Flaws, including highly restricted sample are increasingly recognized.

http://www.psychologytoday.com/blog/the-skeptical-sleuth/201202/cbs-60-minutes-news-treating-depression-sorting-through-the-confu-0

Problem with thsi and many modern meta analyses is that the population entering studies is less sick. This is quite apparent in Anxiety disorders. There is a move now to do a sensitivity analysis for this.

And antidperssants not being that good with mild depression was already well known.

Thanks James and Christopher for your answers. I really liked the Psychology Today piece, which gives a lot of critical additional information.

Christopher, please pardon my ignorance, what is a "sensitivity analysis" in this context and how would it be performed ? Thanks for your insight on the less sick patients, this becomes clear from looking at the publication figure, only very few patients with more severe depression.

Discussions of antidepressants versus pill placebo have gotten emotionally charged on the web and in the scientific literature, some like to use the Fournier et al. data and particularly the data from Kirsch to argue that differences between antidepressants and pill placebo are trivial.

In the interest of being evenhanded, my colleagues and I conducted a systematic review of the literature and edified about 10 or so comparisons between psychotherapy in a pill placebo. These usually occurred in clinical trials with a pill placebo was included to evaluate an antidepressant, not necessarily the psychotherapy. Our review is now accepted contingent on minor revisions, but I can't say too much except that if one applies the same standards to psychotherapy that are being applied to antidepressants versus pill placebo, and one accepts that effect size of .30 is trivial, then psychotherapy would have to be judge trivial in its effect. I don't think anyone is willing to make that point, but those who are arguing to the effects of antidepressants are trivial, need to be inconsistent in order to avoid making this point.

I will be glad to share our paper when we are free to make it available.

A further interesting comment is the one by Bech 2010:

http://www.ncbi.nlm.nih.gov/pubmed/19490744

So, basically, the HAMD17 isn't a good scale to measure antidepressant effect, and this is a fact completely overlooked by Kirsch et al. Intriguing that effect sizes are much larger on the HAMD6, which basically measures the core depressive symptoms (and is a homogenous scale fulfilling item response criteria).

So, besides having a problem with less sick patients in the trials we also have a depression rating scale problem.

The antidepressant drug discovery process has been based on a logical flaw since its inception so it is no surprise that drugs developed using it are little better than placebo - the tragedy is that it has taken us over 60 years to realise this

http://sageinsight.wordpress.com/2012/12/24/the-unsuccessful-antidepressant-drug-discovery-process-indicates-a-catastrophic-systems-failure/

To this welcome discussion I can contribute with a 2011 letter with 46 open access references I had written in 2010 on the topic to the Swiss Medical Bulletin in response to an “Kirschian” editorial by co-editor prof. Stalder *)

Beforehand the “Budapest fallacy” has to be mentioned as fundamental for this topic. We blindly use to subtract the “placebo – responders” from the verum- group, although about half of these would have responded to “verum” as well.

See:

1. Rihmer, Z., et al., Psychiatry should not become hostage to placebo: An alternative interpretation of antidepressant–placebo.... European Neuropsychopharmacology (2012), http://dx.doi.org/10.1016/j.euroneuro.2012.03.002

2. Rhimer Z. Is drug-placebo difference in short-term antidepressant drug trials on unipolar major depression much greater than previously believed? J Affect Disord. 2008;108:195-8.

1. Placebo drugs 1,2,3,4 for long have been the only remedy 5 and we understand only the analgesic ones. 6 Is it therefore „depression“, when the „I will please“ will not work?

7,8 This review argues: this is rather not the question!

2. Specialists dislike antidepressants , also because 40% of depressed are from the bipolar spectrum 9 - and often do respond,

10 but „differently“. 11,12,13 Among these we find up to 1/3 14,15„mixed states“

16 With many choleric-action-prone or sanguinc thought-absorbed clients. The underpinnings of this include different ratios between 17 more transmitter- (hence antidepressant-) or preponderantly ion-channel (hence mood-stabilizer-) responsive “systems”. These in my opinion also seem to steer respectively “internal motivators” and external “motion”. 18 Without a similar bio-psycho-social framework, who’s initial coordinates can be at least at times “read” also from the early antidepressant response

19 antidepressants trigger problematic „mixed state“ holdups. 20 Antidepressants in any case are useless neither in primary care 21 Nor among 2/3 of the STAR*D study 22, nor among the very ill 23,24 , nor in preclinical models and research 25,26 . The at times rather curious 27,28 arguments published by Kirsch

29 instead are questionable.30,31

3. The observer-scored „objective“ placebo-effect-sizes meta-analyzed through 96

studies amounted to 1,85 (having doubled from 1980 through 2005) – „subjectively“ though only to 0,67.

Somehow or other 2/3 of the effect was attributed to „placebo“ – an effect which moreover had predicted half of the effect of the verum drug (independently of severity).

32 Are the changed scores on the interesting 33 Hamilton-depression-scale „HAM-D“

just mirroring a fact, that an increasing number of healers “like to be pleased” while checking the items with a pleasing client?

No, since following Jules Angst these are just means of groups, which enounce little on the topic, and just nothing on the surplus of responders, as Kirsch himself 34 had understood.

4. HAM-D-raw data prove 35,36 that all antidepressants can trigger an extremely uniform normally distributed retrogression which in average (!) manifests 13 days after onset of medication, which is identical to the one occurring in the placebo-group: but in fewer individuals. Melander and other 37 leading experts of four European „FDAs“ co-assembled privately all the 57 approval-RCTs (1984-2003) of SRI and SNRI antidepressants to determine: how many more responded under medication? The answer was 16% - without a bias from study size “n”.38 As commonly accepted, the reciprocal of the variance of “n” acted as a weight.

Kirsch only had accessed comprehensive data from 35 of 47 studies.Severity, as had shown up beforehand 39, and contrary to former conjectures 40,41 apparently had shown no influence whatsoever!

5. This additional “One in six efficiency“ is relevant, since it hides many stronger item-signals, 42 and also because at times quite “antidepressively active” placebos with atropine 43 had been used 44 , and last but not least since „augmentation“ e.g. with (mood-stabilizing) lithium can be used.

6. Kirsch had lost his 2005-debate about fulgurously active drugs purportedly causing undeserved increases in the responders’ box through minute improvements.

Short of one study serving as a “light bearing” of his new linear regression, already 2008 all studies by themselves, as he wrote himself in 2005 45 , showed no relation of responsivity to severity. Furthermore “HAM-D=23” in no way corresponds to „most extremely depressed.“ On top of this he evaluated the overall-HAM-D-average with the arbitrary NICE-tresholds: „>3“ or „Moderate effect-size≥ 0,5“ respectively, requiring more, than his analysis with “1,8” and “0,32” respectively had yielded. How often ideological drill and „expertise“ 46 offer a wonderful placebo against the pain of patients!

REFERENCES: as an attached Word-document

*) Stalder H. Antidepressiva, Placebo und Alternativmedizin. Schweiz Ärztezeitung. 2009;90(41):1600. Reply: http://www.saez.ch/docs/saez/archiv/de/2010/2010-01/2010-01-Leserbriefe.PDF

The references to the above contribution are mostly interactive and to be found in the attached word-doc.

Many psychologists and psychiatrists have now published for years that Antidepressants are "not a viable scientific treatment for depression". In fact, the top psychiatrists on the FDA science committee issued a letter several years about recounting that modern physicians know that an antidepressant is not an effective stand alone "treatment" for depression. In actuality, antidepressants are a "technique" in a more comprehensive treatment plan for depression. We have allowed the less scientific providers, pharm houses, and marketing firms to create the "myth" that "antidepressants are a treatment for depression". They are important, just like other techniques, when used properly and by someone who has completed an "accurate dx" (not just a description of a feeling), a "comprehensive tx plan that is change focused", and family psychoeducation and psychotherapy for the family when needed. This is the scientific, and now widely recognized position, and providers advertise that they may not be the best choice for the patient when they can't do a more depth understanding of the research and discussion of "treatment"!

The concept of "Psychotherapy", "learning", and "potential for change" being trivial, forwarded as an interpretation of studies set up not to evaluate psychotherapy, but to evaluate antidepressants is very faulty logic.  Clearly, neuroplasticity (brain change with learning and experience) is so well established in the literature that we "fund grade schools, high-schools, colleagues, and votec schools all over the world"!  It is very difficult to bring sense to many science communities because of the training we scientists get to "get lost in the small details" and have difficulty with integrative work.

Let's be realistic.  Even Boyce and Hart proved this when they advertised for the most untalented and musically uneducated folks they could find in Variety mag, and later trained them into multiple gold record sales band and cutely named them "The Monkeys" to prove their point about neuroplasticity and the massive ability of people to learn and the brain to change!

Jer