Stopping antiseizure drugs: relapse was likelier with under five years seizure-free

Design

Systematic review and meta-analysis of mostly observational studies

Population

22 studies, 3,933 people with epilepsy withdrawing antiseizure drugs

Primary outcome

Seizure relapse after withdrawal by seizure-free period

Effect

Under 5 years seizure-free: RR 1.49 (95% CI 1.17–1.88)

This systematic review identified 45 studies of antiseizure drug withdrawal and pooled 22 of them, 3,933 patients, to ask how the length of seizure freedom before withdrawal affects relapse.

Across all patients, withdrawal after less than five years of seizure freedom carried a higher relapse risk than after longer periods (RR 1.49, 95% CI 1.17–1.88). In subgroups, the thresholds differed: children with under three years of seizure freedom and adults with under five years both had higher relapse risks (RR 1.70 and 1.52).

The evidence comes from heterogeneous, mostly observational studies, and the subgroup thresholds are exploratory. The finding supports a longer seizure-free interval before stopping in adults than is sometimes used, but the decision remains individual: epilepsy syndrome, EEG, imaging, driving and the patient's own priorities all weigh in.

• Discuss relapse risk explicitly before withdrawing antiseizure drugs, with the seizure-free interval as one factor.
• In adults, a seizure-free period of five years or more was associated with lower relapse risk.
• Use a validated relapse-risk calculator alongside syndrome, EEG and imaging findings.
• Taper gradually and advise on driving and safety during and after withdrawal.
• Agree a plan for what to do if a seizure recurs.

Why it matters

The commonly quoted two-year seizure-free threshold may be too short for many adults.

Don't overread it

Subgroup thresholds by age and region are exploratory and should not be used as fixed cut-offs.

The statistics, in plain English

A relative risk of 1.49 means relapse was about half as likely again after shorter seizure freedom. The confidence interval of 1.17–1.88 excludes no effect, but the pooled studies differed in design and population.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.