Healthy Skepticism International News

January 2008

Antidepressants in children

David Menkes, Peter Mansfield, Jon Jureidini

The meta-analysis reported by Bridge [1] concludes that likely benefits outweigh harms for antidepressant treatment of 3 groups of children and adolescents. In so doing, this study challenges the FDA’s ‘black box’ warnings issued in 2004 regarding suicidality induction by antidepressants in this age group (http://www.fda.gov/fdac/features/2005/105_kids.html).  The analysis points to a favorable benefit/harm ratio based on 60% response to drugs (50% to placebo) and a 1 in 100 excess rate of suicidal ideation/attempts, but some methodological considerations question the authors’ interpretation of these data. 

First, the outcome of individual studies is not adeqately addressed. As Mansfield’s analysis [2] shows, no more than one of the 20 trials on antidepressants in children and adolescents shows benefit on protocol defined outcome measures.

Second, in weighing benefits and harms the authors consider responder status and suicidal ideation/suicide attempts as opposite but equal. Since there were more of the former than the latter they conclude there is more benefit than harm. However patients only had to improve a few points more than in the placebo group to cross the line to be classified as a responder (i.e., a small difference). It is likely that the deterioration to get to suicidal ideation/suicide attempts is a larger difference which may offset or overwhelm the benefit. To summarize: for every 10 adolescents treated, one will slightly improve; for every 11 that slightly improve, one will become suicidal.

Third, the clinical logic of enhancing and exploiting the 1 in 2 placebo response (NNT = 2), and proceeding to drugs only if necessary, is ignored altogether.  This is important in light of evidence that trials have exaggerated antidepressant drug treatment benefits in this age group [3].

Fourth, suicidality is the only harm considered (no mention of common adverse effects, including hostility or drug withdrawal, nor of other potentially serious effects on growth or sexuality). 

Fifth, suicidality is considered per individual rather than per act - some children on SSRIs had 2-3 acts, generally much more serious than those on placebo, but were still counted the same as 1 in this analysis.

Sixth, the authors report that there were no suicides in the trials analysed. Because many patients dropped out or were lost to follow up, it would be more accurate to say ‘none recorded’. Overall in the FDA antidepressant trials database there are at least 5, possibly 6, pediatric suicides. In addition, a recent large case-control study found a significant excess of suicide attempts and deaths in children and adolescents treated with antidepressant drugs [4].  Similarly, re-analysis of trials considered by the FDA confirmed the conclusion that these drugs cause a doubling of suicidality in pediatric populations.

Bridge et al have shown that antidepressant treatment can be helpful in adolescents, particularly those with anxiety disorders. However, the data regarding safety are less reassuring than portrayed in Bridge et al’s conclusion. A particular concern is the rapid and uncritical dissemination of the study’s conclusions by abstracting services such as Journal Watch, which summarized this study for its general medical, pediatric, and psychiatric readers (eg, http://general-medicine.jwatch.org/cgi/content/full/2007/501/1)

View/Hide References

HS Int News index

Page views since 15 March 2010: 6015

Comments

Our members can see and make comments on this page.