Effects of epilepsy treatments on sleep architecture and daytime sleepiness: an evidence-based review of objective sleep metrics.

OBJECTIVE

Sleep is considered restorative, and good quantity and quality sleep is required for memory consolidation and synaptic plasticity. Sleep disorders are common in patients with epilepsy. Poor sleep quality or quantity may worsen seizure control. On the other end, seizures and epilepsy may worsen the sleep quality and set a vicious cycle. In addition, antiepileptic drugs have an effect on sleep architecture. We performed a systemic literature review with a goal to evaluate the effect of antiepileptic drugs and nondrug treatments for epilepsy on sleep architecture to help better understand treatment effects, especially in patients with epilepsy and sleep problems.

METHODS

We searched PubMed and identified studies that evaluated objective sleep outcomes for an antiepileptic drug. We also searched for studies with objective sleep outcomes that evaluated other epilepsy treatments such as epilepsy surgery, vagus nerve stimulation, and ketogenic diet.

RESULTS

The studies were categorized based on evidence class and study population for an individual antiepileptic drug or treatment. We identified that most antiepileptic drugs and nondrug treatments for epilepsy affect sleep architecture.

SIGNIFICANCE

We identified that gabapentin, tiagabine, pregabalin, clobazam, and carbamazepine reduce sleep latency and/or improve sleep efficiency. Phenobarbital, valproic acid, and higher-dose levetiracetam aggravate daytime sleepiness, whereas topiramate and zonisamide do not. Vagus nerve stimulation reduces daytime sleepiness, and ketogenic diet improves slow-wave sleep. Epilepsy surgery may improve nocturnal sleep only in a subgroup of patients with improved seizure frequency. Further studies are needed to evaluate the dose-dependent sleep effects of antiepileptic drugs and nondrug treatments independent of the improvement of epilepsy, and to identify if these changes are clinically significant.