Antidepressant use and risk of epilepsy and seizures in people aged 20 to 64 years: cohort study using a primary care database

BACKGROUND Epilepsy is a serious condition which can profoundly affect an individual's life. While there is some evidence to suggest an association between antidepressant use and epilepsy and seizures it is conflicting and not conclusive. Antidepressant prescribing is rising in the UK so it is important to quantify absolute risks with individual antidepressants to enable shared decision making with patients. In this study we assess and quantify the association between antidepressant treatment and the risk of epilepsy and seizures in a large cohort of patients diagnosed with depression aged between 20 and 64 years. METHODS Data on 238,963 patients with a diagnosis of depression aged 20 to 64 from 687 UK practices were extracted from the QResearch primary care database. We used Cox's proportional hazards to analyse the time to the first recorded diagnosis of epilepsy/seizures, excluding patients with a prior history and estimated hazard ratios for antidepressant exposure adjusting for potential confounding variables. RESULTS In the first 5 years of follow-up, 878 (0.37 %) patients had a first diagnosis of epilepsy/seizures with the hazard ratio (HR) significantly increased (P < 0.01) for all antidepressant drug classes and for 8 of the 11 most commonly prescribed drugs. The highest risks (in the first 5 years) compared with no treatment were for trazodone (HR 5.41, 95 % confidence interval (CI) 3.05 to 9.61, number needed to harm (NNH) 65), lofepramine (HR 3.09, 95 % CI 1.73 to 5.50, NNH 138), venlafaxine (HR 2.84, 95 % CI 1.97 to 4.08, NNH 156) and combined antidepressant treatment (HR 2.73, 95 % CI 1.52 to 4.91, NNH 166). CONCLUSIONS Risk of epilepsy/seizures is significantly increased for all classes of antidepressant. There is a need for individual risk-benefit assessments in patients being considered for antidepressant treatment, especially those with ongoing mild depression or with additional risk factors. Residual confounding and indication bias may influence our results, so confirmation may be required from additional studies.