A new review of antidepressant studies has thrown up another challenge to the US regulatory requirement for ‘black box’ warnings on antidepressant drugs about an increased risk of suicidal thoughts and behaviour in children and adolescents.

“While I support the Food and Drug Administration’s role in monitoring the safety of medications, in this case the FDA should reconsider the black box warning on these medications,” commented David Brent, professor of psychiatry at the University of Pittsburgh School of Medicine and lead author of the study published in the Journal of the American Medical Association (JAMA). The agency says it sees no need for a labelling change.

The FDA introduced the black box warning in October 2004, based on a meta-analysis showing that the average risk of suicidal behaviour or ‘suicidal ideation’ in more than 4,400 children and adolescents taking antidepressants was 4% compared with 2% on placebo. The move was heavily criticised by psychiatrists and others, who warned that it could put young people with depression at greater risk by discouraging off-label prescribing or scaring candidates for therapy and their parents away from treatment.

The review conducted by Dr Brent et al also found an increased risk of suicidal tendencies in children and adolescents taking antidepressants, but it was much smaller. Unlike the FDA, the University of Pittsburgh researchers also looked at the benefits of antidepressant use in this population, concluding that the overall risk-to-benefit profile for the treatment of major depressive disorder (MDD), obsessive-compulsive disorder (OCD) and non-OCD anxiety disorders was favourable.

Marked differences

There were some other marked differences between the FDA review and the latest meta-analysis published in JAMA. For one thing, it included seven additional studies, three of which were not available at the time of the FDA report. Moreover, the researchers used a different statistical approach, which produced a lower estimate of risk.

The University of Pittsburgh study, which was supported by the US National Institute of Mental Health, looked at 27 randomised, placebo-controlled clinical trials of second-generation antidepressants (i.e., selective serotonin reuptake inhibitors plus nefazodone, venlafaxine and mirtazapine) in children and adolescents under 19 years of age with MDD (15 trials), OCD (six) and non-OCD anxiety disorders (six).

In terms of efficacy, pooled absolute rates of response were 61% for participants with MMD on antidepressants, compared with 50% on placebo. This gave a pooled risk difference of 11% and a number needed to treat (NNT) – meaning the number of patients who would need to be treated to prevent an adverse outcome in one patient – of 10. By contrast, the pooled absolute rate of suicidal ideation/suicide attempts in MMD was 3% on antidepressants and 2% on placebo. This generated a pooled risk difference of 1% and a number needed to harm (NNH) – i.e., the number of patients who would need to be treated for one to come to harm – of 112.

For OCD, the pooled rates of response were 52% on antidepressants and 32% on placebo, giving a pooled risk difference of 20% and an NNT of six. Pooled rates of suicidal ideation/suicide attempts were 1% on antidepressants and 0.3% on placebo, making a pooled risk difference of 0.5% and an NNH of 200. In the treatment of non-OCD anxiety disorders, pooled rates of response were 69% on antidepressants and 39% on placebo, giving a pooled risk difference of 37% and an NNT of three. Pooled rates of suicidal ideation/suicide attempt were 1% on antidepressants and 0.2% on placebo, producing a pooled risk difference of 0.7% and an NNH of 143.

While these figures pushed the overall risk-benefit equation for antidepressants in children and adolescents firmly into the positive, the more moderate increase in risk compared with the FDA’s estimates was strongly flavoured by the different statistical approach taken in the new study. Whereas the FDA mainly used fixed-effects models to combine data from the randomised trials (while applying random-effects models as part of a sensitivity analysis), Dr Brent et al relied on random-effects models, which “tended to yield smaller risk ratios and risk differences”, they noted.

When the University of Pittsburgh team re-analysed the data using fixed-effects models, the results were far closer to the estimates in the FDA review. As an illustration, the pooled risk difference for suicidal ideation/suicide attempts across all of the antidepressant trials in the new meta-analysis was 0.7% on a random-effects basis but 1.3% when re-analysed with a fixed-effects model. The pooled risk difference in the FDA review was 1.5% using a fixed-effects model but 0.8% on a random effects basis.

Random effects approach more appropriate

“While either approach to meta-analysis is defensible,” Brent et al commented, “we believe that a random effects approach is more appropriate for pooling data because it does not assume that there is a common effect across all studies. In light of quantitative and qualitative evidence of heterogeneity among trials, this is a more conservative assumption and provides estimates of risk differences that are less vulnerable to heterogeneity.”

Although the University of Pittsburgh review still found an increased risk of suicidal tendencies in adolescents and children taking antidepressants, the pooled risk differences for each specific indication, “while in the direction of risk, were not statistically significant”, they concluded.

Some might argue, they added, that any risk of suicidal ideation/attempts “cannot possibly justify treatment” with antidepressants for these populations. “Instead, we believe that the strength of evidence presented here supports the cautious and well-monitored use of antidepressant medications as one of the first-line treatment options,” the authors said, “with the recognition that efficacy appears greatest for non-OCD anxiety disorders, intermediate for OCD and more modest for MDD.”