Abstract
Background:
Awareness of the risks of prescribing valproate to girls and women of childbearing potential has increased, with some countries strongly restricting prescribing and requiring pregnancy prevention plans.
Objectives:
To explore changes in valproate dispensing patterns and the proportion of women covered by prescription contraceptives, especially long-acting forms, in Australia.
Design:
Population-based cohort study using routinely collected government data on dispensed medications.
Methods:
We analysed a 10% random sample of Australian Pharmaceutical Benefits Scheme (PBS) dispensing data for women aged 15–49 years. We calculated the annual rate of initiating valproate and alternative anti-seizure medications. We assessed what proportion of women were covered by long-acting or other PBS prescription contraception at treatment commencement.
Results:
Between 2013 and 2021, overall valproate initiation declined substantially (from 2.06 to 0.75 per 1000 women per annum), compensated by a rise in the initiation of alternative anti-seizure medications. This reflected a decline in valproate initiated as a first-line treatment, with valproate initiated as a second-line treatment remaining stable. For women initiating valproate, contraception coverage by an intrauterine device or implant increased from 16% in 2013 to 22% in 2021. In 2021, no coverage by PBS prescription contraceptives occurred in 69% of women initiating valproate and 68% of women initiating alternative anti-seizure medications.
Conclusions:
A decline in valproate prescribing has occurred in Australia. At the time treatment commenced, the great majority of women dispensed valproate were not covered by any form of PBS prescription contraceptive, which is of concern.
Plain language summary
In the United Kingdom, restrictions on prescribing valproate to girls and women were strengthened in 2018 along with pregnancy prevention requirements, and further restrictions commenced in 2024. Australia has not taken such steps and recent changes in prescribing practices as well as contraception coverage are unknown.
The researchers used routinely recorded drug dispensing data collected by the Australian government to study the frequency with which women commenced valproate or alternative anti-seizure medications, as well as coverage by prescription contraception.
Between 2013 and 2021, valproate initiation in Australian women declined substantially. In 2021, only 22% of those initiating valproate were covered by highly effective contraceptives and 69% were not covered by any PBS prescription contraceptives.
Clinicians appear to be aware of the need to avoid valproate where there are alternatives. Actions to strengthen pregnancy prevention are required.
Introduction
Valproate is approved in Australia as a medication to treat epilepsy and also bipolar disorder. 1 Valproate has been used for epilepsy since the 1960s as it is highly effective for controlling seizure. 1 Although alternative anti-seizure medications have been developed, valproate is in some cases the only effective medication, notably for patients with genetic generalised epilepsy.2,3 In psychiatry, valproate is used in acute treatment of manic episodes. As a mood stabiliser for long-term use in people with bipolar disorder, valproate is not more effective than alternatives such as lithium or some other anti-seizure medications that are used for this purpose. 4 Prescription of valproate ‘off label’ for other conditions such as pain and migraine occurs in Australia, following a pattern elsewhere.3,5–7
Valproate is known to be teratogenic, with several registry-based studies showing that the prevalence of major congenital malformations, such as heart defects, among children prenatally exposed to valproate ranges from 6.7% to 10.3%8–11 compared with 2% to 3% in the general population. 10 In addition, prenatal exposure to valproate is linked with a higher risk of neurodevelopmental disorders, intellectual disability and autism spectrum among children.12,13 Alternative anti-seizure medications are generally safer, but many also carry increased teratogenic risks, and for some the risk is not yet known.1,14
A decade ago the United States Food and Drug Administration (FDA, 2011/2013) 15 issued warnings about valproate, followed by European authorities (European Medicines Agency, 2014/2018).16,17 A lawsuit 18 led to a stronger response in the United Kingdom and Europe in 2018, 17 with valproate only to be prescribed to girls and women of childbearing potential if epilepsy could not be managed with alternative anti-seizure medications, and a pregnancy prevention programme required for girls and women using valproate. 2 In the United Kingdom from 2024, two specialists need to confirm use is appropriate before valproate can be initiated or continued. 3
Australia has not taken these same steps. Australian practice guidelines and the Australian Product Information for valproate indicates that women of childbearing potential must use effective contraception while receiving treatment. In June 2018, the Therapeutic Goods Administration asked the Advisory Committee on Medicine for further advice. The Advisory Committee reiterated standard prescribing information and noted an ongoing review in Europe of the effectiveness of risk mitigation activities. The Advisory Committee considered that educating prescribers about risks was more appropriate than introducing a pregnancy prevention programme.1,19 At this time, there were barriers to prescribing certain alternative anti-seizure medications in Australia; notably, levetiracetam and lamotrigine were available through the Pharmaceutical Benefits Scheme (PBS) as second-line treatments, 19 until this impediment to safer prescribing was addressed through changes to the PBS in January 2021. 20
The most effective contraception options are the intrauterine device (IUD) and implants, and these do not require daily attention (as for the oral contraceptive pill) or specific action before a sexual encounter (as for condoms or other barrier methods); however, women’s preferences need to be taken into account. 21 For alternative anti-seizure medication, a further consideration is that some are enzyme-inducing, which reduces the effectiveness of hormonal contraceptives such as implants and pills, but not the IUD.8,21 However, there are barriers to IUD use in Australia, including myths and reluctance among women, and a lack of promotion and relevant skills in primary healthcare.22,23 Compounding this, the first prescription for anti-seizure medication is frequently obtained from specialists, 21 who may not be routinely involved with detailed contraceptive counselling and prescribing.
Declines in valproate prescribing have been documented in many countries.6,24–26 Australian clinicians are likely to be aware of international warnings and policy changes in relation to valproate, 19 but it is unclear how prescribing practice has changed for women of reproductive age, and no evidence exists about contraception coverage at the time of treatment initiation. Therefore, we aimed to explore changes in the patterns of valproate and other anti-seizure medication dispensing to women of reproductive age and to assess what proportion of women were covered by PBS prescription contraception at treatment initiation, particularly highly effective methods.
Methods
This population-based open cohort study used data on dispensed medications that is routinely collected by the Australian government. Under the PBS, subsidised medication is provided to all Australian citizens, permanent residents and people from countries with reciprocal health insurance arrangements. We accessed a 10% random sample of the dispensing claims under the PBS, through an approved process designed to make representative data available to researchers. 27 In this instance, data were for women of reproductive age (15–49 years) between 2012 and 2021 (hence the term ‘open’ cohort, as women may enter or exit based on age). The de-identified administrative PBS data included socio-demographic characteristics of patients, prescribing/dispensing dates, medicine names, PBS item codes and quantity. We did not undertake sample size calculations as this is not expected for use of administrative datasets that contain very large numbers of individuals and typically have ample statistical power. 28 Our focus was women aged 15–49 years dispensed at least one anti-seizure medication between January 2013 and December 2021. We excluded some item codes of clonazepam with ‘restricted benefit’, which means that it can only be prescribed to patients receiving palliative care.
Anti-seizure medications
Anti-seizure medications were extracted using the Anatomical Therapeutic Chemical Classification System (ATC) code NO3 (Supplemental Table S1). Using the dispensing dates, we extracted the first recorded dispensing of valproate, or other anti-seizure medications (e.g. oxcarbazepine, lamotrigine, levetiracetam) supplied to women of reproductive age between 2013 and 2021. The dispensing data are drug specific, and our approach intentionally captured prescribing that may have been for other indications (e.g. pain, migraine), although we have used the term ‘anti-seizure medication’.
For the initial analysis of longitudinal trends, we separated valproate from other anti-seizure medications. We further stratified enzyme-inducing anti-seizure medication (i.e. carbamazepine, oxcarbazepine, perampanel, phenobarbital, phenytoin, primidone, topiramate) from non-enzyme-inducing anti-seizure medications (i.e. brivaracetam, clonazepam, ethosuximide, gabapentin, lacosamide, lamotrigine, levetiracetam, tiagabine, vigabatrin, zonisamide, stiripentol, sulthiame) owing to differences in contraceptive recommendations for these medications.
To identify newly initiated treatment with anti-seizure medication, we considered all available data back to January 2012 (the starting point for complete data, irrespective of the cost of the medication). Thus, there was a minimum of 12 months of dispensing history for all women and progressively longer for women who commenced treatment later. Valproate was considered first-line therapy if it was the first anti-seizure medication to be dispensed to a woman in the study period. If it followed dispensing of another anti-seizure medication, valproate was considered to be second-line therapy.
Contraceptive overlap
The PBS prescription contraceptives in our study include implants, levonorgestrel IUD, pills (progestogen-only pills (POP) or combined oral contraceptive pills (COCPs)) and injections. The copper IUD is not subsidised by PBS, so is not included, but it accounts for around 2% of long-acting contraceptives in the general Australian population; 29 in this article, IUD refers to levonorgestrel IUD.
Contraception coverage (the overlapping use of prescribed contraception and anti-seizure medications) at treatment initiation with anti-seizure medication was ascertained by comparing the first record of either valproate, or other anti-seizure medications script, with any contraceptives dispensed before the initial anti-seizure medications dispensing date. For example, where women had an IUD supplied up to 5 years before anti-seizure medications, and that IUD would still be effective after the dispensing date of the anti-seizure medications, they were classified as having ‘IUD overlap’. Where women had any prescription contraceptive dispensed before the anti-seizure medication, but the end date (based on 112 days’ supply of COCP and POP or 90 days’ supply for injection, 3 years for implants, and 5 years for IUD) was after the anti-seizure medications dispensing date, they were classified as pills overlap, injection overlap, implant overlap and IUD overlap. If long-acting and other contraceptives were dispensed concurrently, a hierarchy was applied. In cases where both IUD and implant were overlapping with anti-seizure medications, we chose the contraception issued nearest/closest to the anti-seizure medication commencement date. We considered contraception dispensed from 1 January 2008 to allow ascertainment of IUD dispensing up to 5 years before, and implant dispensing up to 3 years before, the first anti-seizure medications dispensing. (Data for these items were complete before 2012 because the cost of these contraceptives did not fall under the co-payment threshold that meant dispensing of some prescriptions was unrecorded.) In a sensitivity analysis, contraceptive coverage was extended to the include dispensing 60 days after anti-seizure medication.
Outcomes
The outcomes of the study include (1) annual rate of initiation of valproate and alternative anti-seizure medication among reproductive-age women between 2013 and 2021 and (2) the proportion of women covered by PBS contraception at initiation of valproate and alternatives, and also the type of contraception.
Covariates
Socio-demographic characteristics included in this study were age (15–19, 20–24, 25–29, 30–34, 35–39, 40–44, 45–49 years), state/territory where the medication was dispensed, concession card holder (yes/no) and calendar year (2013–2021).
Statistical analyses
The frequency of initiation of a new type of anti-seizure medication was calculated by taking the first dispensing of an anti-seizure medication as the numerator (multiplied by 10 to re-weight to whole-of-population estimates) and the annual estimated resident population data from the Australian Bureau of Statistics for women aged 15–49 as the denominator, and presented as a rate per 1000 women. 30 We removed the prevalent users from the denominator. We estimated the proportion of women using PBS-subsidised contraception at the time of initiating valproate, enzyme-inducing and non-enzyme-inducing anti-seizure medications. We fit logistic regression models to assess the relationship between socio-demographic factors and not having any coverage by dispensed contraception at the initiation of valproate, compared with having contraceptive coverage.
Missing data were less than 0.0005% for all the covariates included in this study; therefore, we present findings of the complete case analyses. We used Stata/MP version 18.0 (College Station, TX, USA) for analysing the data.
We have followed the RECORD statement for pharmacoepidemiology (RECORD-PE) checklist in writing of this article. 31
Ethics approval and consent to participate
This study was approved by the Monash University Human Research Ethics Committee (22877). The analysis was approved, and the manuscript was noted by the Services Australia’s external requests evaluation committee (RMS2349). We used a de-identified administrative dataset; therefore, obtaining individual consent was not possible or expected and was waived by the ethics committee. 32
Results
Between 2013 and 2021, in the open cohort that was analysed 6570 women of reproductive age initiated treatment with valproate and 18,100 commenced an alternative anti-seizure medication. Table 1 shows that in 2021, among those who initiated valproate as first-line therapy, 18% were aged 15–19 years, and 59% were concession card holders indicating disadvantaged circumstances. Between a quarter and a third of dispensed prescriptions were issued by a psychiatrist or neurologist. Prescribing occurred across Australian states and territories, broadly in line with the distribution of the population of reproductive-age women. (The demographic profile of all women in the 10% random sample of those prescribed medication in 2013 and in 2021 is presented as Supplemental Table S3.)
Demographic characteristics and prescriber for reproductive-age women initiating valproate or other anti-seizure medication in 2013 and 2021.
n.r.: not reported due to small cell size as per the requirements of the data custodians; GP: General Practitioner.
Between 2013 and 2021, overall, valproate initiation decreased substantially from 2.06 to 0.75 per 1000 women per annum. Concomitantly, treatment initiation with alternative anti-seizure medications increased from 3.1 per 1000 women in 2013 to 4.0 per 1000 women in 2021. These changes were statistically significant (test for trend, p < 0.01). As shown in Figure 1, the decline occurred in initiation of valproate as first-line therapy (from 1.8 to 0.6 per 1000 women per annum), while valproate commenced as a second-line anti-seizure medication remained consistent (around 0.2 per 1000 women).
Initiation of valproate or other anti-seizure medication as first-line therapy, and second-line use of valproate, among women by year.
In 2013, at the time of initiating valproate, 71% of women were not covered by PBS prescription contraception; this proportion was similar in 2021, at 69%. Lack of change in PBS contraception coverage is depicted in Figure 2. Lack of coverage was similar for all anti-seizure medications. Including contraceptives dispensed up to 60 days after initiation of anti-seizure medications, as a sensitivity analysis, did not appreciably change these results.
Proportion of women with (2a) no PBS prescription contraceptive coverage, (2b) IUD coverage, (2c) implant coverage, or (2d) COCP/POP coverage at the time of initiating different anti-seizure medications, by year. BS: Pharmaceutical Benefits Scheme; IUD: intrauterine device; COCP: combined oral contraceptive pill; POP: progestogen-only pill.
In 2021, those initiating valproate aged 15–19 and 45–49 years were most likely to lack contraception coverage, with odds ratios (ORs) of 2.68 (95% confidence interval (CI) 1.26–5.68) and 2.13 (95% CI 1.04–4.36), respectively, when compared with those aged 30–34 years (Table 2). Concession card status was not associated with contraception coverage at valproate initiation in 2021 (OR 0.97, 95% CI 0.64–1.48).
Characteristics of reproductive-age women initiating valproate and covered by PBS prescription contraceptives at the time of initiation in 2021 (N = 453).
CI: confidence interval; IUD: intrauterine device; OR: odds ratio; PBS: Pharmaceutical Benefits Scheme.
Use of the IUD increased markedly between 2013 and 2021, with coverage by this type of contraception applying to 13% of women initiating valproate at the end of the study period (Figure 2), slightly lower than use among those initiating enzyme-inducing anti-seizure medications (18%). Use of implants increased to a lesser extent than IUDs (Figure 2). The shift to these forms of contraception was accompanied by a decrease in the use of pills (Figure 2). Relatively few women (<2%) were dispensed contraceptive injections (data not shown).
Discussion
Main findings
This population-based cohort study using the Australian national PBS claims data showed that the initiation of treatment with valproate among women of childbearing age declined between 2013 and 2021; it was mainly replaced by an increase in the initiation of other anti-seizure medications. Specialists in neurology or psychiatry issued between a quarter and a third of initial scripts for valproate. Throughout, more than two-thirds of reproductive-age women were not covered by prescribed contraception as recorded by the PBS at the initiation of valproate or other anti-seizure medications. Thus, if we extrapolate our findings to the Australian population, approximately 4500 women initiated valproate in 2021, and about 3000 of these did not have coverage by prescribed contraception. Within the group covered by contraception, IUD use increased markedly over the study period (and implants slightly), replacing oral contraceptive pills.
Interpretation
The decline in initiation of valproate for Australian women is comparable to recent findings from the United Kingdom, Germany, France and Italy.6,25,26 For example, an English primary care database study demonstrated that 3.1 per 1000 of women (12–46 years old) were prescribed valproate in 2004, which declined to 1.6 per 1000 women by 2018. 6 The Australian trend suggests that clinicians are following the international initiatives.15–17 For example, in Australia, studies using the Australian Register for Women on Antiepileptic Drugs also reported decline in valproate use in pregnancy among women with epilepsy. 24 Given that the latest public assessment report from the United Kingdom Medicine and Healthcare Products Regulatory Agency 3 emphasises that women under age 55 years should not be initiated on valproate, with few exceptions (and require that two specialists independently consider and document that no other treatment is effective or tolerated), it is possible that further declines in first-line use of valproate could occur in Australia. However, there are concerns about the implications of discontinuing valproate for some women. The position statement of the Epilepsy Society of Australia cautions that cessation of valproate can be followed by serious uncontrollable seizures especially for women with genetic generalised epilepsy, for whom valproate provides optimal seizures control.33,34 In the United Kingdom following the 2024 restrictions, there have been calls for flexibility in prescribing to prevent life-threatening seizures. 34
Effective contraception to avoid pregnancy is important even in adolescents and those not in an ongoing relationship. 23 The Second Australian Study of Health and Relationships, drawing on a nationally representative sample, showed that by age 16 around 30% of girls had experienced vaginal intercourse, reaching 65% by age 18. 35 Furthermore, in another national survey, one in four Australian women reported having an unintended pregnancy in the past 10 years: of these 57% had not been using any contraception, 11% had used condoms and 26% had been using oral contraceptives. 36 Girls and women with epilepsy may have circumstances that reduce or increase these risks, depending on vulnerabilities, comorbidities and level of independence in daily living. For example, some women may be in 24-h care, with contraception reasonably considered unnecessary. Other women may not be in a heterosexual relationship, or they or their partner may have undergone sterilisation. While the principles of informed choice and autonomy need to be upheld, 34 our results nevertheless raise concerns about contraception coverage, particularly among young women, acknowledging that we do not know about use of contraceptives not recorded in the PBS database. Concerns about low contraception coverage among reproductive-age women initiating valproate have been reported in the United Kingdom (34% in 2018),6,37 where any use of valproate among women of reproductive age must now be in compliance with the Pregnancy Prevention Programme. 3
Five years ago, the Australian Advisory Committee on Medicine did not consider that a formal pregnancy prevention programme for women using valproate was the best course of action. 1 In the United Kingdom, such a programme involves counselling, signing a risk acknowledgement form, pregnancy testing before commencement of medication and annual pregnancy testing. 3 Recent experiences in the United Kingdom and Europe suggest mixed effectiveness of risk minimisation measures, 38 with formal pregnancy prevention programmes difficult to implement, cumbersome to administer and not necessarily well received by women. 39 It has been argued that the United Kingdom approach prioritises the risks to potential children over the safety and quality of life of people with epilepsy. 34
However, mechanisms to increase awareness of the issues and contraception counselling and uptake could be improved in Australia. The Epilepsy Society of Australia has useful resources on valproate and women, 40 epilepsy and childbearing, 41 and these could be promoted more widely. Increasing contraceptive coverage is likely to require a stronger partnership between specialists, general practitioners and others involved in primary healthcare. Greater access to public services where an IUD can be inserted with appropriate amounts of sedation and analgesia is also required, as the pain associated with unsedated insertion reduces acceptability of this method and the costs of attending a private reproductive health clinic that offer sedation services is a major barrier.22,23 Other barriers to uptake of contraception in Australia were the subject of a Senate Inquiry with the 2023 report making many recommendations to improve access, including greater scope of practice for nurses, midwives and pharmacists. 42
Strengths and limitation
The study’s strengths include the use of a nationally representative (i.e. no selection bias), routinely recorded (i.e. no recall bias), high-quality PBS drug dispensing dataset (i.e. low measurement bias compared to self-report) of over 9 years (2013–2021, recent data). All data relate to medication dispensing only and we cannot be certain about actual medication use. We did not have data on contraceptives that are not subsidised by the PBS (copper IUD, selected COCPs, barrier methods, or contraception provided directly by some health clinics or hospitals). We do not know the indications for these medications, and we did not consider alternative medications relevant to bipolar disorder that were not anti-seizure medications.
Conclusion
Overall, valproate initiation among women of childbearing age declined markedly between 2013 and 2021, with an increase in initiation of alternative anti-seizure medications. The major reduction in the use of valproate is in line with the international efforts to avoid use in women of reproductive age. While coverage by highly effective forms of contraception when commencing valproate increased over the study period, the proportion of women lacking coverage by PBS prescribed contraception was unchanged, at about two-thirds of women; this is concerning given known teratogenicity. Our finding suggests that efforts to improve pregnancy prevention and planning for women using valproate are warranted.
Supplemental Material
sj-docx-1-whe-10.1177_17455057251414916 – Supplemental material for Longitudinal Australian trends in the initiation of valproate medication among women of reproductive age and concurrent use of prescription contraceptives
Supplemental material, sj-docx-1-whe-10.1177_17455057251414916 for Longitudinal Australian trends in the initiation of valproate medication among women of reproductive age and concurrent use of prescription contraceptives by Mumtaz Begum, Lynne Giles, Jenni Ilomaki, Alice Rumbold, Michael Davies, Luke E Grzeskowiak and Vivienne Moore in Women's Health
Supplemental Material
sj-docx-2-whe-10.1177_17455057251414916 – Supplemental material for Longitudinal Australian trends in the initiation of valproate medication among women of reproductive age and concurrent use of prescription contraceptives
Supplemental material, sj-docx-2-whe-10.1177_17455057251414916 for Longitudinal Australian trends in the initiation of valproate medication among women of reproductive age and concurrent use of prescription contraceptives by Mumtaz Begum, Lynne Giles, Jenni Ilomaki, Alice Rumbold, Michael Davies, Luke E Grzeskowiak and Vivienne Moore in Women's Health
Footnotes
Acknowledgements
The authors would like to thank Services Australia for providing de-identified Pharmaceutical Benefits Scheme dispensing data and thank the Medical Research Future Fund (MRFF – Grant ID, ARGCHDG000041) for funding this project.
Ethical considerations
This study was approved by the Monash University Human Research Ethics Committee (22877).
Consent to participate
We used de-identified administrative dataset; therefore, obtaining individual consent was not possible or expected and was waived by the ethics committee.
Consent for publication
The analysis was approved, and the manuscript was noted by the Services Australia’s external requests evaluation committee (RMS2349).
Author contributions
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Medical Research Future Fund (MRFF, Grant ID ARGCHDG000041) and LEG receives salary support from a Channel 7 Children’s Research Foundation Fellowship (CRF-210323).
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability statement
The researchers do not own the data. However, request can be made to the data custodian (Pharmaceutical Benefits Scheme) for using their data for research projects, subject to ethics approval. All the authors had full access to all study data (including statistical reports and tables).
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
