Hammad · Archives of general psychiatry 2006 · Individual patient data meta-analysis of randomized controlled trials · n=4582 participants (24 trials)

Suicidality in pediatric patients treated with antidepressant drugs.

Cited 1067 times in the scientific literature.

Level 1 - systematic review of randomized trials

Meta-analysis of individual patient data from 24 randomized placebo-controlled trials.

PubMed 16520440 · doi:10.1001/archpsyc.63.3.332 · record verified 2026-08-26

What was done

The authors conducted an individual patient data meta-analysis of 24 randomized, placebo-controlled pediatric trials submitted to the FDA (23 industry-sponsored programs and the TADS trial) involving 4,582 patients. Indications evaluated were major depressive disorder (16 trials), obsessive-compulsive disorder (4 trials), generalized anxiety disorder (2 trials), attention-deficit/hyperactivity disorder (1 trial), and social anxiety disorder (1 trial). Risk ratios (RRs) and risk differences were calculated for individual drugs, SSRIs in depression, and all antidepressants across all indications.

What was found

No completed suicides occurred across any of the trials. Twenty trials contributed to risk ratio analyses because 4 had zero events in both groups. Only one individual trial (TADS) showed a statistically significant risk ratio on its own (RR 4.62; 95% CI, 1.02-20.92). Across all SSRIs in depression trials, the pooled RR was 1.66 (95% CI, 1.02-2.68). Across all antidepressant drugs and indications combined, the pooled RR was 1.95 (95% CI, 1.28-2.98), with an overall risk difference of 0.02 (95% CI, 0.01-0.03).

Why it matters

This FDA meta-analysis established the empirical foundation for black box warnings by quantifying a small but statistically significant absolute excess (about 2%) in suicidal ideation and behavior among youth treated with antidepressants.

Limits

The findings represent short-term trial exposures and cannot assess long-term suicidality or completed suicide mortality, as zero completed suicides occurred across the sample. Four trials could not contribute to relative risk estimates due to zero events in both treatment and placebo arms.

Cited by