Central nervous system adverse effects of new antiepileptic drugs. A meta-analysis of placebo-controlled studies

Seizure. 2008 Jul;17(5):405-21. doi: 10.1016/j.seizure.2007.12.003. Epub 2008 Feb 8.


Objective: Systematic review and meta-analysis of the most frequent treatment-emergent central nervous system adverse events (CNS AEs) of new antiepileptic drugs (AEDs) from double-blind, add-on, placebo-controlled studies conducted in adult epileptic patients and identification of dose-adverse effect relationships.

Methods: Trial reports found by searching Medline and journals. Outcome was the number of patients complaining of treatment-emergent CNS AEs. Sixteen predefined CNS AEs were considered. Risk differences (RDs) were calculated for individual studies and summary statistics estimated using the random effect model. Predefined CNS AEs in patients treated with active drug (broken down into dose levels) or placebo were extracted and the RDs (95% CI) for CNS AEs were calculated.

Results: Thirty-six suitable studies identified. No meta-analysis was possible for oxcarbazepine and tiagabine (only one study each included). For these drugs RDs were calculated from single studies. Gabapentin was significantly associated with somnolence 0.13 (0.06-0.2) and dizziness 0.11 (0.07-0.15); lamotrigine with dizziness 0.11 (0.05-0.17), ataxia 0.12 (0.01-0.24) and diplopia 0.12 (0.00-0.24); levetiracetam with somnolence 0.06 (0.01-0.11); pregabalin with somnolence 0.11 (0.07-0.15), dizziness 0.22 (0.16-0.28), ataxia 0.10 (0.06-0.14) and fatigue 0.04 (0.01-0.08); topiramate with somnolence 0.09 (0.04-0.14), dizziness 0.06 (0.00-0.11), cognitive impairment 0.14 (0.06-0.22) and fatigue 0.06 (0.01-0.12); zonisamide with somnolence 0.06 (0.02-0.11) and dizziness 0.06 (0.00-0.12). The dose-response relationship was analysed only for those CNS AEs significantly associated with the AED.

Conclusions: No comparison between drugs was possible. One CNS AE was significantly more frequent for levetiracetam, two for zonisamide and gabapentin, three for lamotrigine and four for pregabalin and topiramate.

Publication types

  • Meta-Analysis
  • Research Support, Non-U.S. Gov't
  • Systematic Review

MeSH terms

  • Anticonvulsants / adverse effects*
  • Central Nervous System / drug effects*
  • Confidence Intervals
  • Dose-Response Relationship, Drug
  • Double-Blind Method
  • Epilepsy / drug therapy
  • Humans
  • MEDLINE / statistics & numerical data
  • Odds Ratio
  • Periodicals as Topic / statistics & numerical data
  • Randomized Controlled Trials as Topic


  • Anticonvulsants