Placebo effect, not placebo treatment

In the course of reading the comments in the last several posts, I’ve come upon many mentions of the “placebo effect”. Steve Novella has a few good posts on the placebo effect, but I’d like to take a look at the clinical view.

The placebo effect is a phenomenon often observed in clinical studies. When doing clinical studies, there is often a notable change in subjects response simply by being in the study. This effect is multifactorial, often due to such biases as a desire to please researches, better medical follow up, and others. It is most often a data artifact that arises when studying human subjects.

It may seem, based on certain misunderstandings of placebo, that it would make a nice therapeutic intervention. After all, placebos are, by definition, inert and therefore harmless, so any benefit derived is essentially without risk. But placebo (by definition) not real medicine.

First, what treatment will be used as a placebo? Sugar pills? Homeopathic potions? Acupunture? There are big differences in cost, and some studies show that treatments that cost more work “better”. Is that a good thing?

Second, the placebo “effect” isn’t a real, measurable, reliable effect. There is no way to predict who will benefit or to what degree. Prescribing a placebo is a bit like saying, “Try hard to feel better. If it works, I let my sugar pill take the credit.” Unlike proven interventions, there is no predictable dose, response, or benefit.

There is also an ethical problem with some placebos. Some of the placebo effect arises simply from being cared for (as is seen in many clinical studies), but a few placebo treatments require actively deceiving a patient. For example, if you tell a patient, “Hey, I’m going to give you this homeopathic treatment. It doesn’t really do anything, but maybe you’ll get lucky,”—if you tell a patient that, it is a reasonable guess that some of the placebo effect may be mitigated by sketicism.

As has been pointed out by others, even were a placebo to be somewhat predictable, there is only a limited number of symptoms that might even be amenable. For example, it’s reasonable to posit that some pain relief might be gained from placebos, as pain has a large subjective component. But you won’t see placebos curing cancer, aborting a heart attack, or lowering blood sugar.

Placebo is an interesting phenomenon, but is not a clinical tool. It’s appearance is serrendipitous, and its use unethical.


Comments

  1. I wrote a few lines about the marketing of Obecalp (placebo spelled backwards, of course) to parents last week, and comments seemed pretty evenly split between ‘hey, whatever keeps the kids from crying’ and ‘that’s unethical and runs the risk of encouraging unhealthy attitudes towards pill popping.’

  2. Second, the placebo “effect” isn’t a real, measurable, reliable effect. There is no way to predict who will benefit or to what degree.

    By this criteria, many widely used pharmaceuticals do not produce a “real, measurable, reliable effect.” Many drugs have a quite substantial variance in their effects, with many patients failing to respond. Often, the only way to predict whether or how much a particular patient will respond is to try it. Perhaps this will change as pharmacogenomics becomes more sophisticated, but that promise is yet to be realized, with the exception of a few specific instances, such as some chemotherapeutic agents.

    As has been pointed out by others, even were a placebo to be somewhat predictable, there is only a limited number of symptoms that might even be amenable. For example, it’s reasonable to posit that some pain relief might be gained from placebos, as pain has a large subjective component. But you won’t see placebos curing cancer, aborting a heart attack, or lowering blood sugar.

    Most drugs are effective in only some conditions, so that is hardly unique to placebos. You don’t see acetaminophen curing cancer, aborting a heart attack, or lowering blood sugar, either. Does that mean acetaminophen should not be used?

    I agree that there is an ethical conflict between the physician’s responsibility to provide a patient with the most effective treatment available and the physician’s obligation to be honest with his patient, but I don’t think that you can evade the dilemma by convincing yourself that the placebo effect is not “real.” You cannot escape the subjective in medicine; in many illnesses, the ultimate measure is the patient’s experience. This is a problem that many physicians are confronting. In a recent survey, nearly half of the physicians admitted to having administered a placebo on at least some occasions, and only 12% felt that placebo use should be prohibited.

  3. By the way, here is The Onion’s take on placebos. Be sure to check out the “ad.”

  4. Kagehi

    Its, rather ironic though that mental state “does”, sometimes drastically, effect recovery, so… I think its almost a grey area. At what point does not lying to the patient, by, at minimum, exaggerating their chances of recovery, or otherwise seeking to trigger the effect, actually constitute “doing harm”, by instead triggering the wrong effect and damaging recovery? Its not quite as cut and dry as people would like, and while lying to them all the time, and providing what is false treatment is unethical, so can be reducing their odds of recovery in some cases by “being” honest. The problem because then, how do you trigger the positive effect when most desperately needed, without lying to them at all?

    Fact is, at this point, we have gotten so good at physical damage repair that the *psychology* of the patient can often have as much or more of an impact than the rest of the treatment, and for those things we are still poor at curing, due to their complexity or high odds of poor recovery, even when *using* the best we have, than anything we can do “physically” to help them at that point. So again, there could be cases, where the psychology of a particular patient, or just the overall psychology of the situation they are in, would make being blunt about what is happening, or just insufficiently excited and believable about it, *could* actually harm or kill someone.

    If this wasn’t the case, there wouldn’t be “real” research going on about how much of an effect it can/does have, and how to “ethically” produce the positive results, when ever possible.

    The real problem though isn’t even the ethics of the matter, its the fact that 90% of the people using it have no ethics and are peddling woo, instead of attempting to improve patient recovery and survival. I.e., they are harming/killing more people *by* using it and ignoring real treatments, than they are helping. In the end, we are going to either have to, imho, accept that “some” dishonesty may actually sometimes be good for the patient, or accept that doctors **will** harm or kill patients by being completely honest, both “entirely” do to the simple fact that the body “is” effected by the mind, so its the mental state that has to be treated, or ignored at the cost of resulting failures, along with the body itself.

  5. trrll, your point is interesting, but I think it doesn’t quite hold up. You use a sort of variant on the tu quoque fallacy—just because proven medicines are not panaceas does not help validate placebos. One of the problem with placebos is that it is impossible to know what conditions may respond to a placebo effect and when, making it, as a therapeutic procedure, useless.

    As noted, the effect is multifactorial, and more artifact than physiologic effect.

  6. Kagehi

    Didn’t say it 100% validated placebo, or even done so *at all* other than as a purely psychological aid, which may or may not work for a specific patient. It **does** matter a lot what the psychology of the patient is, much as hypnosis is less useful when directed at someone who isn’t easily hypnotized. Recovery also “does” depend highly on that mental state too.

    Put simply, we don’t know enough at this point to tell when or if to use it, we might need to know more about the patient than they would want us to, in order to figure out if it “is” useful in their case, and it *may* be a toss up whether the ethics of figuring out if it can be used effectively in a particular case outweigh the ethics of not using it at all. Its not like anyone has, until recently, even “attempted” to make real complete studies of the matter, from the perspective of how and if it should be used, not just, “if it may sometimes work.”, which we do know it can, sometimes, or, “if I can make lots of money selling people shit that doesn’t do anything.”, which seems to be the one true unequivocal success about them… And, also the #1 reason it **is** as big an ethical issue as it happens to be. Since no “solid” rules exist, there is no way to *not* cross ethical lines if you opt to attempt it.

    Oh, and just to be clear, I am using the term far broader than what you are. I would consider “anything” that was intended to increase the state of certainty in the mind of a patient that they *will* get better a placebo. Why? Because that is all a placebo does anyway, make people think they are receiving something that helps them, be it a pill, a patch, a positive testimonial on the treatment they are going to get, a doctor flat out lying to them that they will be OK, etc. If it helps or hinders recovery, at all, due to mental state, it is a placebo. And, I think you can agree, with that broad definition, the lines get damn fuzzy. What you may not agree with is *merely* that somehow its less ethical to present a “physical” representation of placebo, such as a pill, or a saline shot, or some other “treatment”, vs. telling them a falsehood about their likely chances.

    Again, I personally think that the former is “only” a bigger issue precisely because there is poor understanding of when, how and if they can be used, and a literally endless line of con artist who don’t give a frack, and *will* use “both” to con people into buying their fake remedies.

  7. qetzal

    I have to mostly side with trrll here. In fact, we do know a number of conditions that consistently respond to placebo effects. People who design clinical trials routinely take that into account when calculating group sizes and expected statistical power.

    It’s certainly true that placebo effects can vary widely from trial to trial and person to person, even for a single condition. But as trrll says, the same is true for many bona fide drugs.

    However, I completely agree with PalMD that this doesn’t help validate placebos as medically appropriate (outside of a trial setting).

  8. PalMD, you have expressed the standard mainstream view of the placebo effect, which is ok, however it is somewhat more complicated than you and the others appreciate.

    I am in complete agreement that placebos have no place in actual treatment of actual patients. I use a different basis for saying that. Even if placebos had a certain statistical chance of working they would still have no place in medicine. I think this was something that came from Steve Novella’s blog. What they do is violate the patient’s sense of agency, that is the patient’s right to make informed decisions about their health care. Deceit by the physician prevents the patient from being informed and so any type of care that hinges on deceit is unethical regardless of the outcome. The second reason placebos have no role in treatment is because it puts the physician in the untenable position of practicing with out a theoretical basis, out there alone in never-never land where no one can (as you say) “have his/her back”. I consider that to be an unacceptable burden to put on a physician (or on anyone).

    On another thread I gave what I consider to be my definition of a placebo treatment, that would be something purported to be a treatment which does not have chemical or physical activity, i.e. such as drugs or surgery. I would consider psychotherapy to be a “placebo treatment”, in that it does not involve anything chemical or physical. The effects of psychotherapy are mediated through communication. I say this as someone who has had 20+ years of psychotherapy and has benefited greatly from it and would recommend it to anyone considering it. It is not meant to disrespect psychotherapy, but rather to put it in the proper perspective.

    I consider psychotherapy to be effective and ethical treatment; I consider subluxation adjustments are not ethical even if they are effective. The difference lies in the theoretical basis for the treatment, what the patient expects, how it is conducted, how it is taught. I appreciate that there are some physicians who consider psychotherapy to be woo-woo medicine in part because psychotherapy can’t be subjected to a double blind placebo controlled study.

    I think there is a problem with the definition of placebo and what treatments have results mediated through the placebo effect and how those therapies are carried out. I think there is an effort to confabulate what is actually meant by the term “placebo effect”, such that proven therapies such as psychotherapy are not covered by that definition while quack therapies such as subluxation adjustment and acupuncture are.

    There is real physiology behind the placebo effect. The effects are not simply mistakes and wishful thinking. Some of the gross physiological effects are mediated through NO and ATP physiology as I discuss on my blog. The subtle effects are more subtle and require more subtle invocation, as in psychotherapy. Once the placebo effect is maximally invoked, there is nothing more that any placebo can do. Once EBM is able to invoke placebo effects reliably and cost effectively, I think there will be no room left for CAM.

  9. …and its use unethical.

    Except in certain, very specific cases of dire need :):

    http://youtube.com/watch?v=d0S7OQ6y5dI&feature=related

    (This was, incidentally, where I first learned about the placebo effect.)

  10. You use a sort of variant on the tu quoque fallacy—just because proven medicines are not panaceas does not help validate placebos.

    I disagree. You are effectively arguing that a placebo effect is not real because placebos exhibit properties that are typical of many (perhaps even most) other proven medications. Accepting an argument as valid when it supports your thesis, but rejecting it when it does not, is rationalization, not reasoning. If you are willing to accept this as a basis for rejecting the use of placebos, then you must also accept it as a basis for rejecting the use of “active” drugs that exhibit the same limitations.

    One of the problem with placebos is that it is impossible to know what conditions may respond to a placebo effect and when, making it, as a therapeutic procedure, useless.

    Why is it impossible? Placebos may be tested in randomized clinical trials just like active drugs. You simply test them against no treatment rather than placebo. For example, one metaanalysis by Hrobjartsson and Gotzsche (who are perhaps the most skeptical investigators when it comes to placebo effects) found significant effects of placebos vs. no treatment with respect to pain relief. A number of other conditions failed to exhibit a response to placebo. So purely from an EBM standpoint, one could argue that the current evidence supports the use of placebos in pain (presumably for patients who do not receive adequate relief from active analgesics alone), but not in other conditions.

  11. trll, there are many highly effective medications for pain relief, medications that are much more effective than placebo.

    Because placebo sometimes works better than nothing is no excuse for using it instead of something that is more effective still.

  12. trll, there are many highly effective medications for pain relief, medications that are much more effective than placebo.

    Because placebo sometimes works better than nothing is no excuse for using it instead of something that is more effective still.

    Indeed there are, but they don’t work for everybody. There are some people who cannot tolerate opiates, which takes the most efficacious class of analgesics out of the running. And there are some types of pain (neuropathic pain, for example) that are not adequately relieved by opiates, or by any other currently available medications. So why shouldn’t a physician add a placebo if the standard medications fail to do the job? I am not necessarily advocating for this approach, as there is still a fundamental conflict with what I see as the physicians duty of honesty. Of course, there may be a middle ground between lying to your patient and denying an effective therapeutic to a patient in pain. A physician could, for example, direct his patient to a homeopathic medication, while telling him quite accurately, “Medical science can see no scientific basis for pain relief by this preparation, yet some people claim that it helps. You might try it and see if it does anything for you.” I wouldn’t be surprised if many of the 45% of Chicago internists who reported having administered a placebo to a patient did something of this sort.

  13. Chris Noble

    Second, the placebo “effect” isn’t a real, measurable, reliable effect. There is no way to predict who will benefit or to what degree.

    I agree with trrll, this isn’t by itself a sufficent reason to dismiss placebo effects.

    However, some people such as Andrew Weil, make such stupid claims about the power of placebos that are completely disconnected from reality.

    One of Weil’s pseudo-arguments is that if you look at any placebo-controlled randomised trial then you will find somebody in the placebo group who shows all the benefits of the treatment. (I don’t believe this claim)

    Weil appears to presume that the benefits to that individual must be due to the power of the placebo. This is hardly a valid conclusion and even if it was then what about the other 99.9% of the placebo group who didn’t show the benefits. Can Weil tell us who is going to benefit from placebo?

  14. Dianne

    Silly anecdote about placebos: Once when I was in medical school I ran out of nonsedating antihistamines on a weekend. At the time, the antihistamines were prescription only and so I couldn’t get any more, not yet having prescribing privileges anywhere. I did have some ibuprofen laying around, so I took that on the off chance that I’d get a placebo effect out of it. Oddly enough, I did. Even though I knew it was the wrong drug and all I was trying for was placebo (although one could argue that suppression of cyclooxygenase might reduce symptoms somewhat, I doubt the biological effect was significant.) So placebos can, at least anecdotally, work even when you know they are placebos, maybe due to a sort of operant conditioning to the expectation that taking a pill will result in feeling better? So for symptomatic relief, maybe one could simply tell the patient “You could try this. It’s a placebo but placebos work for pain relief X% of the time and it certainly won’t do any harm.”

  15. The second reason placebos have no role in treatment is because it puts the physician in the untenable position of practicing with out a theoretical basis, out there alone in never-never land where no one can (as you say) “have his/her back”. I consider that to be an unacceptable burden to put on a physician (or on anyone).

    daedelus2u, I don’t think that I can go along with your insistence on…shall we call it “theory based medicine?” Historically, medicine, and especially pharmacology, has been largely an empirical discipline. While theory is used to guide drug discovery and clinical experimentation, I don’t think that we have reached a level of knowledge where it is reasonable to demand not merely evidence, but understanding, for every therapeutic. Many valuable medications were used long before there was a theoretical understanding of their action. To pick an example in your are of interest, do you think that it was wrong for doctors to prescribe nitrates to patients with angina in the years prior to Furchgott’s discovery of EDRF?

  16. One of Weil’s pseudo-arguments is that if you look at any placebo-controlled randomised trial then you will find somebody in the placebo group who shows all the benefits of the treatment. (I don’t believe this claim)

    Yes, this is an invalid claim, because placebos control not merely for the effects of suggestion and the therapeutic interaction, but also for regression toward the mean. Many medical conditions improve spontaneously with time, and even those that don’t tend to show some variation over time. Since people are most inclined to seek medical aid when they feel particularly bad, there is a statistical expectation that they will feel better in the future whether or not something is done. This is why a valid test of the placebo effect must compare placebo to no treatment at all.

  17. Chris Noble

    Yes, this is an invalid claim, because placebos control not merely for the effects of suggestion and the therapeutic interaction, but also for regression toward the mean. Many medical conditions improve spontaneously with time, and even those that don’t tend to show some variation over time. Since people are most inclined to seek medical aid when they feel particularly bad, there is a statistical expectation that they will feel better in the future whether or not something is done. This is why a valid test of the placebo effect must compare placebo to no treatment at all.

    Not just invalid, but mind-numbingly stupid.

    That doesn’t mean that true believers don’t find it convincing.

    My favorite story on this is from Dr. Andy Weil who would routinely asked his med students to randomly pick any major study off the library shelves and notice that the control group ALWAYS showed some cases of spontaneously healing at least as dramatic as the test group.

    The idea that there is a powerful untapped placebo effect out there waiting to be utilised is psychologically very attractive.

  18. “The idea that there is a powerful untapped placebo effect out there waiting to be utilised is psychologically very attractive.”

    Yes, it is psychologically attractive, and the post hoc hypothesis of “regression to the mean” has no power to explain what ever physiology may be behind it.

  19. My favorite story on this is from Dr. Andy Weil who would routinely asked his med students to randomly pick any major study off the library shelves and notice that the control group ALWAYS showed some cases of spontaneously healing at least as dramatic as the test group.

    Once you start selecting out “responders” from a population after the fact, you run the risk of fooling yourself–essentially subjecting yourself to a version of the “racetrack tout con” in which the con artist gives a large number of people at the track, for free, different “predictions” (actually random guesses) for the winner of the first race, the revisits the subgroup for which that guess turned out to be right, and gives them another set of predictions. If you start with a large enough population, you end up with a handful who have seen the tout correctly “predict” each race in succession, and are now willing to pay handsomely for the next prediction.

    The point is that in any treatment group, about half of them are going to appear to do better than the median, and a few are going to do much better, so it is easy to fool yourself by picking out these “responders.” Of course, one always worries, when a clinical trial comes out negative, that there is actually a subgroup of patients who respond very favorably, but who are lost in the mass of those who do not. But there are statistical methods and study designs to deal with this situation. For example, I’ve seen studies in which the population that seems to respond is then subjected to a double-blind placebo discontinuation trial, in which half of them have the experimental drug switched to a placebo. One could obviously do something similar with placebo versus no treatment, randomly selecting half of the placebo “responders” for discontinuation of placebo administration.

Leave a Reply

Your email address will not be published. Required fields are marked *