Comments: Lexapro For Teen Depression Studies: Not Much Efficacy
Great investigative work to get all the studies involved!
Posted by Sara at March 24, 2009 12:16 PM
Thanks. Philip! Now we know that given the choice between Lexapro and a glass of water, the water would have the same effect - and, unless it contains residual pharmaceuticals, the water might well be safer.
Posted by Lilly NC at March 24, 2009 01:03 PM
effect size: if the standard error reported in the poster is equal to [the standard deviation, divided by (the square root of the sample number)], then a standard deviation is roughly 8.2. I looked up mean and std dev really quickly on this CDRS. Another study had Std dev of 7.8, roughly equal to 8. So, a std dev of 8 at basline sounds like a decent numner to work with here.
Effect size: a common effect size is change score divided by std deviation. So, for this study, esc kids had about 22 point change. 22/8 = 2.75.
The average kid in this study changed 2 and a half standard deviations on this depression scale?
That just does not add up.
If the study recruited kids with very high depression scores (possible), then they may have had a lot of regression to the mean operating. can't say without digging into CDRS more to know typical standard deviations. But effect size of 2.75 is almost unheard-of in depression treatment.
Posted by MedsVsTherapy at March 24, 2009 01:24 PM
As a matter of fact, both studies you mentioned as failed can be interpreted as positive, and one of them probably was interpreted as positive in the Forest's FDA submission. My guess is that the European study (Journal of Clinical Psychopharmacology) was not formally in the FDA submission.
JAACAP abstract notes: "In a post hoc analysis of adolescent (ages 12-17 years) completers, escitalopram significantly improved CDRS-R scores compared with placebo (least squares mean difference = -4.6, p = .047)."
JCP study abstract also had a tentatively positive result in the pertinent sub-group: "For those patients not receiving psychotherapy, there was a higher percentage of Kiddie-SADS-P responders with citalopram (41%) versus placebo (25%) and a significantly higher percentage of MADRS responders and remitters with citalopram (52% and 45%, respectively) versus placebo (22% and 19%, respectively)."
You can also glean some details of the unpublished studies from the Forest database of clinical trials at http://www.forestclinicaltrials.com/CTR/CTRController/CTRCompletedListStudies. Look at studies SCT-MD-15, SCT-MD-32 and SCT-MD-32a.
Posted by TSC at April 5, 2009 06:55 PM
RE: TSC
The two studies' authors clearly indicate failure in the tests they designed. But for some reason, you are trying to twist the results by highlighting certain incidental data that was touched upon in the study (JAAPAC's post hoc analysis; JCP's analysis of patients not receiving psychotherapy which the study clearly states was less than a third of the initial study group).
You then point readers to unpublished studies from Forest Labs- whom, you fail to mention, manufactures the drug.
Adding to this your basically anonymous screenname, you'll forgive me for having difficulty seeing your post having been written from an neutral point of view.
Posted by Alan K. at September 7, 2009 06:02 PM