massive issue . . . [T]hat would make everything like a million per cent more difficult.

Philomena

Women in rural areas with little public transport and those who required a vehicle for work were especially concerned about the isolation and inconveniences caused by losing their driver’s licence in the event of a seizure.

Strategies for balancing risks and benefits to themselves and their babies in relation to medication

Many of the participants reported thinking about whether or not they should have children in view of having epilepsy. Most women had planned their pregnancies (n = 22) and two women had been advised against having children because of their condition (one by a doctor and one by their partner). Philomena reflects on her decision to have a baby, stating:

You think about what effects the drugs have on the baby, what if I had a fit . . . There are all of those considerations, but for me none of them outweighs actually having a baby.

In relation to taking AEDs, women made their medication management decisions by weighing up the risks to the health of their baby, their own health and other aspects of their life. Some women felt it was particularly important to keep taking their medication during their pregnancy to minimise the risk of seizures:

I can’t help but think I’m so scared that I’ll have a seizure and somehow or another I will end up hurting either the baby or myself . . . And I think your priorities completely change, you do want to do what is best for your baby, but at the same time if you’re harming yourself you can’t be doing what’s right for your baby. If I fall and if I have a seizure, then I’m risking both anyway. I think you’ve just got to look at it practically.

Fatimah

The risk of me being on the medication is minimal [to the baby] . . . But the risk of me falling down a flight of stairs could kill her really.

Shelly

Thus, although participants saw continuing their AEDs during pregnancy as potentially harmful to the health of their babies, such risks had to be balanced with what some saw as a more likely and harmful risk of having a seizure that could occur if they stopped or reduced their medication. Some women did choose to stop or reduce their medication in the first trimester because of the perception that the fetus was at heightened risk of malformations at this time:

When I found out I was pregnant with [my first child], I did stop taking [the medication] ‘til after my 3-month scan, and I did do this time as well for the current pregnancy . . . I went to the [clinic] at about 10 weeks’ pregnant and [the epilepsy nurse] said, ‘Have you had any seizures?’. I said, ‘No’. She said, ‘You know you’re more susceptible, etc., etc.?’. And I said, ‘Yes’. And she said, ‘Well I’m going to leave you to it’. And then 10 days later I had a bout of seizures, so that just brought me up to my 12-week scan, so I’m back on my medication now, properly.

Laura

Women’s decisions about taking medication in pregnancy to prevent seizures had to be balanced with other aspects of their lives, such as the need to keep their driving licence. One participant, Sandra, had a seizure during her first pregnancy, and during her second pregnancy she stated:

[The seizure] cost me my driving licence. I was a community midwife [in a rural area] . . . It was the most depressing time without a driving licence. I didn’t really know how much it meant to me, and it put me off getting pregnant again because the thought of losing my driving licence again, it was too much of a risk.

Sandra

Sandra believed she had the seizure because her epilepsy specialist midwife had failed to increase her AED dosage during the first pregnancy. She reported that for this current pregnancy her epilepsy specialist midwife had agreed to increase her medication, but if the midwife had not, Sandra would have self-managed her medication and increased her dosage to prevent a seizure.

Perceptions of maternal responsibility in the context of having epilepsy

Although women’s decisions ranged from stopping, reducing, maintaining and increasing their AEDs over the course of their pregnancies, overwhelmingly the rationale for such decisions was based on a feeling of maternal responsibility towards their babies. This is illustrated in the following two contrasting extracts, one from a woman discussing her decision not to take her AEDs in the first trimester, and one woman on why she continues to take her medication during her pregnancy:

If I took the tablets all the time and something happened and [the baby] comes out deformed in some way or had something wrong with it, then I would think that I’ve been a selfish person and feel terrible for doing that when I know I can cope for so long without them.

Mary, on not taking AEDs

[T]here is the odd night I’ll get into bed and every night [think] ’have you taken your tablets? Yes’. But I will remember if I lay there long enough. I’m like, ‘Oh, I didn’t take them’. But it’s not just me I’m thinking about now. So yes, it is a little bit more important. Because I’m not just taking them for me.

Veronica, on taking AEDs

Both women made decisions regarding their medication based on feelings of maternal responsibilities towards their unborn children. The types of seizures that the women experienced also informed their choices. Prior to having full tonic–clonic seizures, Mary would often experience symptoms including auras. Thus, if she had any of these early warning signs she would take her medication. In contrast, Veronica did not have such warning symptoms prior to having a seizure.

Women continuing or increasing their medication to prevent seizures in pregnancy still had concerns about the chance AEDs could affect their baby’s health. These concerns derived from their role as mothers and carers, now looking after the health of their unborn child as well their own. Clara stated, regarding the possible teratogenic effects, that:

[C]hances are very, very low, but then there’s always still that chance. And it’s not just you that you’re talking about any more, it’s an extra person, which took me by surprise at how differently I probably feel about that because, you’re like, it’s not just me any more, it’s another little person.

Participants’ feelings of maternal responsibility were also evident in the form of guilt associated with the possibility of having a baby born with health problems, as illustrated in the following extract:

I worried a lot that if there were problems with the baby that it would be my fault. It wouldn’t necessarily be my fault, but it kind of is if you know what I mean. I worried that the drugs that I took . . . I did have quite sleepless nights thinking, ‘Oh God, what if something happens to the baby and they’re born with defects that I’m going to have to explain to it’. That worries me.

Nicole

Influences on decision-making in the management of epilepsy and pregnancy

Women had a range of influences on their decision-making in the management of epilepsy and pregnancy. In relation to their medication, many made decisions regarding their dosage in consultation with their neurologist, epilepsy nurse or epilepsy midwife, but were also informed by knowledge of their own body, history with the condition and seizure warning symptoms. Women’s decisions, as discussed in Perceptions of maternal responsibility in the context of having epilepsy, were also shaped by considerations for the well-being of their baby. Many participants had partners and key family members (such as mothers) who provided them with care and support in the management of their condition. However, partners and family members did not play key roles in influencing women’s AED management choices. As Fiona replied, in relation to her medication, ‘At the end of the day I make the decisions for what I want and what’s best for me and the child’.

Some women reported using the internet to research the teratogenic effects of the particular AEDs they took, but viewed such information as supplementary to the professional advice of their medical team.

Postnatal reflections on the experience of pregnancy and childbirth, and the management of epilepsy

Women’s perspectives on pregnancy and managing epilepsy were shaped by labour and pregnancy outcomes. The majority of women reported giving birth to babies who, from birth to the point of the postnatal interview, did not have any apparent health problems linked to taking AEDs. For such women, their views on having a pregnancy while managing epilepsy were largely positive. This is reflected by Jeannette, who stated, 6 weeks after giving birth to a healthy baby:

I just don’t think if a woman has epilepsy [that she] should be scared to have a baby because [she] could be like me, everything is fine, no worries at all. So I wouldn’t even worry about thinking, ‘Oh I can’t have a baby because I’ve got epilepsy’. Because it’s twaddle really.

One participant reported early health problems with her baby that she was concerned could be linked to AEDs. Amy had a son ‘born with shaky arms and legs’ and some possible visual impairment. She stated that ‘[At] first, I thought, “Oh was that to do with the medication?” Could that have had an influence because I take these tablets?’. She consulted paediatricians who attributed these complications to an ‘immature neurological system’ and assured her that her son would ‘grow out’ of the ‘shakes’. Amy stated that she now felt ‘pretty reassured’ as the doctors did not believe the medication caused her son’s health impairments and because she is prescribed ‘one of the safest drugs and take[s] quite a relatively low dose’.

Although seizure during labour had been a concern for some women while they were pregnant, only one participant experienced a seizure during childbirth. Many women reported negative experiences in postnatal wards as a result of a lack of sleep and staff shortages. Some women reported feeling ‘abandoned’ and ‘vulnerable’ to seizure in the postnatal ward. They expressed a desire for a partner or family member to remain on the ward to help care for them and their newborn infant. Although a few wards allowed partners/family members to remain outside of visiting hours, this was not universally practised across all wards. One woman reported that she felt that she would not have another baby because of her poor postnatal care experience.

In relation to caring for their newborns, many women reported having to make accommodations as a result of their epilepsy. This was evident in sleeping patterns, breastfeeding strategies and day-to-day care practices for infants. For some participants, sleep deprivation was a seizure trigger. Thus, to ensure they had enough sleep, some had partners, family members or night nannies take the lead in baby care and feeding overnight. Some of these women expressed feelings of guilt that they were not ‘proper’ mothers because they were not doing night feeds.

Many women found mixed feeding (combining breastfeeding and bottle feeding) an effective way to ensure that their babies got the health benefits of breast milk while also being able to ‘top up’ with formula milk. Women felt that this feeding strategy helped babies sleep for longer periods in the night, allowing women to get more sleep. Bottle feeding also allowed partners and family members to share in the feeding duties. A few women reported receiving conflicting advice from health professionals regarding the safety of breastfeeding while taking AEDs.

Precautionary practices in the day-to-day care of newborns, such as breastfeeding or nappy changing while sitting on the floor, refraining from bathing a baby alone and using a car seat when carrying a baby on stairs, were less likely to be taken up by women with a history of well-controlled seizures.

Reasons for declining participation in the randomised trial

Eleven women declined to participate in the RCT. About half (n = 6) of this group chose not to take part in the RCT because the randomisation process was not acceptable to them. They were concerned that they may be streamed into the clinical features monitoring strategy of the trial, and would have their medication dosage increased only after a seizure and not when their blood levels decreased, with negative consequences for them and/or their baby. Such women reported that they felt they would lose control over the management of their epilepsy:

I thought it would be easier to control my seizures if I didn’t go into the study.

Clara

For me it would be quite a bizarre choice not to know what was happening to the levels of lamotrigine in my blood, and not to intervene.

Sandra

These six women also expressed concern that they would be at more risk of a seizure if they were randomised:

Should I require additional medication, that wouldn’t necessarily be prescribed if I was part of the wrong part of the trial.

Nicole

If I could choose which group I was going into that would be fantastic, but I was told I couldn’t choose which group I was going into. So it wasn’t worth the risk [of having a seizure].

Sandra

The five other non-randomised women chose to decline trial participation for the following reasons:

1. work and time commitments (n = 1)

2. lived too far from hospital for required monthly antenatal visits (n = 1)

3. fear of needles (n = 1)

4. stopped taking AEDs prior to pregnancy and did not want to resume (n = 1)

5. believed she did not have epilepsy (n = 1).

Strengths and limitations of the qualitative study

As this study was carried out alongside the EMPiRE trial, including only women who chose to have children, there is a risk of bias as those having children may have more well-managed seizures and fewer negative symptoms and side effects associated with their medication. Approximately one-third of women of childbearing age living with epilepsy in the UK consider not having children, or having fewer children, because of their condition. 40

Additional limitations of this qualitative study include the use of a self-selected sample and an inability to capture participants’ experiences of seizure during pregnancy. These seizure experiences were captured only retrospectively through postnatal interviews.

Despite these limitations, the study’s strength lies in the original contribution it makes to further understanding women’s experiences of epilepsy, pregnancy and reproductive health, in which there has previously been a dearth of robust, in-depth qualitative research. 29 To our knowledge, this constitutes one of only two studies to directly examine women’s experiences of pregnancy while managing epilepsy. 30

Women’s views for declining participation in the randomised trial

Findings from the qualitative study regarding women’s rationale for declining the trial may also help understand reasons for slow recruitment. Eleven women who participated in the qualitative study declined the RCT; of these, approximately half reported that they found the process of randomisation unacceptable. They expressed concern that randomisation would lead to a ‘loss of control’ over the management of their epilepsy as they could potentially be streamed into the clinical features monitoring strategy of the trial, and would have their medication dosage increased only after a seizure and not when their blood levels fell. They believed that not increasing their medication when their blood levels dropped could potentially lead to a seizure and they or their baby could be harmed. Thus, women’s concerns regarding preventing seizures in pregnancy and maintaining control over their medication regime could also underpin slow recruitment to the trial.

Reductions in serum antiepileptic drug level in pregnancy and seizure deterioration

Serum AED concentrations often decrease during pregnancy. Physiological changes in pregnancy alter AED pharmacokinetics and AED concentrations. There is reduced gastric tone and motility, and increased plasma volume, renal clearance, albumin levels and protein binding. 9,14,42,43 The reductions in serum AED levels are considered to aggravate seizures. 44 Monitoring of serum AED levels in each trimester and after delivery has been recommended by the American Academy of Neurology based on consensus as a good practice. 44 In the UK, however, the Scottish Intercollegiate Guidelines Network’s guideline does not recommend regular AED monitoring in pregnancy because of a paucity of evidence. 12 Our systematic review on the effect of AED monitoring strategies in pregnant women with epilepsy on AED showed lower rates of seizures with therapeutic drug monitoring than clinical features monitoring strategies. 45 The studies were not randomised, or controlled, results were heterogeneous and there was imprecision, with small numbers of women making findings unreliable.

In our trial, we did not observe any differences in seizure rates and time to first and to multiple seizures with the two strategies. Although the point estimates of hazard to time to first seizure and first tonic–clonic seizure, and to multiple tonic–clonic seizures, showed a trend towards favouring therapeutic drug monitoring, the findings were not significant.

Antiepileptic drug exposure in pregnancy to mother and fetus

The measurement of total AED exposure enabled us to delineate the likelihood of excess exposure under therapeutic drug monitoring, in which dose escalation is expected in response to a known decrease in serum AED levels. However, no differences were observed between the groups. It is likely that, when serum AED levels fell, the intervention in the therapeutic drug monitoring group comprised either close monitoring or dose escalation, whereas, without information on serum AED level, clinicians escalated drug doses in response to their clinical monitoring. Despite similar average AED dose exposure in both groups, the cord blood levels of the commonly prescribed AEDs lamotrigine and levetiracetam were higher in newborns in the therapeutic drug monitoring group. The developmental quotient of infants of mothers exposed to lamotrigine in pregnancy compared with women without epilepsy and women not on AEDs appeared to be similar in a small study. 46 There is limited evidence to assess the effect of levetiracetam or AED polytherapy on long-term neurodevelopment. 47

Quality of life in women using antiepileptic drugs

The effects of seizures extend into daily living, resulting in a loss of driving licence, and are known to have a negative impact on employment and relationships, and reduced QoL. 48 We found no differences in the scores for QoL between the two groups. The additional information on serum AED levels in pregnancy, and the subsequent management based on it, did not appear to adversely affect women’s QoL.

Recommendations for clinical practice

Women with epilepsy on AEDs require management in a multidisciplinary setting, with a team involved in both care of their pregnancy and their seizures. The standards of care for individuals with epilepsy vary widely across the UK. 49 This is particularly relevant for pregnant women with epilepsy. Our survey of epilepsy specialists (n = 29) in the UK showed that one-third of participants managed women with epilepsy on AED with regular therapeutic monitoring of drug levels, one-third adjusted doses based on clinical features only and the rest used therapeutic drug monitoring occasionally. 50 Given the wide CIs, reflecting the imprecision, the absence of differences between the two strategies and similar rates of seizures in women with stable and reduced serum AED levels, we are not able to advocate routine therapeutic drug monitoring monitoring in pregnancy.

Although we randomised only half the number of women required to provide the definitive answer, given the wide imprecision in the CIs, we do not expect one strategy to be shown to be significantly effective compared with the other, even if we had managed to recruit to target. We calculated the fragility index, which is the number of currently randomised women who should have been seizure free, to show a 25% decrease in seizures, as postulated in our sample size calculation. Only 36 women should have suffered seizures, compared with the observed number of 48 women, a significant number that should have been reduced. We also calculated the additional number of women needed to show statistical significance for the effect size as observed. Assuming that the observed effect size and SD holds, we estimate that the number of women that would need to be recruited to demonstrate a significant effect of therapeutic drug monitoring and clinical features monitoring on time to first seizure and time to any seizure was 1038 and 1302, respectively.

The risk of seizure deterioration was not significantly different in the non-randomised group with stable serum AED levels and those with a decrease of ≥ 25%, reinforcing the lack of benefit of routine drug monitoring. Furthermore, we observed a significant increase in the cord blood levels of lamotrigine and levetiracetam in women whose AED doses were managed based on therapeutic drug monitoring. Given the above findings, in the absence of firm evidence on long-term neurodevelopmental outcomes in infants exposed to AEDs, particularly the newer AEDs such as levetiracetam, caution is required prior to routine dose escalation based on serum AED levels alone.

Our qualitative study findings have led to the following recommendations on care for pregnant women living with epilepsy.

• Pre-conception information: pre-conception counselling interventions need to continue to work towards better identifying, and reaching out to, women who are not accessing this information, and to provide consistent information on whether or not they should start a family.

• Antenatal care and medication management: women’s varied positions on the spectrum of seizure types need be more fully recognised by health professionals, as this informs how women understand and manage their condition. It should also be recognised that women’s decisions to stop, reduce, maintain or increase their AEDs over the course of their pregnancies are based on a rationale of maternal responsibility towards their babies.

• Postnatal care: ward practices need to facilitate appropriate access for partners and families of women with epilepsy in order to manage their condition following delivery.

• Supporting mothers with epilepsy: women would benefit from more advice and information concerning modifying their caregiving practices (changing nappies on the floor, never bathing babies alone, etc.), as these practices were often not considered fully until after the babies were born.

Recommendations for future research

Given the difficulties in achieving the target sample size, despite recruitment in more than 50 centres, conducting future randomised trials with a large sample size will be challenging. Any such trials will need to take into account the core outcomes needed for minimal reporting to enable meaningful evidence synthesis. The risks of seizure deterioration for various threshold levels of decrease in AEDs need further evaluation. A robust risk assessment method, using an individualised prediction model by taking into account mothers’ clinical characteristics including type and duration of seizure, type of AED and change in serum AED levels, will help to identify those women who need close monitoring in pregnancy. Importantly, the long-term neurodevelopment of the infants born to mothers in both randomised groups, and any impact on health-care costs, need further evaluation.

In relation to the qualitative research on pregnant women’s experiences of managing epilepsy, additional research is needed with women with less controlled and/or more frequent seizures. As the qualitative study was carried out alongside the EMPiRE trial, which included only women who had chosen to have children, there is a risk of bias as those women having children may have better-managed seizures and fewer negative symptoms and side effects associated with their medication. Furthermore, there is a need for a more integrated approach in future research in this area to provide a more comprehensive picture of the clinical and experiential aspects of taking AEDs.