The Island of Doubt

ResearchBlogging.orgAnother study purports to find that, for most people, Prozac and the other members of the antidepressant family of pharmaceuticals are no better than sugar pills. Expect Big Pharma to object, but not too loudly. At least, don’t expect them to expend too much effort and money denouncing the findings. We’ve heard this before, and it would seem that neither patients nor the doctors that prescribe antidepressants care much about whether or not the drugs actually do what their makers claim they do.

You can go all the way back to 1998 to find studies casting doubt on the efficacy of antidepressants. Here’s a summary in New Scientist of a meta-analysis published that year:

…the drugs were only 25 per cent more effective. In addition, they suggest that even that 25 per cent could be due to an additional placebo effect derived from the side effects caused by the antidepressants, which alerted patients to the fact that they were receiving an active drug rather than a placebo

The new report, another meta-analysis of previous studies, including those that were never published, concludes that only in the most severely depressed patients did the drugs actually improve mood, and even then we’re not talking about much of an improvement.

Drug-placebo differences increased as a function of initial severity, rising from virtually no difference at moderate levels of initial depression to a relatively small difference for patients with very severe depression, reaching conventional criteria for clinical significance only for patients at the upper end of the very severely depressed category.

Drug-placebo differences in antidepressant efficacy increase as a function of baseline severity, but are relatively small even for severely depressed patients. The relationship between initial severity and antidepressant efficacy is attributable to decreased responsiveness to placebo among very severely depressed patients, rather than to increased responsiveness to medication.

One might argue that both the study referred to in New Scientist, which appeared in Prevention & Treatment, and the new one, published this week in PLoS Medicine, were led by the same researcher, one Irving Kirsch of the University of Hull in the U.K.

But he’s not the only one coming up with dispiriting results. His new paper is co-authored by half a dozen American and Canadian scientists. And, more importantly, “the authors received no specific funding for this study.” In fact, Kirsch has worked as a consultant for Squibb and Pfizer, two of the world’s leading manufacturers of drugs, including those about which his team is casting such depressing aspersions. Squibb, for example, makes Serzone (nefazodone).

Furthermore, the new study takes care of publication bias, which tends to favor studies that find a positive outcome for a drug. We all know what happens to most studies with a negative conclusion. This has impressed news editors. One day after publication, the study is all over the science sections. The Sydney Morning Herald‘s lead, for example is “Millions of people taking commonly prescribed antidepressants could be wasting their time and money, research suggests.”

But will the drug makers (or their shareholders) lose much sleep over this apparent blow to their big money-makers? Probably not. First of all, it’s hard to underestimate the power of the placebo effect. When people feel like a pill is changing their metabolism or physiology ;;;;; because it actually is ;;;;; they don’t much care that the effect has nothing to do with the selective serotonic uptake that the drug is supposed to inhibit. And so they feel better.

Many doctors know this and despite their skepticism prescribe Prozac or Seroxat for the same reason. Placebos can work. And it’s not only doctors dealing with moody patients. Sports medicine has figured this out. In a recent study in The Journal of Neuroscience (27(44):11934-11939), researchers found that

After repeated administrations of morphine in the precompetition training phase, its replacement with a placebo on the day of competition induced an opioid-mediated increase of pain endurance and physical performance, although no illegal drug was administered.

There are ethical questions in both situations. But for me the most interesting question is, how much effort should be put into spreading the word that Prozac et al, aren’t any better than placebos?

It’s important because many of us here at ScienceBlogs like to point out that most “alternative” medicines are basically placebos masquerading as the genuine article. Lots of traditional Chinese medicines fall into that category, and yet literally billions of people turn first to powdered seahorse or tiger bones because that’s what they know best. So in the interests of avoiding hypocrisy, we in the Enlightened West should be just as willing to draw attention to the failings of conventional drugs, no?

In fact, the only difference I see between bogus traditional Chinese medicine and many products of Big Pharma, is the former are often made from endangered species ;;;;; the more threatened the better. By contrast, the only “harm” associated with prescribing conventional placebos may be restricted to our collective trust in Big Pharma. Although that’s hardly a bad thing.



  1. #1 Scott Simmons
    February 26, 2008

    One does find this a bit shocking … I participated myself in a double-blind trial of Luvox for depression. The physicians running the study prescribed it for me after the study was over, and the improvement I’d experienced during underwent a sharp decline. When I went back, I found that they’d prescribed the lesser of the possible doses I could have been on during the study period, as they didn’t know what I’d actually been taking. When they upped the dose to the maximum I could have been taking, I again began to recover. It sure didn’t seem like a placebo effect to me!

    Of course, there were significant side effects, so I could have been aware subconciously of whether I was taking the drug, and how much. (The side effects were severe enough that I eventually quit taking it …) And of course, anecdotes do not equal evidence. :)

  2. #2 Abie
    February 26, 2008

    These studies raise another (possibly naive) question for me: is there a way to harness the placebo effect or the mechanisms behind it?

  3. #3 Dunc
    February 26, 2008

    Well, as someone with many friends who have been prescribed SSRIs, all of whom have experienced either unpleasant withdrawal symptoms or side-effects (yeah, I know, anecdote != data), I’d kinda think you might want to go with a placebo that doesn’t have either.

  4. #4 Nestor
    February 26, 2008

    The problem with this meta-analysis (as with all meta analyses), is that they are reporting average effects as pooled from all studies. It says nothing about responders vs. non-responders (although they may have reported this data but the finding is not being picked up by the media – and I’m too busy [lazy] to print and read the original article). So, putting the entire placebo issue aside, and assuming we are seeing true effects of the SSRIs, the real question in what % of people with mild depression experience a significant improvement in symptoms? People have reported from 10-30%. pretty low! BUT, the real question is what predict such positive response? 10% positive response rate in a debilitating condition affecting millions of people is pretty encouraging, specially if we can better identify who is likely to be in such 10% group and continue to research possible solutions of the other 90%. Hey, I’m not defending big pharma, but I’m afraid of how the media may jump to the conclusion that “meds are not effective”.

  5. #5 Slug
    February 26, 2008

    At this point, thanks to government-mandated redistribution of pharmaceutical patent rights following an arbitrary period in addition to semi-socialist managed care system, you can buy the majority of the anti-depressants for less than the cost of the rival Chinese medicines because they’ve gone generic.

  6. #6 Scott H
    February 26, 2008

    Nestor’s comment about responders/non-responders is interesting, just because my experience is very similar to Scott Simmons’s. No change with the first prescription; slightly increased dosage; two weeks’ delay before recovery became noticeable. That just doesn’t seem like the sort of pattern I would expect from a placebo effect.