Nov 08 2007

Placebo Myths Exposed

In a recent Slate article, Darshak Sanghavi exposes the mythology surrounding the placebo. He correctly points out that the common belief that the “placebo effect” is primarily a mind-over-matter real physical healing is nothing more than a myth. Reviews of the medical literature show that for subjective symptoms, especially pain, there is a decrease in the the reporting of such symptoms by about 30% with placebo, or inactive, treatment. However, for biological or objective outcomes, like survival from cancer, there is no significant or measurable placebo effect – sugar pills do not increase your chance of surviving cancer.

Rather than a placebo effect, there is, rather, a trial effect – namely the fact that a patient is enrolled in a clinical trial will improve their outcome. This is due to a host of factors – being in a trial means getting more medical care and attention, and it inspires patients to think more about their care and thereby to take better care of themselves, for example by being more compliant with their treatments. In those trials where a placebo group was further compared to a group without a real treatment or placebo treatment – an observation group – there is no difference in outcome between placebo and observation groups.  So getting the placebo had no advantage – the belief that one was being treated was not enough to cause a real biological healing.

Why, then, Sanghavi asks, are we so insistent on placebo-controlled trials? While I agree with Sanghavi that we need to listen to the evidence, and be clear that there is no biological response to placebo treatments, I think he goes too far in recommending that placebo-controlled groups are not necessary. The real reason for a placebo is not to rule out a real biological effect from the belief of being treated, but to assure proper blinding of those who are assessing the outcomes (which may include the subject themselves, the treating physician, or a third observer not otherwise involved in the trial).

Proper blinding is necessary to eliminate bias in assessing the outcomes of the trial. Bias is a huge problem in clinical trials. The overall pattern of research is that the less well controlled a trial the more it will reflect the biases of the researchers. The best blinded and controlled trials tend to be the most reliable (in terms of predicting future trends in the research literature).

In addition, scientific medicine is being infiltrated by ideologically driven beliefs and practitioners. Under the umbrella of so-called alternative medicine, there is a host of dubious and unscientific treatments, all clamoring for scientific respectability, with true-believer advocates doing the research. We need to move toward more rigorous trials designs and higher standards of blinding – not weaker. Sanghavi is living in a world of scientific researchers testing plausible treatments under reasonable conditions. But we need a system that can deal with the nonsense as well, otherwise they will exploit the weaknesses in the system to further infiltrate medicine.

Also, I am not letting the pharmaceutical and supplement industry off the hook – review of literature shows that industry sponsored research is more likely to support the desires of industry. Again – bias has an effect, and the only way to weed it out is through methodological rigor, including blinding of trials.

So while I agree with Sanghavi that we need to dispel the placebo myth, and I also agree with him that in some cases (like doing sham surgery as a placebo) we can control trials in other ways, for the most part we still need placebo-controlled trial designs to maximize blinding and thereby minimize bias in clinical trials.

8 responses so far

8 thoughts on “Placebo Myths Exposed”

  1. daedalus2u says:

    I completely agree, there are a lot of misconceptions and myths about the placebo effect. I see it as a completely real effect mediated solely by physiology, and have posted a blog about how I understand the physiology of it.

    As I see it, it derives from the normal allocation of metabolic resources into the myriad different tasks that cells must do, the two extremes of “fight or flight” and “rest and relaxation”. Metabolic resources of each cell are limited, and what is used to repair cells cannot be used to “run from a bear” (the extreme example of fight or flight that I use). Shutting down long time constant pathways to conserve ATP for short term consumption is a critical stress pathway. I think that is what ischemic preconditioning is, the turn off of ATP consuming pathways that can be delayed for a while (during the acute ATP crisis of the ischemic events). The pathways that are turned off are not unnecessary, they are needed, just not during a short ATP crisis where other more important pathways need the ATP they would otherwise consume. The ischemic preconditioned state can’t be a state that cells can survive in long term (otherwise they would evolve to be in that state continuously).

    I think that “stress”, invokes these same ATP conservation pathways, to make more ATP available for immediate consumption. That means less ATP for the non-voluntary pathways (such as healing) that can be deferred until later, until after the ATP crisis, or until after the “stress” has passed. If the ATP crisis doesn’t end, then the organism goes down the low ATP death spiral as cellular processes degenerate from insufficient repair. Physiology is extremely robust, but not infinitely robust. Insufficient repair will eventually degrade the cell such that the DNA is not sufficiently protected and once the DNA is damaged, the cell has no way to fix it or do without. I see the inappropriate invocation of the ischemic preconditioning stress response as a major final common pathway in many degenerative diseases.

    I see the placebo effect as the neurogenic switching from the “fight or flight” state to the “rest and relaxation” state. The “standing down” of emergency diversion of ATP from repair to “running from a bear”. The “fight or flight” state can be invoked neurogenicly, that is what athletes do to “psych themselves up”. The “rest and relaxation” state can be invoked neurogenicly too, that is the “relaxation response”.

  2. saburai says:

    Steve, don’t forget that the use of placebos in clinical trials is heavily promoted by placebo manufacturers. See here, for example:

  3. TheBlackCat says:

    We read a paper a while back where they did fMRI while people were getting a placebo pain treatment. There were measurable differences in the part of the brain that controls the endogenous opiate pathways, the body’s own pain-suppression system. This would both indicate a mechanism by which the placebo affect could take place as well as explain why only pain seems affected by placebos.

  4. jonny_eh says:

    If the brain can control sense responses like pain and nausea, then why can’t a false belief in the relief of the cause of the sense reduce the pain/nausea?

    How else can you explain the apparent (but not actual) effectiveness of homeopathic treatments? To me, it seems like if it’s for things like pain or nausea relief, which is all in the brain, then it makes sense. If it’s for fighting an infection or cancer, then obviously no placebo can make a difference.

  5. daedalus2u says:

    Actually nausea is not only in the brain, there is a lot of involvement of the enteric nervous system.

    The point of this study is that a false belief in the anti-nausea effects of a placebo actually makes nausea worse, while a false belief that the inert material will make the nausea worse actually makes it better. I think this data reflects on the mechanism(s) by which placebos and nocebos actually have physiological effects.

    Endogenous opiate pathways are one mechanism for neurogenic control of gain in pain sensitivity. No doubt there are others.

    Virtually every organ is enervated. Presumably there is an evolved reason for that enervation that involves some sort of neuronal feedback and “control” of some aspect(s) of the physiology of that organ. Even organs that are completely de-enervated at transplant (such as the heart), do become re-enervated to some extent in some cases. There was a recent study that showed that ablation of pancreatic sensory neurons in NOD mice via capsaicin reversed autoimmune diabetes.

    I think that different disorders need to be examined separately to determine what (if any) placebo or nocebo effects there are. Immune system function is very complicated. Different components are distributed all over the body and so necessarily communicate with each other via multiple channels. Disruption of that communication is going to degrade function. I suspect that degradation in communication via NO mediated communication pathways is the main effect of the low basal NO that I am studying. But whether that disrupted communication manifests itself as an overactive immune system as in autoimmune sensitization, allergies, asthma, or anaphylaxis, or as an insufficient immune response and insufficient clearance of some infections or tumor cells (or both simultaneously) is unclear. Studying systems that are under feedback control is quite difficult because they adapt and change as you try to measure them.

  6. jonny_eh says:

    daedalus2u, I don’t understand the stuff you write, too much jargon. It sounds interesting, but I can’t understand it. Could you please write at the same level as the blog author that you comment on? (i.e. if you think Dr. N would explain a certain word when using it, you should too)

    “Endogenous opiate pathways”

    I could look these up, or figure out the meaning based on context, but I’d rather just give you this tip.

  7. wertys says:

    The 30% placebo response figure is itself a myth that dates from the earliest research into placebo response in the ’50s. In fact the size of the placebo response varies according to a number of things including (a) whether it is an active placebo or an inert one. The example of this is an experiment where rats were injected daily at the same time with insulin. After a week of this, saline was injected, which has no known physiological activity, but the response of the blood sugar in the rats was the same as if they had been given insulin. In human trials, a placebo treatment which can produce some physical effects, such as amitrityline or a beta-blocker can be included in trials to provide an active placebo.
    (b) Whether it is a ‘procedure’ or a tablet. British researchers published a trial last year of placebo tablet vs sham acupuncture and the acupuncture won by a mile, though both groups improved a little.
    (c) Other expectations. Capsules are more effective placebos than tablets, especially if granules are inside. Multicoloured capsules beat most other oral placebos, and as mentioned above, physical treatments have much bigger responses than tablets.
    (d) the patient themselves. Despite predictions, there is no predictable placebo responder. People who respond to placebos are no more gullible than anyone else. The Italian researcher Benedetti has recently shown that people suffering from dementia which reduces the connections of the frontal lobe to the rest of the brain have a much lower placebo response to pain relief. This is hypothesized as being because the influence of expectation is reduced on the rest of the brain.
    (e) Timing of the placebo. Placebos seem to work better when the symptoms are severe. One possible explanation for this is the phenomenon seen in drug trials of ‘regression to the mean’. This refers to the fact that people seek treatment when their problems have become bad enough to warrant it. the natural history of many conditions is that the symptoms wax and wane over time, so a few percent of any sample will regress to the mean over a study period, and the more dramatic the symptom (such as fever or pain) the likelier it is to have reduced in time whatever treatment is given.

Leave a Reply