- Design
- Systematic review and meta-analysis of randomised trials
- Population
- 1,000 children with new-onset or inadequately controlled epilepsy in 8 trials
- Primary outcome
- Seizure freedom
- Effect
- RR 0.95 (95% CI 0.83–1.09), I² 59.5%; low certainty
A meta-analysis pooled eight randomised trials with 1,000 children with new or poorly controlled epilepsy. Seizure freedom did not differ between levetiracetam and valproate (5 trials, 655 children; RR 0.95, 95% CI 0.83–1.09), though heterogeneity was moderate (I² about 60%) and the certainty low; the result was sensitive to one trial.
Time-to-recurrence analyses favoured valproate but the authors judge them exploratory, because censoring was not handled properly. Levetiracetam was associated with fewer adverse events overall and less weight gain. Behavioural effects of levetiracetam and the liver safety of either drug remained uncertain.
The practical reading is that efficacy is not a strong reason to prefer either drug, so the choice turns on seizure type, side-effect profile, sex and future pregnancy potential.
- Seizure freedom was similar on levetiracetam and valproate in pooled paediatric trials.
- Expect less weight gain with levetiracetam; ask about mood and behaviour change.
- Weigh valproate's teratogenic risk carefully in girls who may later become pregnant.
- Match the drug to the seizure syndrome; generalised epilepsies may still favour valproate in boys.
Why it matters
It removes efficacy as a tiebreaker, putting the weight on harms.
Don't overread it
Low-certainty evidence that is sensitive to one trial cannot establish equivalence.
The statistics, in plain English
A risk ratio of 0.95 with an interval from 0.83 to 1.09 crosses 1.0, so the trials found no clear difference. I² of about 60% means the trials disagreed more than chance would explain, which lowers confidence in the pooled number.
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.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for neurology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free