the Age of the Decepticons…

Posted on Thursday 18 June 2015

The Treatment for Adolescents with Depression Study [TADS] was an NIMH funded clinical trial including 439 adolescents that ran from September 1998 though March 2004. The study compared Placebo [PBO], Prozac [FLX], Cognitive Behavior Therapy [CBT], and Prozac+Cognitive Behavior Therapy [COMBO] for 12 weeks in a blinded RCT. But that’s not all folks.  They then broke the blind and followed the subjects for another 24 weeks. PBO non-responders could then choose FLX, CBT, or COMBO:
by March JS, Silva S, Petrycki S, Curry J, Wells K, Fairbank J, Burns B, Domino M, McNulty S, Vitiello B, and Severe J
Archives of General Psychiatry. 2007 64[10]:1132-1143.

CONCLUSIONS: In adolescents with moderate to severe depression, treatment with fluoxetine alone or in combination with CBT accelerates the response. Adding CBT to medication enhances the safety of medication. Taking benefits and harms into account, combined treatment appears superior to either monotherapy as a treatment for major depression in adolescents.
by Benedetto Vitiello, Susan Silva, Paul Rohde, Christopher Kratochvil, Betsy Kennard, Mark Reinecke, Taryn Mayes, Kelly Posner, Diane E. May, and John S. March.
Journal of Clinical Psychiatry. 2009 70[5]: 741–747.

RESULTS: Forty-four patients [10.0%] had at least 1 suicidal event [no suicide occurred]. Events occurred 0.4 to 31.1 weeks [mean +/- SD = 11.9 +/- 8.2] after starting TADS treatment, with no difference in event timing for patients receiving medication versus those not receiving medication. Severity of self-rated pretreatment suicidal ideation [Suicidal Ideation Questionnaire adapted for adolescents score > or = 31] and depressive symptoms [Reynolds Adolescent Depression Scale score > or = 91] predicted occurrence of suicidal events during treatment [P < .05]. Patients with suicidal events were on average still moderately ill prior to the event [mean +/- SD Clinical Global Impressions-Severity of Illness scale score = 4.0 +/- 1.3] and only minimally improved [mean +/- SD Clinical Global Impressions-Improvement scale score = 3.2 +/- 1.1]. Events were not preceded by increased irritability, akathisia, sleep disturbance, or manic signs. Specific interpersonal stressors were identified in 73% of cases [N = 44]. Of the events, 55% [N = 24] resulted in overnight hospitalization.
CONCLUSIONS: Most suicidal events occurred in the context of persistent depression and insufficient improvement without evidence of medication-induced behavioral activation as a precursor. Severity of self-rated suicidal ideation and depressive symptoms predicted emergence of suicidality during treatment. Risk for suicidal events did not decrease after the first month of treatment, suggesting the need for careful clinical monitoring for several months after starting treatment.
Among the bizarre misrepresentations in Clinical Trials of psychiatric drugs during the Age of the Decepticons, this one may take the grand prize. The full story is covered in significant I… and significant II…, but here’s the punchline:

« truncated » · « recolored » · « click for original »

This is figure 1 from that last paper. A more truthful version of the legend would read:

As much as I appreciate someone finally mentioning psychological factors ["Most suicidal events occurred in the context of persistent depression and insufficient improvement without evidence of medication-induced behavioral activation as a precursor"] as even existing, they’re just trying to dissuade us from looking at figure 1 and seeing what it actually says: Prozac caused suicidality in a significant portion of adolescents in the NIMH funded TADS trial. No statistical tests needed to see that. And what person with suicidal thinking or attempts ever didn’t have an explanation?

This disconnect between the narrative and the data came to my attention from the posts by David Healy [Coincidence a fine thing] and Robert Whitaker [The Real Suicide Data from the TADS Study Comes to Light]. I ran across them vetting an absurd paper by statistician, Robert Gibbons [Suicidal Thoughts and Behavior With Antidepressant TreatmentReanalysis of the Randomized Placebo-Controlled Studies of Fluoxetine and Venlafaxine], in one of his many attacks on the Black Box Warnings [a warning that the TADS trial actually supports]. Both Healy and Whitaker [and I] attribute the discovery of this damning chart amid the din of TADS papers to Swedish Investigator Göran Högberg [his comments from our 2012 email conversation are in my significant I… post].

So finally, Högberg‘s finding makes it into the medical literature [years after the patent ran out]:
by Högberg, Göran | Antonuccio, David O. | Healy, David.
International Journal of Risk & Safety in Medicine. 2015 27[2]:85-91.

Completed suicides are a major cause of death in adolescents in Sweden. Forensic analysis of completed suicides in children and adolescents shows there is one completed suicide per 1000 children taking a selective serotonin re-uptake inhibitor [SSRI]. In order to elucidate these events we undertook a study of the results and reporting of suicidal events in the Treatment of Adolescents with Depression Study [TADS]. We conclude that a major, albeit underreported, finding in the TADS was the significant increase of suicidal events in the adolescents on antidepressant medication in comparison to the group on placebo medication. The proportions of suicidal events were 11% and 2.7% respectively. This increased risk of suicidal events might be related to the high incidence of medication with an SSRI in the group of completed suicides among Swedish adolescents:
« from Högberg et al’s version » · « recolored »
DecepticonIt’s years late, and it’s softly presented, but it’s still something of a breakthrough. It’s what we’re hoping for from Data Transparency, but they didn’t have to petition for the data or jump through any hoops as it was just hiding there in plain sight. And it’s a republication of a deceptively presented NIMH funded study, setting the record straight. It may seem frivolous to name these studies that distort the Clinical Trial findings after characters from the SciFi Transformer comics and movies, but the term Decepticons just fits them to a tee [deceptive icons]. The TADS trial produced some twenty publications, and is widely quoted recommending SSRI treatment for depressed adolescents. It’s also used in most arguments against the Black Box Warning, as in the paper quoted above. Perhaps NIMH Director Insel might grace us with a blog on Högberg’s paper?
    June 19, 2015 | 1:18 PM

    Something I do not understand, maybe someone can explain it. Why are there 2 red bubbles in the placebo group before the 12 week marker? How are those ideation on SSRI if the initial study length was 12 weeks?

    June 19, 2015 | 1:22 PM

    I haven’t figured that out either. It’s on every version, and so far I can’t find anything in the text that explains it. I’m going to send that question to the authors.

    June 19, 2015 | 2:24 PM

    An eagle-eyed reader found it in Suicidal Events in the Treatment for Adolescents with Depression Study [TADS].

    Some patients who had been randomized to CBT (N=2) or PBO (N=9) and had a suicidal event were in fact on SSRI edication at the time for the event having started antidepressant treatment because of non-response to the randomly assigned treatment. For these patients, tune to event was recomputed by taking the tune when SSRI medication was initiated as the starting point. Using this approach, for all the patients on SSRI at time of the event (N=36). the time to event was 10.0 ± 7.7 weeks (range 0.7—28.9), which was not statistically different from that for the patients not on antidepressant medication at the time of the event (N=8: 5.8 ±3.2 weeks, range: 0.4-10.4; Wilcoxon two-sample test, z=-1.2[ p=0.22).

    June 19, 2015 | 2:39 PM

    I get it now. So aside from being misleading in its original labeling the bottom of the chart also represents two completely different timelines, depending on what group the patient is in. This is like a study in how to design an incomprehensible chart. If you wanted to figure out at what point after initiating treatments someone had a suicidal event, you could do that with a separate chart. This chart seems to be trying to do two things at once and completely messing up both of them as a result. Or maybe the purpose of making a chart like this was to deceive anyway.

    June 19, 2015 | 6:18 PM

    In the Suicidal Events TADS study, it could be researchers were looking for obvious “evidence of medication-induced behavioral activation.”

    See Hirose, S. (2003). The causes of underdiagnosing akathisia. Schizophrenia bulletin, 29(3), 547-558. Free pdf at

    From what I’ve seen, added to those factors identified by Hirose should be that even adult patients do not have the vocabulary to accurately convey how they feel. The internal sensations of akathisia are not easily described.

    June 19, 2015 | 6:43 PM


    Without the raw data, there’s really no way to fix it. I agree that in this case, two charts are better than one.

    June 22, 2015 | 2:56 PM

    Thanks for pointing out these important “inconsistencies.” As someone who reviews studies for journal publication, I’ve had to get better at spotting data manipulations and clever attempts to emphasize or de-emphasize certain things. These can appear not only on pharma studies but also from people trying to support new types of therapy. Thanks to the previous commentators who figured out how some patients got SSRI when randomized to placebo. That one vexed me too.

    However, I’m not sure this chart is as monumental as described. As I understand the study, the blind was broken after 12 weeks and patients were allowed to switch treatments. This may reflect the study authors’ own bias, but it sounds like the patients whose depression was getting worse on placebo or CBT were started on an SSRI. Putting all the causality here on the medication, then, is a bit dubious. That said, there was better evidence for more suicidality related to SSRI use during the acute 12 week period but, as was mentioned in the post, this was acknowledged in the original TADS publication.

    Another interesting phenomenon I’ll point out, is how certain folks seem so convinced SSRIs can have this huge powerful negative effect on the brain while discounting any chance that for some people it really helps. Can’t it be that some are helped, some harmed, and some neither one? Is this not the case for almost all prescription medications?

    June 22, 2015 | 3:22 PM

    Another interesting phenomenon I’ll point out, is how certain folks seem so convinced SSRIs can have this huge powerful negative effect on the brain while discounting any chance that for some people it really helps. Can’t it be that some are helped, some harmed, and some neither one? Is this not the case for almost all prescription medications?

    I completely agree…

Sorry, the comment form is closed at this time.