What is already known on this topic

Evidence of an increase in suicidal behaviours in children and adolescents prescribed antidepressants resulted in a Food and Drug Administration (FDA) ‘black box’ warning in 2004 for this age group. This was extended to young adults up to the age of 24 years in 2007. A link with suicide has not been clearly demonstrated for any age group, nor have differences in the risk of these behaviours according to the type of antidepressant. Large cohort studies overcome the limitations of trials to establish these links and while not clearly demonstrating differences in risk according to the class of antidepressant, there is emerging evidence of an increased risk according to the individual agent, with venlafaxine being a particular concern.1

Methods of the study

This study is a cohort study of 238 963 patients from UK general practices aged 20–64 years. Patients had a first recorded diagnosis of depression (using Read codes of terms indicative of major depression as used in general practice—http://systems.hscic.gov.uk/data/uktc/readcodes) between 1 January 2000 and 31 July 2011 and were followed up until 1 August 2012, although most analyses are restricted to the first 5 years of follow-up. The association between antidepressant use and suicide attempt or self-harm and suicide (based on death certificates and medical records) was examined using Cox proportional hazards models to calculate HRs adjusting for potential confounders.

What does this paper add

• This study is the largest prospective study of its kind to investigate suicide and suicide attempt and self-harm separately and according to both class of antidepressant and specific antidepressant. It uses reliably recorded information on prescriptions throughout the follow-up period so that antidepressant exposure can be treated as a time-varying exposure accounting for patients starting and stopping treatment.

• Restricting inclusion to those with a diagnosis of depression associated with an increased risk of suicide, reduced indication bias. Indication bias occurs when prescription of the medication is for a condition that is associated with the outcome of interest.

• Results confirm earlier studies2 showing no difference in risk between selective serotonin reuptake inhibitors and tricyclic antidepressants in adults.

• Compared with citalopram (a selective serotonin reuptake inhibitor), the risk of suicide was significantly higher for those on mirtazapine and the risk of attempted suicide and self-harm was significantly higher for those on venlafaxine, trazodone and mirtazapine.

Limitations

• While there has been adjustment for a range of confounders, there may be other confounders not accounted for. Particular antidepressants may have been prescribed in response to patient characteristics not adjusted for in the analysis and these characteristics may account for the association between that antidepressant and the increased risk.

• Depression severity is not routinely recorded in general practice so that adjustment of this potential confounder was based on Read codes. Further, this was accounted for at baseline only and not in terms of changes in severity over time.

• While very large, this study may still not be adequately powered, given suicide is a rare event; some have suggested that a sample size of more than 300 000 is required.2

What next in research

Measures of potential moderators (eg, severity of depression, previous suicide attempts, risk factors for suicide-related behaviours) and mediators (agitation) should be included in future prospective cohort studies and clinical trials to elucidate the nature of the associations between specific antidepressants and increased risk of suicide, suicide attempt and self-harm. The timing of the association is also important. While this study demonstrated higher rates just after starting/stopping treatment, previous studies have shown that the highest rates are just before starting an antidepressant concluding that, overall, antidepressant treatment is probably improving outcomes, given depression itself is associated with an increased risk of suicide and suicidal behaviour.2

Do these results change your practices and why?

For those working with children and adolescents, where there is little evidence of clinically meaningful benefits in children and adolescents,3 and FDA warnings about the use of antidepressants, recommendations have long existed regarding close monitoring for the emergence of suicide-related adverse outcomes. The results from this study highlight the need for those working with adults to also be vigilant to the risk of suicide and suicidal behaviour after prescription of an antidepressant. This increased risk needs to be actively managed, including clear communication of the risks and benefits of antidepressant medications.4