significant II… »
« suspicious mind[s]…
significant I…

Posted on Wednesday 22 February 2012

There has been no recent issue as hotly contested as the question of whether the SSRIs can induce suicidal thinking, particularly in youth. In 2004, the FDA added a black-box warning to the SSRIs, resulting in a fall in prescribing, and a polarized literature on this topic that lives on. Recently, the focus has been on the Treatment for Adolescents with Depression Study [TADS], first published in 2004. I walked in on an email discussion of several recent articles and found myself kind of confused. So first, I thought I’d gather the relevant links for anyone who might want to read along [most articles available on-line]:

A. There’s a full bibliography from TADS on the ClinicalTrials.gov site [first reference]. These articles are the ones I thought were relevant to the discussion of suicidality:
  2. Fluoxetine, Cognitive-Behavioral Therapy, and Their Combination for Adolescents With Depression
    by the Treatment for Adolescents With Depression Study (TADS) Team
    JAMA. 2004 292(7):807-820. [full text on-line]
  3. Treatment for Adolescents with Depression Study (TADS): SafetyResults.
    by Emslie G, Kratochvil C, Vitiello B, Silva S, et al.
    Journal of the Amer Acad of Child and Adolesc Psychiatry. 2006 45(12):1440-55.
  4. The Treatment for Adolescents With Depression Study (TADS)
    Long-term Effectiveness and Safety Outcomes
    by The TADS Team
    Archives of General Psychiatry. 2007 64(10):1132-1143. [full text on-line]
  5. Suicidal Events in the Treatment for Adolescents with Depression Study (TADS)
    by Benedetto Vitiello, Susan Silva, Paul Rohde, Christopher Kratochvil, et al.
    Journal of Clinical Psychiatry. 2009 70(5): 741–747. [full text on-line]
B. There was a recent article by Gibbons reanalyzing TADS that reached some contended conclusions. My two response blogs have links to Gibbons’ previous articles, other takes on the question of SSRI suicidality, and various opinions about the effect of the 2004 black box warning:
  1. Suicidal Thoughts and Behavior With Antidepressant Treatment
    Reanalysis of the Randomized Placebo-Controlled Studies of Fluoxetine and Venlafaxine
    by Robert D. Gibbons, Hendricks Brown, Kwan Hur, John M. Davis, and J. John Mann
    Archives of General Psychiatry. Online February 6, 2012. [full text on-line]
C. Recent commentary from David Healy and Robert Whitaker about TADS and Gibbons’ article:
  1. Coincidence a fine thing
    by Dr. David Healy
    February 8, 2012
  2. The Real Suicide Data from the TADS Study Comes to Light
    by Robert Whitaker
    February 20, 2012
All of this post’s discussion hinges on the reference A.5. above, the Vitiello et al article that has this conclusion in the abstract:
    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.
Whether you’ve read the voluminous literature on the topic about suicidality from SSRIs in adolescents or not, you’re likely aware that feelings run strong on this topic. At one pole are people who contend that the pharmaceutical companies and their supporters have suppressed the suicidality data motivated by a wish to sell drugs. At the other pole are those who say that it’s not true, and that vital treatment is being withheld from adolescents, and suggest this increases suicidal rates. None of the active participants in this debate have changed their minds along the way. So back to article A.5. It has this figure [Figure 1] of the timing of suicidal ideation and suicide attempts from the TADS study [PBO=Placebo, CBT=Cognitive Behavior Therapy, FLX=Prozac, and COMB=CBT+Prozac:


[reformatted for clarity]

This is the same graph, except I’ve removed the "suicidal ideation" cohort, leaving only the "suicidal behavior" [suicide attempts] group:

By count, there were 18 suicide attempts. 17 of them were by subjects on Prozac.  To be fair, Vitiello et al were reporting that they did not see agitation prior to the suicide attempts, implying that drug induced akathisia wasn’t the cause of the suicide attempts. To my reading, this 17/18 suicide attempts being on Prozac wasn’t discussed. The reason that people in all groups were on Prozac after 12 weeks had to do with the study design [see below].
    TADS was a publicly funded, randomized clinical trial to evaluate the acute (12 weeks) and long-term (36 weeks) effectiveness of fluoxetine (FLX), cognitive-behavior therapy (CBT), and their combination (COMB) in the treatment of adolescents with major depressive disorder. A placebo condition was included as a control for the first 12 weeks, after which all treatments were unblinded. Patients who were deemed responders or partial responders, as documented by a score of minimally improved or better on the Clinical Global Impression- Improvement scale (CGI-I), continued with their assigned treatment up to the end of the 36-week trial. Non-responders were discontinued from study treatment and managed as clinically indicated while still participating in the remaining assessments throughout week 36.
That means that after 12 weeks, many [?] of the non-responders in the top two groups were put on Prozac and followed to 36 weeks. For completeness, here’s the data cropped for suicidal ideation only:
It still looks pretty bad for Prozac. 18 of the 26 patients with suicidal ideation [without attempt] were on Prozac. So the totals [suicidal ideation + suicidal behavior] was 36 of 44 suicidal events occurred in subjects on Prozac. So that’s Figure 1 from the paper. It seems clear enough. Now here’s Table 1 from the article:


[reformatted to fit]

Table 1 seems like it must come from another article. Take a look at it and compare it to Figure 1 and see if it makes any sense [it won’t]. See if the footnotes help [they don’t]. That’s part of what the David Healy and Robert Whitaker articles are about. Trying to make some sense of this morass [C.1. and C.2.]. Both authors give credit to Göran Högberg PhD from Stockholm, Sweden who figured this out. In his own words [from an email response]:
    TADS is worth a study on its own. The main, meagre results, are hidden in a mass of texts, take home messages and cutting edges … not any new data, just careful reading of what there is in the study. The result is shocking, because it really means that important risks are hidden. You can do the calculations yourself. Check it.

    TADS was not a 36 week RCT, it was a 12 week RCT. In this article they claim in the abstract that it was a 36 week RCT. By doing that they can miscalculate the percentage with suicidal events. Suicidal events were defined as ideation or sucidal behaviour = added together as suicidal events. In the study after 12 weeks subjects were re-assigned, non-responding CBT and placebo participants were offered fluoxetine. We can clearly see the number of suicidal events – the black boxes and circles in figure 1. But they are not correctly assigned. If we are to give a percentage of subjects on placebo, all the subjects first on placebo and then on fluoxetine should be added to the fluoxetine group, leaving only true placebo. All the subjects who first had CBT and then got fluoxetine should be on the group of combined treatment. Both fig 1 and table 1 thus give false reporting, as percentage is calculated on the grouping of the first 12 weeks, not the full 36 weeks. The percentage of suicidal events in subjects on only fluoxetine, all fluoxetine, only CBT and combined CBT/fluoxetine is shown in my table.

    There were all in all 44 suicidal events over the 36 weeks, 36 of those events on fluoxetine. And in the abstract there is nothing of this mentioned.
Obviously, this data has been massaged. The clear association of suicidality in these adolescents on Prozac [16%] compared to ones who weren’t [4%] has been lost in the clouds and isn’t mentioned in an article titled Suicidal Events in the Treatment for Adolescents with Depression Study (TADS) – an NIMH funded study. I call that significant…
  1.  
    Altostrata
    February 22, 2012 | 12:56 PM
     

    What we have now is psychiatry researchers, having built their careers and, in some cases, fortunes, on flawed research enthusiastically recommending various psychiatric drugs, are in a rear-guard action to defend their legacies.

    Do they even know they’re jimmying the data? Maybe not, their anosognosia being so acute they can’t even do math anymore, or have lost the ability to distinguish categories.

  2.  
    Mickey
    February 22, 2012 | 1:06 PM
     

    a·no·sog·no·si·a       -nsg-nz–, -zh
       noun
       Unawareness of one’s disease, disability, or a defect.

    My new favorite word!

  3.  
    Altostrata
    February 22, 2012 | 1:10 PM
     

    You’ll love this NY Times article series by Errol Morris:

    The Anosognosic’s Dilemma: Something’s Wrong but You’ll Never Know What It Is
    http://opinionator.blogs.nytimes.com/2010/06/20/the-anosognosics-dilemma-1/

  4.  
    Bernard Carroll
    February 22, 2012 | 1:49 PM
     

    Your term ‘massaged’ and Whitaker’s term ‘hiding in plain sight’ are to the point here. The authors of the Vitiello paper have plausible deniability – after all they did report the data in the Figure. But they make the reader work to see the data clearly. That’s usually called obfuscation. Notice, for instance, how they do not forthrightly label the Figure as ‘suicide attempts on fluoxetine’ and ‘suicide attempts not on fluoxetine.’ Plus, fluoxetine in the Figure legend becomes the pastel, generic term SSRI. Did they learn this stuff from Madison Avenue?

    The numbers speak for themselves: the Odds Ratio for suicide attempt being associated with current fluoxetine intake rather than no drug is 17. The rate ratio is 15 (7.5% versus 0.5%). The probability is 0.0005. Suppose the data had come out showing the reverse, that the drug has a huge preventive action against suicide attempts – do you think that finding would have been front and center in the publication? Duh.

    There may be clinical reasons for this strong association of suicide attempt with on-drug condition in the TADS project. For instance, quite a few attempts occurred in patients who were switched to drug because of non-improvement with their earlier treatment. So the on-drug cases from week 12 to week 36 may have been more at risk. But looking into that is a second stage of discussion… first, the big effect has to be acknowledged, but it was obfuscated by Vitiello and co-authors.

    As an aside, I am constantly amused to observe that physicians have no problem with the idea that some drugs that have a generally beneficial action on the brain can have unpredictable side effects such as the motor dyscontrol (dyskinesias) caused by l-DOPA in Parkinson disease, but the same people go postal when it is suggested that something similar might happen in the form of emotional and behavioral dyscontrol (agitation, akathisia, and suicide attempts) in some patients with SSRI drugs.

  5.  
    stephany
    February 22, 2012 | 4:19 PM
     

    The connection of no meds ever vs on Zoloft was suicidal reaction in plain sight, when it happened to my child in 2001. All of the doctors saw it. Too bad their KOLs and data was based on lies and deceipt, otherwise they may have known not to give a teen (13!) Zoloft, or even though history was charted, the PROZAC at age 17. Always the same reaction, and thankfully my child did not kill herself on the drugs, just hanging out a car door on the freeway and other dreadful drug-induced, never repeated off the drugs events and behaviors.

    I wrote my own abstract in 2001 showing the connection of SSRI use and increased suicidality, it’s in my boxes with the other momentos of those horrific days where doctors said they knew what they were doing.

    Thank you, quacks!

Sorry, the comment form is closed at this time.