Citation

Smith EG. J. Affect. Disord. 2009; 114(1-3): 143-148.

Affiliation

Department of Psychiatry, University of Massachusetts Medical School, 55 Lake Avenue, Worcester, MA 01655, United States. Eric.Smith5@va.gov

Copyright

(Copyright © 2009, Elsevier Publishing)

DOI

10.1016/j.jad.2008.06.018

PMID

18692250

Abstract

BACKGROUND: This study sought to determine from a recent meta-analysis of pediatric antidepressant trials if a general property of antidepressant medications--the multiple-dosing medication half-life--is associated with risks for suicidal ideation or behavior. METHODS: Relative risks for suicidal behavior (ideation, attempt, or preparation) for seven antidepressants were obtained from both the FDA's initial and published versions of their pediatric antidepressant meta-analysis. The correlation between the relative risk for suicidal behavior and antidepressant half-life was examined using a nonparametric test, Spearman's rho. RESULTS: A significant correlation (rho=0.929; p=0.003) was observed for the initial analysis, as previously reported by Weiss and Gorman. The correlation was robust to a change in the suicidality ranking for the longest half-life medication, fluoxetine, that occurred when results from the Treatment of Adolescent Depression Study (TADS) were included in the published meta-analysis (rho=0.786, p=0.036). LIMITATIONS: In addition to limitations common to meta-analyses, our analysis has additional uncertainties including the fact that adult, rather than pediatric, antidepressant half-life data were used due to an unavailability of published information. In addition, risks for suicidal ideation/behavior may vary for reasons other than half-life (e.g. study eligibility criteria, illness severity or responsiveness to treatment, diagnoses, etc.). CONCLUSIONS: The risk of suicidal ideation or behavior in short-term antidepressant trials involving children or adolescents, as defined in the recent FDA meta-analysis, appears to be potentially at least partly associated with antidepressant half-life. Although any relationship is tentative, approaches to investigating several potential candidate mechanisms for any association are discussed.

Language: en