Intended for healthcare professionals

Editorials

# Antidepressants and suicide

BMJ 2008; 336 (Published 06 March 2008) Cite this as: BMJ 2008;336:515
1. Gregory Simon, investigator
1. 1Center for Health Studies, Group Health Cooperative, Seattle, WA 98112, USA
1. Simon.g@ghc.org

Treatment is probably too sporadic to affect overall suicide rates

Two accompanying observational studies by Biddle and colleagues and Wheeler and colleagues add to the recent literature regarding changes in antidepressant use and changes in suicide rates.1 2 The current controversy began in2003,whenreanalyses of data from clinical trials raised concerns that antidepressants might precipitate suicidal thoughts or attempts at suicide. Children and adolescents starting treatment with several newer antidepressants had a 4% risk of developing suicidal ideation or behaviour, compared with 2% in those receiving placebo.3 However, clinical trials cannot determine whether antidepressants increase or decrease the risk of genuine suicide attempts or death from suicide because these outcomes are, fortunately, too rare. No deaths from suicide and few attempts at suicide have occurred to date in trials of antidepressants in adolescents,3 so any clinical trial that could reliably assess effects on death from suicide would require several times more participants than have been included in all such trials to date.

Any study of serious suicide attempts or death from suicide must fall back on observational analyses of data from large populations. What population based studies gain in sample size, however, they lose in ability to account for bias or confounding. For example, if we simply compare suicide rates between people who are using and not using antidepressants, we would erroneously conclude that antidepressants increase risk. That conclusion would ignore the fact that symptoms of depression (including suicidal ideation) are the reason that antidepressants are prescribed. Although more sophisticated observational studies try to account for bias or confounding, they are never as good as randomised trials.

Bearing in mind these limitations, several observational studies have suggested that antidepressants might actually reduce the risk of attempted suicide or death from suicide. In adolescents and adults, the risk of suicide declines sharply after starting treatment with antidepressants.4 Areas with higher rates of antidepressant prescribing tend to have lower suicide rates.5 As the use of antidepressants in adolescents increased in the United States between 1990 and 2003, suicide rates declined. And when warnings led to decreased antidepressant use between 2003 and 2004, the suicide rate in US adolescents increased for the first time in a decade.6

More recent observational studies, however, raise doubts about an association between suicide rates and changes in antidepressant prescribing. The most recent US data suggest that adolescent suicide deaths began to decrease again between 2004 and 2005.7 The two accompanying observational studies show a lack of connection between antidepressant prescribing rates and suicide rates in adolescents and young adults in the United Kingdom.1 2 Suicide rates declined when antidepressant use steadily increased but continued to decline when the use of antidepressants fell sharply in 2003 and 2004.

In truth, it would be surprising if antidepressants had any effect—positive or negative—on the risk of suicide in the general population. In the US, only half of adults and a quarter of adolescents who have a major episode of depression in any given year start taking an antidepressant.8 Nearly half of all adults who start treatment discontinue after just a few weeks.9 Only 3% of adolescents dying by suicidein New York City had toxicology data showing recent use of antidepressants.10 Sustained use of antidepressants is probably too rare to have much overall effect on risk of suicide in people living with depression.

The shifting association between antidepressant prescribing rates and suicide rates argues for caution in interpreting other ecological associations. Although we can see that changes in suicide mortality are associated with changes in social conditions or population rates of substance use, ecological associations do not imply causality. We tend to search for or believe only those associations that fit in with our expectations.

Observational data do clearly show that warnings in the UK and the US reduced rates of antidepressant prescribing, especially in adolescents. Between 2003 and 2005, antidepressant use in adolescents declined by about 20% in the US, 30% in the Netherlands, and 40% in the UK.1 2 6

Unfortunately, we can find no evidence that regulators’ urgent recommendations for closer monitoring of treatment led to any improvement in practice. The US Food and Drug Administration urges weekly or biweekly follow-up after starting antidepressant treatment. National quality measures in the US indicate that only a fifth of patients starting antidepressants have even three follow-up visits over three months.9 Furthermore, follow-up rates have shown no improvement over the past five years. We can only hope that regulatory warnings will eventually have as much effect on the quality of treatment as on the quantity.

## Footnotes

• Research, doi 10.1136/bmj.39462.375613.BE
• Research, doi 10.1136/bmj.39475.603935.25
• Competing interests: During the past five years Dr Simon has received $750 (£382;€507) from Wyeth Pharmaceuticals for consultation regarding antidepressant adherence and$1137.50 from Bristol Myers Squibb for consultation regarding suicide during psychotropic drug treatment. He has no active or planned consulting arrangements and no other financial interests to declare.

• Provenance and peer review: Commissioned; not externally peer reviewed.

View Abstract