Data-driven study reveals falling use of antiseizure drug in pregnancy

University of Liverpool researchers have contributed to the largest study to date on sodium valproate use. This significant report shows decreased use in women of childbearing age and during pregnancy, as well as young men, due to safety concerns.

The new research suggests that the use of sodium valproate by young women, including during pregnancy, has continued to fall, in line with changes to policy following safety concerns. Researchers at the BHF Data Science Centre at Health Data Research UK (HDR UK) analysed whole-population data from England and Wales, marking the largest ever study on this medicine.

Sodium valproate – an important drug used to treat epilepsy and bipolar disorder – has a long and controversial history due to safety concerns, particularly for pregnant women, as it can lead to miscarriage, birth defects and developmental conditions including ADHD and autism.

The researchers analysed de-identified NHS health records across England and Wales to review current and new uses of sodium valproate, as well as other anti-seizure medications, dispensed before and during the COVID-19 pandemic. Ethical access to the data, held securely in NHS England’s Secure Data Environment service and the SAIL Databank Trusted Research Environment, was enabled via the BHF Data Science Centre.

The large-scale study, published in BMJ Medicine, suggests that sodium valproate use by young men is also decreasing. The decline follows new evidence suggesting a potential link between valproate use by men and a small increased risk of a range of neurodevelopmental disorders in their children.

By securely accessing linked data on a national scale, researchers from the universities of Liverpool, Swansea, UCL, Edinburgh, Strathclyde, King’s College London, Keele and Leicester were able to produce a map illustrating regional variation in sodium valproate dispensing across England and Wales. This showed clear differences in dispensing patterns of sodium valproate to women of childbearing potential by local authority district – with higher rates in the north west and across east coastal regions – highlighting potential inequalities worth further investigation.

This research demonstrates the enormous potential for data science approaches to studying the use of medicines, detecting safety issues and patients at risk. These data-driven methods could also help us monitor broader changes in prescribing.

Professor Reecha Sofat, Associate Director at the BHF Data Science Centre, professor of clinical pharmacology and therapeutics at the University of Liverpool, and senior author, said: “For the first time, we can really assess the real-world use of medicines by different groups following changes in policy.

“We believe that expanding our approach – for example, by linking anonymised mums’ health data to their children’s health records – is a real opportunity to improve evidence about medicines used during pregnancy. Using real-world data in this way means that we can avoid the challenges of traditional research methods, reducing time taken to make findings, and improving inclusion of underserved groups.”

The findings suggest that use of sodium valproate in women of childbearing potential decreased, while use of most other anti-seizure medications increased, suggesting doctors were responding to the MHRA guidance. Between 2019 and 2022, new initiation of sodium valproate per 100,000 women fell from 7 to 5 in women aged 15-19, and 11 to 7 in women aged 20-29. It also fell from 14 to 7 in women aged 30-39. Pregnancy rates decreased from 6.0 to 5.2 per 1,000 women dispensed sodium valproate over the same period.

There was no evidence that epilepsy-related deaths increased in women aged 15-49 during 2015-2022, but there was some evidence for a slight increase in men during the period April 2018 to December 2022. However, the authors caution that further investigations of this emerging trend are needed, before any conclusions can be drawn.

Dr Caroline Dale, Research Fellow at the University of Liverpool and first author, said: “Pregnant women are often excluded from traditional research methods, such as clinical trials, given understandable concerns about unknown risks – this means that we have huge gaps in our knowledge of the outcomes of safe drug use in this group.

“Our data-led approach is non-invasive and could be expanded to understand maternal health in pregnancy better, linking this to child health records that would help us to understand the real-world effects of medicines.”

The paper, 'Use of sodium valproate and other antiseizure drug treatments in England and Wales: quantitative analysis of nationwide linked electronic health records' was published in BMJ Medicine. (DOI: 10.1136/bmjmed-2023-000760).