To err is human… David Colquhoun is no super hero!

Like it or not, I’m at the same time an osteopath, an engineer, a father of four, a football fan interested in philosophy and social sciences, an amateur musician, a fiction reader and a science reader, and as my twitter profile states, I’m also sometimes a researcher and sometimes a reviewer.

The reason I write this post is because I also enjoy reading skeptics, including David Colquhoun (DC), of course.

Before going further I have to make clear that, even if I’m going to be critical with some of the skeptics’ works (which is nothing more than an incentive to extend their work), I consider Colquhoun, Singh, Goldacre and others’ contributions essential in the complex and chaotic progression of human knowledge. Knowing one part of the Complementary and Alternative Medicine (CAM) world from the inside – probably not the worst part – I have to confess that what these authors say about CAM is most of the time relevant (I’ll probably develop some other time the slight but nevertheless important difference between ‘most of the time’ and ‘always’, ‘healthy skepticism’ and ’skeptic fundamentalism’). If I ever come to criticize some aspects of skeptics’ critics, that doesn’t mean I’m standing up for any form of quackery. There is no excuse for trickery, people’s awareness should be raised and any form of abuse needs to be exposed.

This blog is neither intended in defending CAM and, in order to finish disillusioning the dogmatics that may come across my posts, I must say that my children my wife and I are all vaccinated. (Usually that’s as much an effective argument to drive back dogmatics as garlic to repel vampires)

Finally, I’m not a native english speaking person, which may explain possible oddities in the way I combine english words and idioms.

If scientists have generally less prejudices than the other men, they are nevertheless more keen to stick to the ones they have

Jean-Jacques Rousseau,  les Confessions,  VII

It is hard to disagree with evidence and, on the whole, I used to agree with Colquhoun’s views until this recent post (Placebo effects are weak) about the placebo effect and regression to the mean. In a few words, DC recently (re)discovered the power of what Sir Francis Galton called regression towards the mean (I shall discuss later how he could have re-discovered it earlier). DC then combined it with a systematic review showing the powerlessness of the placebo effect to declare a win by default over all the transient snags and hitches of life many therapists claim to cure. A final and conclusive triumph over ages of quackery: Oppenheimer was an amateur !

DC’s post is still an interesting reading because :

  • it calls attention on an important statistical feature, regression to the mean, that is probably more complex than the bare idea many of us have of it, me included
  • it implicitly invites the reader to improve his knowledge of the placebo effect and the multiple forms it may take
  • it recalls us that even great minds (no irony here) make mistakes, that no one is above criticism and that you cannot take for granted what a scientist says without paradoxically becoming yourself a sanctimonious believer, a potential prey for… skeptics.

    Complexity in science - the hardest path

    Complexity in science – the hardest path

But pretending that the placebo effect is weak, given the amount of evidence showing the opposite, is surprising. Here are my objections, and part of the evidence supporting them:

Identifying real psychobiological placebo responses

Box 1: Colloca & Benedetti 2005 – Nature Reviews. Neuroscience

1. DC could have had his eureka-‘regression to the mean’-moment ten years ago

In a 2006 post (BBC 2 and the Open University on Alternative Medicine) DC cites an article published in Nature Reviews – Neuroscience Placebo and painkillers written by two well-known researchers in the placebo effects, Colloca and Benedetti.

Besides the fact that the article is a seminal piece of work on placebo that is well worth a read, it contains an explanatory insert (see pict. – Box 1) where they discuss the issue of regression to the mean and suggest a possible explanation of the lack of power findings in Hrobjartsson & Gøtzsche’s review.

2. Hrobjartsson & Gøtzsche’s review is weak, placebo effects are strong

Systematic review is a brilliant concept but not totally free from potential flaws. The first one lies in the fact that it remains a human activity and even if the process of systematic review is intended to smooth out the human factor, you cannot completely get rid of it. Things are even more eloquent when they come close to complex phenomena like the placebo effects. It seems that one cannot carry out an accurate and useful systematic review without having previously gained a thorough knowledge of the reviewed topic. In spite of the respect I have for Gøtzsche’s work about over-diagnosis and unncessary medicine, the review he led with Hrobjartsson looks like researchers dropping Napalm on a field and then pretending flowers don’t exist.

Ben Goldacre is the kind of skeptic I prefer because he doesn’t content himself with denouncing bad science but is also pragmatically committed to improve future research. He writes « I think the real scientific story of the connections between body and mind are infinitely more interesting than anything concocted by the miracle-cure community, but here it is enough to remind you that the placebo effect is very powerful (…) » (Bad Science, 2008, p. 138)

In a top ranked journal, Ben Colagiuri discusses how participant expectancies may limit the validity of DC’s blue-eyed-boy-solution-to-any-serious-problem-in-the-world : double-blind placebo-controlled RCTs. Colagiuri explains in detail three potential biases that may considerably limit the validity of double-blind placebo-controlled RCTs (the article is available here):

  • effective blinding is too rarely assessed, and probably often broken, in many RCTs
  • placebo and drug effects may not be additive like we naïvely used to think
  • RCTs’ conditions of drug delivery do not reflect how the drug is delivered in daily care (placebo effect value may be different in daily care conditions).

Whoever takes time to carefully read the above cited articles by Colloca and Benedetti, Colagiuri, plus this most recent and comprehensive review (The placebo effect: from concept to genes) would realize that this time DC has got it wrong. I’m not here to blame him, science makes progress thanks to successes and mistakes, and I invite whoever never failed to cast the first stone. What is absolutely necessary, given his blog’s reputation, is to prevent readers from following him too close towards what looks like a dead-end this time, and invite them to explore more deeply the evidence about the placebo effect and the complex neurological mechanisms which underlie it.

Marco Gabutti (@mgabutti), Normandy, 31st January 2016