Seizure Recurrence After Withdrawal in Children With Epilepsy

What was studied

This prospective cohort study followed 318 children with epilepsy during antiseizure medication withdrawal. All had been free of clinical seizures for at least 24 months and had completely normal EEGs before withdrawal.

Researchers measured seizure recurrence and used a statistical model to identify clinical factors linked to recurrence. They built these factors into a web-based tool intended to estimate an individual child's risk.

What they found

Seizures returned in 82 children (25.8%). The reported mean time to recurrence was 4.70 ± 13.96 months. Among 35 children undergoing a second withdrawal attempt, 11 (31.4%) had a recurrence.

After accounting for the other factors in the model, recurrence was associated with status epilepticus (hazard ratio [HR] 2.718; 95% confidence interval [CI] 1.193–6.194; p=0.017), febrile seizures (HR 2.300; 95% CI 1.214–4.356; p=0.011), and longer active epilepsy duration (HR 1.022 per month; 95% CI 1.016–1.027; p<0.001).

Compared with seizure freedom of 24 months or less before withdrawal, seizure freedom of 25–35 months was associated with a lower recurrence hazard (HR 0.231; p=0.001), as was seizure freedom of at least 36 months (HR 0.232; p=0.002). Both represent about a 77% lower hazard, not a 77-percentage-point drop in the chance of recurrence.

The model's ability to distinguish children with higher versus lower recurrence risk was reported as acceptable: its C-index was 0.808, or 0.769 after correction for overly favorable performance estimates from the data used to build it.

Limits of the evidence

This observational study identifies associations; it cannot prove that waiting longer before withdrawal causes a lower recurrence risk. The findings apply to children who had at least 24 months without clinical seizures and normal pre-withdrawal EEGs, not necessarily to children with abnormal EEGs or other patient groups.

The abstract does not report the overall follow-up duration, age range, epilepsy types, withdrawal schedules, or how active epilepsy duration was defined. Without a stated follow-up period, 25.8% should not be interpreted as a one-year or lifetime recurrence risk. Confidence intervals were not provided for the seizure-free-duration comparisons. The abstract also does not report testing in a separate patient population or whether using the web tool improves care.

For families and caregivers

In this selected group, 25.8% of children had seizures return despite at least 24 months of seizure freedom and a normal EEG before withdrawal. A normal EEG therefore did not guarantee that seizures would stay away. The findings may help inform discussions about withdrawal by considering seizure history and seizure-free duration together, but the tool does not establish whether withdrawal is appropriate for a particular child.

What to watch next

Testing the model in independent groups of children, with clearly stated follow-up periods, would strengthen evidence that its individual risk estimates are reliable.

Terms in this summary

Antiseizure medication

Medicine used to prevent or reduce seizures.

EEG

A test that records the brain's electrical activity; a normal result does not rule out future seizures.

Prospective cohort study

A study that follows a group forward in time to observe outcomes.

Status epilepticus

A prolonged seizure or repeated seizures without recovery between them; it is a medical emergency.

Febrile seizure

A seizure associated with fever in a young child.

Hazard ratio

A comparison of how quickly an event occurs over time. A value above 1 indicates a higher hazard, and a value below 1 indicates a lower hazard; it is not an absolute probability.

95% confidence interval

A range showing statistical uncertainty around an estimate; wider intervals mean less precision.

C-index

A measure of how well a model ranks people by risk. A value of 0.5 is about chance, and 1.0 is perfect ranking; it does not by itself show that predicted probabilities are accurate.