http://www.nejm.org/doi/suppl/10.1056/NEJMct0806114/suppl_file/nejmct0806114_appendix.pdf

there is this supplementary listing of the major studies that we ofen cite…notice closely that within the results column, they show multiple differences “favoring acupuncture” over sham is how they write it, but it is not statistically significant….just writing it that way shows how flawed they see this….no it could not just be a trend caused by some noise…it must favor accupuncture? Not sure if I am reading into this, but I would make sure to qualify it as “not statistically significant” or better yet, “no difference detecte”

The problem with your definition is that when you test specific treatment modalities against “placebo” versions of the treatment modality (as in real acupuncture with needles vs sham acupuncture with toothpicks), you can end up concluding “toothpicks work too!”, rather than “acupuncture is a placebo”.

If the doctor goes to give you a sugar pill, but palms it so you don’t really get it, you only think you get it, and you have a therapeutic effect. Was that a placebo effect because you didn’t really get a placebo, the doctor tricked you into thinking you got it. I think using such a narrow-hyper technical specification of the placebo effect is not useful if we are going to understand the physiology behind it.

A placebo effect can come from non-health care providers too. If you shared a hospital room with someone who was loud, abusive and bullying, you would very likely do worse than if you shared a room with a kind gentle and nurturing person. You may not have the expectation that you will do worse with the abusive bully, but you will.

If you did a clinical trial with abusive bullies and kind and nurturing room mates, you would see greater improvement with the kind and nurturing person. I suspect enough improvement that if you could bottle it, it would be worth tens of dollars per day due to reduced hospital stay length. If it reduces stay length by 1%, and hospital beds cost $1000 per day, that is a $10 per day savings.

http://www.physorg.com/news199986147.html

This would also indicate that different groups would have different responses to different placebos. (I believe there is experimental evidence to back that statement)

Bedside manner is a tricky bordeline case, I would admit. I could go either way on that.

Good bedside manner could produce confidence that their doctor is going to help them, which would fall under the more typical ‘expectation effects’ definition I’ve advocated. On the other hand, it may just relax them and calm them down, which has independent benefits. Whether we call all such auxilliary psychological responses to treatement ‘placebo’ is a fuzzy case: I could go either way. But that doesn’t mean it is a clear cut or canonical instance of placebo effect, which after all is the doctor giving you a sugar pill and saying it is morphine.

The crux of placebo is belief in the animal that something will produce a certain desired result. All other uses, by my lights, are secondary.

I hate arguments via dictionary citation (e.g., the Pais-indle strategy), but for what it’s worth, OED defines placebo as:
‘A drug, medicine, therapy, etc., prescribed more for the psychological benefit to the patient of being given treatment than for any direct physiological effect; esp. one with no specific therapeutic effect on a patient’s condition, but believed by the patient to be therapeutic (and sometimes therefore effective). Also: a substance with no therapeutic effect used as a control in testing new drugs, etc.; a blank sample in a test. ‘

Bold added by me. This is what I have been saying is crucial.

What ever the placebo effect is, it affects a great many different aspects of physiology. It must be very complex and integrated into the rest of physiology. Something that complex can’t suddenly appear without evolutionary precursors.

I disagree that placebo effects necessitates the expectation of improvement. If that were true, then organisms that had no expectations could not exhibit placebo-like effects. The evolutionary precursors of placebo effects didn’t spring up de novo with the first appearance of a nervous system complex enough to have expectations of improvement.

The definition of placebo effect that I am using is positive therapeutic effects not mediated through drugs or surgery. To me, the component of improvement that is produced by a good bedside manner is mediated through the placebo effect (because it is not due to drugs or surgery). A neutral bedside manner is not as good, and a harsh bedside manner is actively bad for patients. If you treat a sick animal harshly, it will do poorer than an animal that is treated gently. That is especially true for social animals like horses, dogs, and mice. That differential therapeutic result is not mediated through drugs or surgery, it is a placebo.

If we did a clinical trial with one group treated harshly and one group treated with a good bedside manner, the good bedside manner group would do better. EBM would say, aha! A good bedside manner is an effective treatment. SBM would say, no, you didn’t use the right control group, you need to provide the equivalent of a “placebo good bedside manner”. You apply a “placebo good bedside manner”, and the effect goes away demonstrating that a good bedside manner is a placebo.

I understand that many people don’t want to call a good bedside manner a placebo. They need to get over that bias about semantics.

And I frankly question what you are saying. If I don’t control for temperature, but temp covaries with my treatement so I wrongly think my treatement has an effect when it is really temperature, that isn’t placebo effect. It would simply be a confounding variable.
Animals that cannot build expectations cannot exhibit placebo effects. If confounding variables were sufficient to show placebo effects, then plants and single cells could display placebo effects, as could iron bars for that matter.

I do a lot of reading of clinical literature and I have never seen it used so broadly. But whatever, I am not one to get too bogged down in semantics. As long as terms are clear and used consistently (i.e., no equivocation) then it isn’t that big a deal how you define the words.

By the operational definition of placebo in trials, animals do have placebo effects.

If you are going to use a more narrow definition (not include observer bias, for example) then you have to also control for the things you are excluding – and trials rarely do that. When they do, btw, the placebo effect shrinks to nothing, or just subjective reporting.

It is important to understand the disconnect – using the size of the placebo effect measured in trials, which includes a variety of effects including bias, regression to the mean, and observational artifacts – to conclude that a subset of the placebo effect (by an arbitrarily narrow definition) is real.

It seems reasonable to say that associations can produces placebo-like effects in nonhuman animals. However, if someone wants to be more strict and insist that the animal must “recognize the intentional effort to treat” for a placebo effect to exist that would also be reasonable. And the latter is very uncommon, if not nonexistent, in nonhuman animals.

Interesting topic!

“The point is that in animals the placebo effect (i.e., expectation of effects) must be built up via some kind of conditioning.”

Most also exclude conditioning in the definition of the placebo effect.

David Ramey in the science based medicine blog:
http://www.sciencebasedmedicine.org/?p=263

…there’s no evidence whatsoever that animals can benefit from, or even experience, placebo effects.

…there are many explanations for how a placebo-like effect might be explained in animals. Take conditioning…

The point is that in animals the placebo effect (i.e., expectation of effects) must be built up via some kind of conditioning.

I originally claimed that acupuncture studies in animals eliminates the possibility of placebo effects, but that was incorrect as pointed out by Steven. However, I should have said it is much easier to block that explanation of data, as to generate such effects takes a bit of work on the experimenter’s part. For instance, associate morphine and needle insertion in one group of animals, then test that group on needle insertion alone versus a naive group. My bet is the morphine group would show greater effects of acupunture needle insertion, and this would be a placebo effect.

Because of the positive effort needed to create an expectation in the animal, it is basically effortless to block placebo effects in animal studies.

That’s why the results in mice in the Nature study impressed me: no placebo in that study, but demonstration of localized transient analgesic effects of using acupuncture needles.