Article type
Year
Abstract
Background: Adverse Events (AEs) are reported in randomised controlled trials (RCTs) of epilepsy and non-epilepsy trials using the same antiepileptic drug (AED). Meta-analysis of these trials would increase study power.
Objectives: To determine if AE outcomes from RCTs of AEDs across non-epilepsy indications (neuropathy or migraine) can be meta-analysed with data from epilepsy trials.
Methods: We searched databases for RCTs meeting inclusion criteria. AEDs included were topiramate, gabapentin, valproate, oxcarbazepine, lacosamide, lamotrigine, carisbamate, zonisamide and pregabalin. Extracted data was analysed using RevMan V5.0. Common AEs analysed were; dizziness, ataxia, headache, fatigue, nausea, somnolence, withdrawals due to AE and any AE. Summary statistics of effect size were calculated using the Mantel-Haenszel method. Statistical heterogeneity was assessed using a random effects model. To test between indications we used a fixed effects model and calculated an I2 statistic.
Results: We included 106 RCTs for analysis. Effect sizes varied with drug and outcome. When dizziness was analysed, test between indications showed no statistical heterogeneity (I2 = 0$%) for gabapentin, topiramate, lacosamide and lamotrigine. However, heterogeneity was significant (I2 = 59%) for trials of oxcarbazepine. When somnolence was the AE outcome, heterogeneity was insignificant for oxcarbazepine (I2 = 8%), lacosamide (I2 = 0%) and topiramate (I2 = 0%), but significant for gabapentin (I2 = 56%) and lamotrigine (I2= 60%). When nausea was analysed there was no heterogeneity (I2 = 0%) for lacosamide, oxcarbazepine and topiramate, but significant heterogeneity existed for gabapentin (I2 =41%) and lamotrigine (I2 =40%). In instances where there was significant heterogeneity, the size of relative risk was greater in the non-epilepsy indications.
Conclusions: AEs of AEDs from non-epilepsy trials could be used in meta-analysis given the absence of statistical heterogeneity for some interventions and outcomes. Effect sizes were larger in the non-epilepsy indications overall. Further meta-regression would unmask any dose effect on heterogeneity and effect size.
Objectives: To determine if AE outcomes from RCTs of AEDs across non-epilepsy indications (neuropathy or migraine) can be meta-analysed with data from epilepsy trials.
Methods: We searched databases for RCTs meeting inclusion criteria. AEDs included were topiramate, gabapentin, valproate, oxcarbazepine, lacosamide, lamotrigine, carisbamate, zonisamide and pregabalin. Extracted data was analysed using RevMan V5.0. Common AEs analysed were; dizziness, ataxia, headache, fatigue, nausea, somnolence, withdrawals due to AE and any AE. Summary statistics of effect size were calculated using the Mantel-Haenszel method. Statistical heterogeneity was assessed using a random effects model. To test between indications we used a fixed effects model and calculated an I2 statistic.
Results: We included 106 RCTs for analysis. Effect sizes varied with drug and outcome. When dizziness was analysed, test between indications showed no statistical heterogeneity (I2 = 0$%) for gabapentin, topiramate, lacosamide and lamotrigine. However, heterogeneity was significant (I2 = 59%) for trials of oxcarbazepine. When somnolence was the AE outcome, heterogeneity was insignificant for oxcarbazepine (I2 = 8%), lacosamide (I2 = 0%) and topiramate (I2 = 0%), but significant for gabapentin (I2 = 56%) and lamotrigine (I2= 60%). When nausea was analysed there was no heterogeneity (I2 = 0%) for lacosamide, oxcarbazepine and topiramate, but significant heterogeneity existed for gabapentin (I2 =41%) and lamotrigine (I2 =40%). In instances where there was significant heterogeneity, the size of relative risk was greater in the non-epilepsy indications.
Conclusions: AEs of AEDs from non-epilepsy trials could be used in meta-analysis given the absence of statistical heterogeneity for some interventions and outcomes. Effect sizes were larger in the non-epilepsy indications overall. Further meta-regression would unmask any dose effect on heterogeneity and effect size.