News blog

Antidepressants ‘no better than placebo’

In all but the most severely depressed patients, there is no evidence that new-generation antidepressants work any better than placebos. So says a large metanalysis that combed through 47 published and unpublished studies on several Selective Serotonin Re-uptake Inhibitors (SSRIs), published in PLoS Medicine yesterday. This class of drugs includes Prozac.

The bold headline ‘Antidepressants don’t work’ obviously caught the attention of the press (Google news search yields at least 250 hits today; almost all from the UK) and the public, and for good reason. It’s important to keep in mind that professionals already knew these drugs don’t have a huge effect for many (more below). But the extent to which the drugs were found to be effective as a placebo is striking. They did seem to work in extreme cases, but not for the reason one would expect: “The main finding among severely depressed people was that they ceased to respond to a dummy placebo pill, not that they showed a particularly heightened response to anti-depressants." (BBC)

Current UK National Institute for Health and Clinical Excellence (NICE) guidelines already advise non-medicinal treatments for mild depression, including exercise and sleep management or psychological interventions. “Antidepressants are not recommended for the initial treatment of mild depression, because the risk–benefit ratio is poor,” they state. “Where mild depression persists after other interventions, or is associated with psychosocial and medical problems, consider use of an antidepressant.”

As Anne Robinson tells us in a Guardian post, there is “no need to panic”. “People who are already on antidepressants and getting better should stay on them and then tail them off gradually when they feel ready. Those who are considering taking them will want to think twice. But none of that’s new and none of it is reason to panic.” This ‘not new’ statement sits rather ironically aside the Guardian’s choice for front page.

It’s with moderate depression that the case gets interesting. NICE guidelines say: “In moderate depression, offer antidepressant medication to all patients routinely, before psychological interventions.” It is unclear whether that should now be changed. Lead study author Irving Kirsch from the department of psychology at Hull University, who was one of the consultants for these guidelines, says the new analysis suggests prescriptions “might be restricted even more” (Guardian).

(see UPDATE below the fold)

The news illuminates two worrying trends: the vast over-prescription of drugs such as antidepressants, and the vast quantity of unpublished data on drugs that often never sees limelight.

The Telegraph’s Q&A tells us the number of prescriptions written for anti-depressants has tripled in 15 years; perhaps more because of over-diagnosis and over-reliance on drugs than because of any upswing in unhappiness.

And as always, the fact that drug companies are not obliged to publish negative results remains a worry. From the Telegraph again: “Drug companies have to submit all their clinical evidence to drugs watchdogs in order to gain a licence but they do not have to publish negative results to the wider medical community.” The researchers in this instance used the Freedom of Information Act to get their hands on more data from the US FDA.

The main comments against this metanalysis are generally from drug companies or industry reps, as here, from the BBC:

Dr Richard Tiner, of the Association of the British Pharmaceutical Industry, said there was no doubt that was a “considerable placebo effect” from anti-depressants when treating people with mild to moderate symptoms. But he said no medicine would get a licence without demonstrating it was better than a placebo. Dr Tiner said: “These medicines have been licensed by a number of regulatory authorities around the world, who looking at all the evidence, have determined that they do work better than placebo.”

To me this simply begs a question: how much better than placebo does a drug need to be, and in how many people, for it to be licensed?

From the archive:

All our previous stories on antidepressants.

UPDATE: The Guardian has gone big with this story, again, today:

Ben Goldacre, ‘bad science’ columnist, hits the issue of drug regulation hard.

Psychoanalysts trace the history of Prozac in this lengthly mental health piece. And, in news, they look at a big UK push to give money money to psychoanalysts (an alternative to drug treatment for depression).


  1. Report this comment

    Dr David E Clarke said:

    It is well established that antidepressants take weeks or months to work (eg: Duman, RS. Neuron 55: 679, 2007). Yet the analysis of antidepressant efficacy conducted by Prof Irving Kirsch and colleagues utilized short term clinical trials of the following durations: 6 trials for eight weeks, 33 trials for six weeks, 2 trials for five weeks, and 6 trials for four weeks (a total of 35 trials). One would expect only threshold responses with marginal statistical significance and some trial failures. Demonstration of long-term efficacy is required for registration in the EU.

  2. Report this comment

    Gregory Weisbrod said:

    “That may be true for you”, goes the old Irish saying. I have been taking one of the SSRI’s for my panic attacks and have since noticed a rather comfortable emotional disengagement from my chronic depression.

    True, the thoughts are still there, but without their gloomy force. In fact I realize I should be quite cheerful, as everything is just peachy.

    Over 100 million USA scripts in ‘06 can’t be wrong.

  3. Report this comment

    Aivar Päären said:

    Better diagnostic is key here – better instruments, differenciated criteria ( for men, adolescents, children …), different approaches etc. Important is to differenciate other related problems that overlap and give the same clinical pictures. When we succeed to find “pure biological depression” have we bigger chance to get good response with SSRI. Study design is much more important than to down the symptoms. Good luck!

  4. Report this comment

    Dave Allsop said:

    This story highlights the larger problem of western world capitalist consumerism infiltrating the very fabric of our souls. This type of medication should be prescribed as the very last resort for only the most severe cases, and even then only after an extensive program of physical activity and cognitiv behavioural therapy has been prescribed and tested on the patient. These companies are making heaps of money on the back of peoples inability to cope in the modern world – it should be the governments responsibility to regulate more firmly, and to put in measures such as the ones I described to demonstrate that they care about their consituents. Perhaps every GP should also be trained in CBT and sports science…they have large brains I am sure they could cope…

  5. Report this comment

    Ricardo K Almeida said:

    After 4 years of psicotherapy to no avail I took an SSRI for 2 months without any effect on depression. Two weeks after interruption I was better than never. This pattern repeated three times, the last one after a 3 month treatment with an increase in drug dosage after the 45th day. I wonder there is a delayed negative placebo effect on literature?

  6. Report this comment

    Dominique Brunet, Ph.D. said:

    “Believing makes all the difference!” should be the title. However, although taking anti-depressants such as prozac ends up with a placebo effect, a placebo effect has a “real” biochemical basis(see Tor D.Wager’study published by Nature in 2007, placebo induces secretion of endorphins and enkephalins in the brain). Therefore, if the patient feels better after taking such anti-depressants, why not? However, humans being as they are, now that it is known a certain category of anti-depressants doesn’t have the physiological effect it is supposed to have, people will not feel better after taking such medication, these drugs being deprived of their psychological effect!

  7. Report this comment

    Dr Jennifer Wild said:

    The NICE guidelines for depression, originally published in 2004 and amended in 2007, say that “for many patients there is little clinically important difference between antidepressants and placebo, and that placebo response is greatest in mild depression” (p.61, Full Guidelines). Kirsch’s study shows the same: the placebo response is greatest for milder forms of depression and minimal for very severely depressed people. NICE go on to recommend antidepressant drugs for moderate to severely depressed people. This is somewhat strange given that the guidelines first state there are no differences between drugs and placebo and then recommend giving drugs to treat depression. Maybe they suggest this because the drugs are cheap? The guidelines also say that “antidepressants are as effective as psychological interventions, widely available, and cost less.” Thankfully the programme for Improving Access to Psychological Therapies has kicked off, which will make psychological interventions more easily available to people who are suffering with depression. In the meantime, NICE need to update their guidelines, make them consistent, and possibly consider recommending placebo treatments – surely they’re the cheapest intervention of all?

  8. Report this comment

    Pharmacy said:

    The term antidepressant is sometimes applied to any therapy (e.g. psychotherapy, electro-convulsive therapy, acupuncture) or process (e.g. sleep disruption, increased light levels, regular exercise) found to improve clinically depressed mood. An inert placebo tends to have a significant antidepressant effect, so establishing something as an antidepressant in a clinical trial involves demonstrating a significant additional effect

  9. Report this comment

    Dental Recall said:

    I think that it depends on which anti depressant you are talking about. Because there are some that are proven to work. However, there is a small mental aspect to the whole idea that a placebo can affect.

Comments are closed.