Womens health in focus: Real-world data on valproate prescriptions during pregnancy – a cohort study in Catalonia (Spain)

Introduction

Valproate and its related compounds are approved in Spain for the treatment of epilepsy, encompassing primary generalised epilepsy, partial epilepsy, secondary generalised seizures and West and Lennox-Gastaut syndrome. Additionally, it holds approval for the treatment of acute mania in select cases among individuals with bipolar disorder.1–3 The teratogenic risk associated with the use of valproate in pregnant women is well established.4–7 Therefore, its use in women of childbearing age is restricted to prevent valproate exposure during conception and pregnancy. The best-known malformations are neural tube defects, but valproate has been shown to cause neurodevelopmental changes and other congenital malformations.8 9

There have been several warnings regarding the use of valproate in women of childbearing age: In October 2014, the European Medicines Agency (EMA) Pharmacovigilance Risk Assessment Committee (PRAC) issued recommendations for valproate prescribing. These recommendations emphasised that valproate and related substances should not be used in girls, women of childbearing potential and pregnant women unless alternative treatments are ineffective or not tolerated. Specifically, they should be contraindicated in prophylaxis of migraine attacks in pregnancy and women of childbearing potential who are not using effective methods of contraception.10 In March 2018, the PRAC recommended update risk mitigation measures (RMMs), introducing a Pregnancy Prevention Programme (PPP) and advocating for more stringent measures.11

There are several studies which have explored risk awareness, prescription patterns12–14 and adherence to the RMMs in different European countries. As an example, the multidatabase longitudinal study carried out by Abtahi et al,15 spanning from January 2010 to December 2020, assessed the impact of prevention programmes for valproate-containing drugs. Despite a reduction in valproate prevalence, there was no statistically significant reduction in incident use in women of childbearing age. Poor adherence to contraception during valproate treatment and high rates of concomitant pregnancy were observed. Implemented measures had limited impact, as indicated by data from five European databases. In the same field of study, Jödicke et al
16 reported on valproate use in women aged 12–55 in the EU (European Union), analysing data from six electronic healthcare databases. Both studies suggest a decline in valproate use among women of childbearing age in most European databases, with an increase in alternative treatments, such as lamotrigine and levetiracetam among other drugs.

The findings mentioned above, along with other studies,17–19 highlight the significance of regulatory measures and electronic healthcare databases in assessing prescription patterns during pregnancy.20 In this context, Lestón et al developed an algorithm within the System for the Development of Research in Primary Care (SIDIAP) database to identify pregnancy episodes from January 2011 to June 2020, which seems to hold the potential as an efficient database for investigating drug safety during pregnancy and its implications for the offspring.21 In this context, assessing valproate prescription patterns during pregnancy has considerable value, due to the lack of drug-safety trials in this population, the well-documented impact on the offspring and the institutional efforts to prevent prescriptions in pregnant women.

Discussion

This study describes exposure to valproate during pregnancy, as well as the characteristics of the exposed pregnancies. We assessed the prescription of valproate in a cohort of pregnant women in Catalonia, dividing pregnancies by trimesters (pregnancy intervals). Of 99 605 identified pregnancies (in 76 459 women), at least 302 pregnancy episodes were exposed to valproate (276 women exposed in total). We also evaluated the impact of the RMMs recommended by the EMA.10

Our study shows a decrease in the prevalence and cumulative incidence of prescriptions of valproate as pregnancy progresses (prescriptions decline along pregnancy intervals). It is noteworthy that there is a subsequent increase in prescriptions after the ending of pregnancy, even though the drug is excreted in breast milk.3

Cumulative incidence by trimester has not been analysed in any other study on record. The notable rise in incident cases (eight cases) during pregnancies entering the third trimester is striking. This increase could be associated with the reintroduction of the drug in patients who have not responded well to alternative medications or even due to the escalation of epileptic seizures in the third trimester.31 Furthermore, there might be a misconception that the risk of malformations is primarily linked to exposure in the first trimester, overlooking the potential risk associated with developmental disorders.

Overall prevalence, prevalence by trimesters and prescription patterns, align with findings from research carried out in other countries,14 20 especially those corresponding to Italian databases. For instance, during a 10-year analysis of pregnancies with valproate prescriptions in Italian regions such as Tuscany and Emilia-Romagna, there were 172 cases in Emilia-Romagna and 490 cases in Tuscany. In comparison, 353 pregnancies in the UK were exposed to valproate.14 The similarities and respective differences towards other countries could be related to both the sociodemographic characteristics of the countries but also may be due to the peculiarities of the pregnancy detection algorithms in different databases. Each region in the aforementioned study14 used a different EHR system and a distinct algorithm to identify pregnancy episodes, which could be a limitation.

Regarding usage patterns, when assessing the prevalence of all ASMs globally, there is a clear decrease in use throughout pregnancy by trimester across all the countries assessed. For instance, the use decreases from 5.9‰ in the first trimester of pregnancy to approximately 2‰ in the third trimester, being valproate a 28.6% of those prescriptions.14

However, as opposed to other studies, the overall prevalence of valproate prescriptions differs from the results of studies in other settings such as Denmark.32 Our prevalence results indicate a significantly higher prevalence compared with studies that focus on claims rather than prescriptions.33

Our research also reveals that the measures imposed by the PRAC of the EMA in 2014 had a significant impact.10 These measures stipulated that valproate and related substances (valproic acid, valproate and valpromide) should not be used in girls, women of childbearing potential and pregnant women unless alternative treatments are ineffective or not tolerated. Additionally, valproate and related substances should be contraindicated for the prophylaxis of migraine attacks in pregnancy and in women of childbearing potential who are not using effective contraception. These guidelines were updated in 201811 to include a PPP. However, the impact of these latter measures could not be assessed in our study due to the lack of sufficient observational time.

Nevertheless, our study’s data demonstrate the effects of these regulations. There is evidence indicating a shift in the prevalence of valproate prescriptions during pregnancy after 2014, mirroring findings in similar studies that focus on both pregnancy episodes and women of gestational age.14–16 While some investigations15 concentrated on measures instituted in 2018 (with a different methodology), our study, constrained by limited information and time points, chose to evaluate those implemented in 2014. Consistent with trends observed in preceding studies, the use of alternative antiepileptic medications, such as levetiracetam and lamotrigine, witnessed a concurrent rise.20 32

Because the current database lacks a mother–child link, it was not feasible to analyse the impact of valproate on offspring during pregnancy, as done in prior studies.34 35 Efforts to conduct this analysis will be made in future research. Despite this limitation, we managed to analyse pregnancy outcomes, revealing a predominant occurrence of vaginal deliveries, followed by abortions (spontaneous and induced). This aspect of the research distinguishes it from other cohort database studies.14

Concerning health issues (table 1), we should highlight that the most frequent health issue associated with a valproate prescription is epilepsy, followed by anxiety. It is noteworthy, in turn, that 37.08% of pregnancy episodes presented, health issue associated with the year prior to the pregnancy onset, ‘nicontine dependence’. This diagnosis would not only affect the health of mothers during pregnancy but could also have implications for the newborn, which should be closely evaluated.

In comparison to similar research conducted in other countries, the health issues identified closely resemble those observed in other European nations such as France and the UK,14 with a notable prevalence of epilepsy and anxiety disorders. Conversely, in the USA, pregnant women exposed to ASM exhibited a higher prevalence of psychiatric disorders, followed by epilepsy and pain diagnoses.36 It is also noteworthy that migraine does not appear among the associated comorbidities when compared with other studies,16 but this could be because it is not an indication of valproate in Spain. Nevertheless, there is a considerable quantity of missing information on the health issues, and thus, these results must be evaluated with caution.

The aforementioned limitation is common among studies based on data extracted from EHRs, which were originally designed for clinical purposes. As an additional example, within our dataset, certain instances of Voluntary Interruptions of Pregnancy may be erroneously categorised as spontaneous abortions. Abortion in EHRs is not consistently recorded and different models for its register protecting women’s privacy may be difficult. Also, the correct classification of abortion in spontaneous, elective or induced, and the outcome registered in SIDIAP did not specify the abortion type, so cases could be spontaneous abortions or induced/elective ones.37 Consequently, we encounter challenges in accurately distinguishing between these categories, leading us to tally them collectively.

Moreover, the study’s reliance on an observational design hinders the establishment of causal relationships, emphasising correlations over direct causation. While the study focuses on prescription data for ASM, potential drawbacks exist, such as overlooking dosage and posology. Additionally, the exclusion of billing data introduces the possibility of information gaps. It is important to note that the present study lacks information on pregnancies that are followed up in hospitals (referred from primary care due to a high risk of complications) or in private settings.

At the time of conducting the present study, data on maternal breast feeding were not available, preventing us from determining whether it occurred in pregnancy-exposed episodes resulting in a live-born child. The potential impact of parental valproic acid intake on newborns is currently under evaluation in other studies,38 and various regulatory agencies have issued recommendations to healthcare professionals on this matter.39 40 Both areas represent intriguing fields for further exploration and investigation.

On the other hand, our study is fortified by a validated methodology, leveraging a validated database22 23 and the creation of an algorithm21 to ensure the reliability of the data.

It is noteworthy to mention that, globally, there are relatively few studies specifically focused on medication patterns in cohorts of pregnant women. The current study focuses on a cohort of pregnant women in Catalonia and their respective prescriptions for antiepileptic medications. No studies of similar characteristics had been conducted previously in this specific population, and to our knowledge, none have been carried out in Spain. Therefore, this study sheds light on the use of valproate in Spain in pregnant women and allows us to compare our findings with other countries.

Furthermore, considering the impact of the decrease in valproate prescriptions, we can confirm the effectiveness of the strategies employed by local authorities to adhere to RMMs, such as warnings in prescription programmes for medical doctors.41 Similar strategies could be implemented in other countries or regions if necessary. While these results may seem promising, the RMMs aimed to prevent any pregnant woman or those of childbearing age from being prescribed valproate. However, the challenging management of the condition, unplanned pregnancies and the lack of information regarding the associated risks with such medication hinder the achievement of this goal.

The findings of the present study align with the results of previous studies. An attempt has been made to profile in-depth the characteristics of pregnancies exposed to valproate, but, however, the use of different algorithms for pregnancy detection in each of the databases of other studies,14 as well as the methodology employed, hinder the comparison and generalisation of the results obtained, to the detriment of available scientific evidence. We should aim for greater consensus in the analysis methodology to maximise the pharmacological safety of pregnant patients, who already have a limitation of evidence of drug use for ethical reasons.

Emphasising the critical nature of decision-making in prescribing medications during pregnancy, our study underscores the need for careful consideration and informed choices. The impact of such decisions influences maternal and fetal health outcomes. It highlights the pivotal role healthcare professionals play and emphasises the importance of a thoughtful and evidence-based approach to prescribing.