According to the findings of a nested case-control study, the use of newer antiepileptic drugs (AEDs) associated with a high frequency of depression in clinical trials appears to also increase the risk of self-harm or suicidal behaviour.  The data do however have a number of limitations, which need to be borne in mind when interpreting the results.

The authors note that the US FDA issued a safety alert in January 2008 on an increased risk of suicidal thoughts and behaviour associated with AEDs.  This was based on a meta-analysis which grouped AEDs together as a class and critics have suggested that this could have masked differences between the individual agents.  The purpose of this study therefore was to determine the risk of self-harm or suicidal behaviour associated with the use of different AEDs, and whether this risk is only evident in newer AEDs that have been associated with a higher frequency of newly occurring depressive symptoms in clinical trials.

This observational study used data from the United Kingdom General Practice Research Database (GPRD) to identify patients who received at least one prescription for AEDs for epilepsy or non-febrile seizures during the study period (Jan 1990-Sep 2005).  Cases were those patients with a medical code for self-harm or suicidal behaviour; those who died were potential cases if suicidal thoughts were recorded <4 weeks of their death.  Records of potential cases were blinded to treatment with AEDs and validated manually.  Up to 20 controls were selected for each case from the cohort members, matched according to age, sex and year of cohort entry.   

AEDs were classified into 4 groups: barbiturates, conventional AEDs (carbamazepine, valproate, phenytoin, ethosuximide, acetazolamide), and newer AEDs with low (lamotrigine, gabapentin, pregabalin, oxcarbazepine) or high (levetiracetam, tiagabine, topiramate, vigabatrin) potential of causing depression. Adjusted odds ratios (OR) were calculated using conditional logistic regression.  This classification of the new AEDs was based on the frequency of occurrence of depressive symptoms in clinical trials (≤1% and >1%, respectively).  Exposure was defined as current (a prescription lasting into the 14-day period before the index date), recent exposure (prescription supply ended 15-183 days before the index date), and past exposure (supply ended 184-365 days before the index date). 

A total of 44,300 patients fulfilled the cohort definition, among which 453 validated cases of self-harm or suicidal behaviour were identified and matched to 8,962 controls.  The main findings were as follows:

• Current use of newer AEDs with a high potential of causing depression was associated with a 3-fold increased risk of self-harm/suicidal behavior (adjusted OR=3.08; 95% CI 1.22–7.77) as compared with no use of AEDs during the last year.

• Use of barbiturates (OR=0.66; 95% CI 0.25–1.73), conventional AEDs (OR=0.74; 95% CI 0.53–1.03), or low-risk newer AEDs (OR=0.87; 95% CI 0.47–1.59) was not associated with an increased risk. 

• In an analysis of individual AEDs, an increased risk estimate was only observed for current use of levetiracetam (2 cases [0.4%] and 8 controls [0.1%]; adjusted OR 6.42; 95% CI 1.24-33.36)

• In the stratified analyses, an increased risk of self-harm or suicidal behaviour associated with current use of new AEDs carrying a high risk of depression was only observed in patients with psychiatric co-morbidity (OR = 8.48; 95% CI 1.86–38.58)

The authors discuss a number of limitations to their study, for example the low number of exposed cases, the lack of validated information on the severity of epilepsy (which may potentially be a confounder), and lack of data on actual use of AEDs (only prescription data available but no details of compliance).  They say that the risk estimate for newer AEDs with a higher risk of causing depression needs to be confirmed in additional studies. In addition the risk estimates for individual AEDs was based on low numbers of exposed cases and should be interpreted with caution. 

The authors of an accompanying editorial also caution interpretation of the results and point out three major concerns: the diagnosis of epilepsy has never been specifically validated in the GPRD cohort; for a large part of the study period the high-risk AEDs were licensed only as add-on therapy for partial epilepsy (suggesting patients received them as part of polytherapy regimens, therefore they may have had drug-refractory epilepsy); and detailed description of psychiatric co-morbidity is lacking. They comment that the current study “represents a good initial attempt but the issues are probably more complex than is at first apparent. In fact, the high psychiatric comorbidity in the GPRD population investigated (especially alcohol and drug abuse, which are closely related to suicidal behavior) may suggest an increase in “background noise” that may invalidate the results. More appropriate studies are needed in order to clarify this and come back to the first goal of a physician: primum non nocere”.