Abstracts

Rationale: There is controversy surrounding the association between increasing antiepileptic drug (AED) load and the occurrence of side effects (SE). We investigated the association of AED load and the occurrence of side effects (SE) in the context of AED types, psychosocial factors, and clinical factors.Methods: The outpatient epilepsy program in a large Canadian health region prospectively gathered data from adult patients first assessments. Self-reported occurrence of SE was collected from patients on AED(s). AED load was calculated as the sum of the prescribed daily dose/defined daily dose (DDD) ratios where DDD is the assumed average maintenance daily dose of a drug for its main indication. Clinical factors included age, gender, medical comorbidity, history of epilepsy surgery, generalized versus focal seizures, epilepsy duration, mono or poly (>1) AED therapy, and seizure freedom in the prior year. Psychosocial variables included recreational drug use, alcohol use, current paid employment, presence of caregiver, having ever received psychiatric treatment, and history of learning disorder, anxiety, depression, and attention deficit disorder. AEDs were grouped into carboxamides (carbamazepine or oxcarbazepine), benzodiazepines (clonazepam or clobazam), barbiturates (primidone or phenobarbital), phenytoin, valproic acid, levetiracetam, topiramate, and gabapentin. The association between the risk of SE and these variables was investigated using Poisson regression with robust variance estimation. The results are expressed as risk ratios (RR, p-value). A sub-group analysis focusing on patients who had tried >1 AED investigated the factors associated with experiencing SE on all AEDs tried. Results: Of 963 patients taking AEDs the mean AED load was 1.32 DDD (SD=1.03 DDD) and 17.8% reported SE. Of patients who had tried ?2 AEDs 13.7% had SE on all AEDs. After adjusting for confounders the risk of SE increased with AED load (16% per 1.0 DDD increase, p=0.029). DDD and current poly AED therapy were associated and so only DDD was included. SE were less likely in patients with a history of learning disorder (0.48, p=0.011); however, when adjusting for AED types, AED load was no longer associated with SE. Instead, SE were more likely with phenytoin (1.41, p=0.039) and topiramate (1.94, p=0.004), while a history of learning disorder remained protective (0.51, p=0.022). Patients having been on ?2 AEDs were more likely to have SE on all AEDs if they used alcohol (1.88, p=0.012). Conclusions: Prior studies have concluded that AED load is not associated with increased risk of SE but that certain psychosocial factors are more important. However, this could be explained by high covariance between AED load and number of AEDs. We found that AED load is associated with increased risk of SE in addition to psychosocial factors. However, the type of AED is a more important determinant of SE than AED load alone. Further investigation is needed to determine whether AED types are associated with SE only in certain patients and if certain factors are protective of SE on given AEDs.