Abstracts

Rationale: The FDA recently linked antiepileptic drug (AED) exposure to suicidality based on meta-analysis of randomized clinical trials. Unfortunately not all AEDs were represented in the analysis, the analysis did not account for the role of mental health conditions, and older patients were poorly represented in the analysis. Methods: We used national VA databases to identify veterans 65 years and older who had new exposure to an AED between October 2004-September 2006. Individuals without exposure were included as potential controls. We created a propensity score predicting new AED prescriptions to match cases of suicidality (identified using ICD-9 codes) and controls. Co-variates adjusted for in the propensity model included individual diagnoses for physical and mental health conditions, use of anti-depressant and antipsychotic medications, prior suicidality, prior psychiatric hospitalization. We matched each suicidality case to up to 10 random sample controls. Results: From the population of approximately 1.5 million older veterans we found 609 cases of suicidality. Conditional logistic regression analysis found that seven of the nine AEDs examined in this study were not associated with an increase in suicidality, when potential confounders were taken into account. Compared to individuals without AED exposure, only patients receiving lamotrigine (OR 14.4; 95% CI 3.1-65.6) and valproate (OR 4.5; 95% CI 2.4-8.2) were at elevated risk for suicidality. In this cohort where all patients with epilepsy were newly diagnosed, epilepsy did not increase risk for suicidality (OR 2.22; 95% CI 0.3-14.9). Conclusions: After rigorous control for psychiatric comorbidity, our data suggest that all AEDs are not associated with suicidality in indivduals with incident AED exposure. Only lamotrigine and valproate were significantly associated with suicidality compared to no AED exposure. While it is possible that this increased risk for individuals with lamotrigine or valproate exposure is due to undiagnosed depression, it is not likely that an AED would be the only depression treatment for such a condition. Moreover, confounding may be considered for a relationship with an odds ratio less than two, however, the odds in this case are over four, making confounding significanlty less likely to explain this relationship. Because our population is predominantly male, our findings suggest that additional analyses are needed to examine the relationship between AED exposure and suicidality among other populations affected by the FDA warning.