Design: Retrospective cohort, US Medicaid claims 2016–2021
Population: 21,923 people with epilepsy of child-bearing potential on susceptible hormonal contraception
Primary outcome: At least 14 days' overlap with an enzyme-inducing antiseizure medicine
Effect: 32.6% overlapped; median 218 days

This retrospective study of US Medicaid claims from 2016 to 2021 identified 110,976 people with epilepsy of child-bearing potential taking antiseizure medicines. Of 21,923 prescribed a hormonal contraceptive whose effect can be reduced by hepatic enzyme induction, 32.6% had at least two weeks of overlap with an enzyme-inducing antiseizure medicine. The median overlap was 218 days.

Overlap was more likely in people with intellectual disability (adjusted OR 2.33), drug-resistant epilepsy (1.73) and those over 35 (1.45) — groups who may be on older drugs, see several prescribers, or have less say in contraceptive choices.

Carbamazepine, phenytoin, phenobarbital, oxcarbazepine and topiramate at higher doses reduce the effectiveness of combined pills, progestogen-only pills and implants. This is a long-known interaction, and the study shows how often it is still missed.

In India, carbamazepine, phenytoin and phenobarbital remain widely prescribed, which makes the same gap likely here. Claims data cannot show counselling, condom use or pregnancy intention, so not every overlap was a failure of care.

Ask every person of child-bearing potential on an antiseizure medicine which contraception they use.
Check whether their antiseizure medicine is an enzyme inducer, including topiramate at higher doses.
Where an inducer is needed, discuss a copper or levonorgestrel intrauterine device or medroxyprogesterone injection.
Give extra attention to patients with intellectual disability, drug-resistant epilepsy or several prescribers.
Record the contraceptive method on the epilepsy plan so every prescriber sees it.

Why it matters

A well-known interaction remains common enough to put one in three of these patients at risk of contraceptive failure.

The statistics, in plain English

The 32.6% is a descriptive proportion from claims data, not an estimate of unplanned pregnancies. The adjusted odds ratios show which groups were more likely to have overlap; an odds ratio of 2.33 means more than double the odds, after accounting for other recorded factors.