White Coat Underground

Placebo is not what you think it is

If I read one more crappy article about placebos, something’s gotta give, and it’s gonna be my head or my desk. Wired magazine has a new article entitled, “Placebos Are Getting More Effective. Drugmakers Are Desperate to Know Why.” Frequent readers of skeptical and medical blogs will spot the first problem: the insanely nonsensical claim that “placebos are getting better”. This not only “begs the question,” but actually betrays a fundamental misapprehension of the concept. I’ve written several times about the nature and ethical implications of placebos, but it’s time for a serious smackdown.


In clinical studies, “placebo” refers to a “treatment” which, compared with the test treatment, is inert. If I want to test a blood pressure pill for basic efficacy, I could simply give it to a group of people and see what their pressures are before and after taking the medicine. But that wouldn’t account for any blood pressure changes that occur simply from being part of the study or by chance alone. Such effects include paying better attention to one’s health and habits, trying to please the doctors, and other less clear and tangible effects.

So, to see what the real effect of the new pill is, it can be tested against a identical-appearing dummy pill, which we call “placebo”. Neither the experimenters nor the subjects know which pill is which. Then, the change in blood pressure between the real and placebo pill groups, and perhaps a no-pill group as well, can be compared to see if and to what extent the new pill lowers blood pressure; the key point being that this allows us to see how much of the blood pressure lowering effect is due to the new pill vs. other undefined factors be they random or due to being in a study.

The placebo “effect” is just that—an effect observed because of a particular situation. It does not show that there is some special benefit to sugar pills, but that when we observe people, we can measure changes that are not always due to our intended intervention. It is not possible to create strong or weak placebos, since the placebo effect is a measure of poorly defined effects and of chance alone. It is what it is.

The author of the Wired article doesn’t get it in a very profound way. I’ll give you a few laughable examples.

The fact that taking a faux drug can powerfully improve some people’s health–the so-called placebo effect–has long been considered an embarrassment to the serious practice of pharmacology.

ORLY? That’s not even wrong. Clinical pharmacology research depends on placebo controls to make sense of the data. The fact that certain data is affected in placebo groups does not, for example, mean everything we understand about the cytochrome P450 system is wrong.

He asserts that more and more drug studies are crapping out due to the placebo effect:

It’s not that the old meds are getting weaker, drug developers say. It’s as if the placebo effect is somehow getting stronger.

No, it’s not like that at all. Perhaps the studies are just that well done, or maybe the drugs being developed suck, or maybe companies are studying more candidate drugs and screening for efficacy. Just about any explanation that doesn’t involve aliens is better than “placebo is getting stronger”.

He goes on to talk about how placebo has become a crisis of the industry, but I have another explanation: it’s not “placebo” that’s the problem. If drugs in testing cannot outperform placebo, then the researches have done a good job of testing the drugs honestly. If the researchers are failing to develop drugs that beat placebo and the company’s bottom line is suffering, it’s not the fault of the sugar pill. Sometimes it’s either difficult or impossible to develop an effective medication. Failure is inevitable. It’s how science works. If the CEOs don’t like it, they have to either make up the data, or find a new business model.

Comments

  1. #1 Greg Laden
    September 3, 2009

    and other less clear and tangible effects. Like showing up for the trial because you’ve felt your BP has been high lately, and it has, but it was transient.

    A trial that asked for volunteers who have a cold NOW to take a pill will show that most patients recover within a few days no matter what is in the pill.

  2. #2 Kate from Iowa
    September 3, 2009

    Finally! I made it past all the stupid in the second sentence! Yay me! As for Wired…WTF’nH? I guess I shouldn’t be surprised anymore.

  3. #3 CRM-114
    September 3, 2009

    What would a placebo be for a test of a homeopathic treatment?

  4. #4 dg
    September 3, 2009

    the sugar in the pills must be contaminated by pharmaceuticals from runoff. the only solution is to use organic sugar.

    also, i’m sure chelation can help. and kinoki footpads.

    medicine is more fun when it’s all just random bullshit!

  5. #5 Tex
    September 3, 2009

    I dunno. There may be something to this.

    This is just anectdotal evidence, but on the TV show King of the Hill Dale Gribble claims he was once an unwitting participant in a government test of mind-control drugs, and now he is addicted to placebos.

  6. #6 daedalus2u
    September 3, 2009

    There may be two things that are going on where placebos seem to be working better.

    First, the better the trial is done, the better the blinding of patients and clinicians, the more the placebo effect of the treatment and the placebo match. The cost of trials is going up and up, so everyone is doing them better and better. The “easy” drugs were discovered long ago, the ones that are being worked on now are more difficult.

    Second, I subscribe to the hypothesis that the placebo effect is triggered by the neurogenic production of NO, which I discuss in my blog.

    http://daedalus2u.blogspot.com/2007/04/placebo-and-nocebo-effects.html

    I also subscribe to the hypothesis that there is an increasing epidemic of low nitric oxide in modern societies, and this epidemic of chronic low NO exacerbates many disorders that are characterized by low NO. Because these disorders are characterized by low NO, they are particularly sensitive to anything that increases NO, including the neurogenic increase of NO by the placebo effect.

    The disorders that are characterized by low NO are not going to be helped by anything other than increasing the NO level of the individual. They are actually due to physiology working correctly, good regulation around a bad setpoint, a setpoint skewed in a characteristic direction by the low NO level.

    You can’t raise NO levels via non-natural means. NO physiology is too “stiff”, it is regulated by so many feedback systems that it is very hard to perturb and essentially impossible to perturb without adverse side effects. You may be able to raise it in one pathway, but the feedback will then lower it in another and you get side effects. A good example of those side effects is in Viagra. Viagra doesn’t raise NO levels, what it does is inhibit the phosphodiesterase that cleaves the cGMP that sGC makes in response to NO. Viagra prolongs the effects of NO. What happens then is that the feedback regulation of NO in the pathways affected by Viagra causes a reduction in the amount of NO produced, which then reduces the effects of NO mediated through non-cGMP pathways. This is why a single dose of Viagra causes adverse effects in people with sleep apnea. Breathing is triggered (in part) by NO/NOx pathways that involve S-nitrosothiols, not cGMP. With less NO/NOx, the other breath triggering pathways (low O2 and high CO2) have to get further out of range to trigger breathing, so sleep apnea gets worse.

  7. #7 Steve Silberman
    September 3, 2009

    PalMD, I have to ask if you read my article beyond those sentences that riled you up so much, because I write at length about the history and use of placebo controls in clinical trials, the Hawthorne effect, the error that Henry Beecher made in claiming to see a placebo effect in treating diseases that wane with time, and the fact that care and attention from medical staff plays a key role in the response of placebo control groups — in fact, nearly all of the issues you mention in this post. I make it very clear that I’m not claiming that sugar pills themselves are somehow mysteriously getting more “effective.”

    I certainly appreciate and share your disgust with most press coverage of the placebo effect, but I submit that you’re being a little quick on the trigger here. I went into much more detail of the various factors contributing to high responses in placebo control groups (which I’m very aware is distinct from true placebo response) than most articles for the mainstream press — such as problems with the Hamilton Rating Scale for Depression, inconsistencies in trial ratings, the power of DTC advertising, and cultural dynamics between trial volunteers in impoverished countries and trial staff.

    Just saying: my article is more mindful of these issues than you give it credit for. Hopefully your readers will judge that for themselves.

  8. #8 PalMD
    September 3, 2009

    Respectfully, Steve, you’re entire piece belies a complete lack of understanding of the topic. I appreciate the historical perspective, but I don’t think I’m quote-mining.

    For example:

    In other words, one way that placebo aids recovery is by hacking the mind’s ability to predict the future. We are constantly parsing the reactions of those around us—such as the tone a doctor uses to deliver a diagnosis—to generate more-accurate estimations of our fate. One of the most powerful placebogenic triggers is watching someone else experience the benefits of an alleged drug. Researchers call these social aspects of medicine the therapeutic ritual.

    The placebo effect is artifact. It is not reliably reproducible as its own “thing” rather than the sum of its parts. It cannot be removed from its context or it ceases to be “placebo”.

    I think the piece has a lot of potential, but it is too full of misunderstanding as it stands.

  9. #9 PalMD
    September 3, 2009

    I should also point out that if people you interviewed were “blaming” a “placebo effect” for bad results, they are either bad scientists or feeding you a line.

  10. #10 Steve Silberman
    September 3, 2009

    PalMD, the quote you cite above is from a section of the piece that deals exclusively with academic research on the placebo response itself — not with the data from placebo control groups in clinical trials, which is the “context” you describe. In most clinical trials, as I assume you know, there is almost never a real “no treatment” arm, which would be required to separate out true placebo response from other statistical noise in the data. That’s why companies like Merck, Lilly, Pfizer, and Johnson & Johnson are collaborating on the NIH project I talk about in the piece — the Placebo Response Drug Trials Survey — to parse the data from decades of trials and try to figure out what’s behind the increasing number of failures in the face of “unexpectedly high” response in placebo control groups, like this one, from just a few days ago: http://online.wsj.com/article/BT-CO-20090901-708321.html.

    Believe me, I appreciate the points you’re making. That’s why I went into such detail.

  11. #11 PalMD
    September 3, 2009

    hmm..well, lemme reread the piece in light of what you’ve said…

  12. #12 Charlotte
    September 3, 2009

    I’m half-convinced. While I think the article definitely makes too much of the ‘healing power of the mind’, ‘placebo effect getting stronger’ angles, there are interesting possibilities here about how social conditioning affects expectation, the medicalisation of states of mind, and (not mentioned in the article) the adequacy of animal models used to develop these drugs. Not to mention the ethical implications of doctors prescribing placebo treatments.

    There’s some interesting stuff there – but it could have been put better.

  13. #13 PalMD
    September 3, 2009

    THis is a nice counterbalance:

    In his eagerness to promote his template for clinical trials, Beecher also overreached by seeing the placebo effect at work in curing ailments like the common cold, which wane with no intervention at all. But the triumph of Beecher’s gold standard was a generation of safer medications that worked for nearly everyone. Anthracyclines don’t require an oncologist with a genial bedside manner to slow the growth of tumors.

    But then something like this:

    Benedetti often uses the phrase “placebo response” instead of placebo effect. By definition, inert pills have no effect, but under the right conditions they can act as a catalyst for what he calls the body’s “endogenous health care system.” Like any other internal network, the placebo response has limits. It can ease the discomfort of chemotherapy, but it won’t stop the growth of tumors.

    This is half-right. I don’t agree that inert pills can be used successfully (or ethically) for some endogenous health care system. Inert pills, as you’ve pointed out, are inert, and effects seen are usually artifactual.

  14. #14 Steve Silberman
    September 3, 2009

    Well, PalMD, I suggest you take a look at the work of researchers like Fabrizio Benedetti, Andrew Leuchter, and Tor Wager, who have been studying the mechanisms of placebo response (that is, distinct from statistical noise in trials) for some years now. Placebo effects are ALWAYS artifactual as far as clinical trials are concerned, as you so rightfully point out here. But they’re not artifactual in studies designed to elicit those mechanisms. Benedetti’s book “Placebo Effects” is rigorous in pointing out the difference, and is a great overview of this body of research: http://bit.ly/3tzIjK

    Unraveling the contribution of those mechanisms from the clutter of data in trials is the goal of the Placebo Response Drug Trials Survey I talk about in my piece.

    Anyway, keep up the great work. I enjoy this blog a lot. Thanks!

  15. #15 Hildy
    September 4, 2009

    Steve:

    You don’t discuss the classic work by Hrobjartsson and Gotzche comparing placebo to no treatment arms.

  16. #16 Steve Silberman
    September 4, 2009

    That’s true. Note that I don’t discuss Irving Kirsch’s antidepressant meta-analyses either. That’s because both sets of studies are highly controversial, and neither are accepted as “classic,” if classic means unimpeachable and widely accepted.

    To my mind, the Hrobjartsson and Gotzche study, ironically, overreached in the same way that Beecher’s original paper did, by looking for a placebo effect in too wide a range of disorders. How wide? This wide: “Hypertension, asthma, anemia, hyperglycemia, hypercholesterolemia, seasickness, Raynaud’s disease, alcohol abuse, smoking, obesity, poor oral hygiene, herpes simplex infection, bacterial infection, common cold, pain, nausea, ileus, infertility, cervical dilatation, labor, menopause, prostatism, depression, schizophrenia, insomnia, anxiety, phobia, compulsive nail biting, mental handicap, marital discord, stress related to dental treatment, orgasmic difficulties, fecal soiling, enuresis, epilepsy, Parkinson’s disease, Alzheimer’s disease, attention-deficit–hyperactivity disorder, carpal tunnel syndrome, and undiagnosed ailments.”

    Nail biting? Poor oral hygeine? Mental handicap? Fecal soiling? Bacterial infection? Herpes? Imagine if they had been examining the effects of a real drug. What drug would be effective for that wide a range of conditions? Answer: Not a drug in this world. That study was front-loaded against finding a placebo response.

    In my original drafts, I talked about the complexities of this issue. But given the realities of the economy and publishing, my story was cut by at least a third in edit. I still maintain that Hrobjartsson and Gotzche’s first study was not essential to a serious consideration of this work, particularly since, in a later follow-up (http://www.ncbi.nlm.nih.gov/pubmed/15257721?ordinalpos=1&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_Discovery_RA&linkpos=3&log$=relatedreviews&logdbfrom=pubmed), they conceded that they found a “statistically significant” placebo effect in pain trials, though they hedged by saying it might be due to observer bias.

    I took pains in my article to point out that the placebo response is only relevant within certain domains: pain, depression, anxiety, Parkinson’s, and other disorders that engage the higher cortical centers.

    Antibiotics are not failing in clinical trials because of unexpectedly high placebo response. But drugs for the conditions I mentioned are. Two this week alone, in fact, including this one, which made news this morning, and suffered from the kind of geographic variations in placebo response that I wrote about:

    http://bit.ly/3ERNPD

  17. #17 Tom Cooper
    September 4, 2009

    Steve,

    This is Pulitzer quality work. Congratulations on a fantastic, critical, and highly informative piece. For people who are already in the so called mind/body camp, especially those who are open to Eastern thought and complementary medicine, much of what you say is hardly surprising.

    To the blind defenders of a flawed orthodoxy I find many of their criticisms highly amusing. Having studied the philosophy of science and philosophy of medicine I am thrilled to see your article getting the attention it deserves.

    One closing thought. I personally suffer from chronic debilitating pain due, in part, by 2 failed cervical fusion surgeries. About a month ago, I started doing biofeedback, which is helping me to meditate. At the same time I am detoxing off of the prescription morphine to supposedly help with my neuropathy.

    While meditation does not make the pain disappear, it does make it far less debilitating. My doctors, friends, and especially my wife are as blown away as I am as to the positive effects of mind in helping me to finally recover.

    Thank you for a great piece.

  18. #18 Steve Silberman
    September 5, 2009

    Hildy, I wrote a lengthy post, with links, to answer your question about the Hrobjartsson and Gotzche study, but somehow it has not cleared the moderator filter yet, and PalMD seems to have migrated this discussion to another blog of his.

  19. #19 PalMD
    September 5, 2009

    No, multiple links get caught in the spam filter…i’ll go rescue it.

  20. #20 Steve Silberman
    September 5, 2009

    Thanks, PalMD.

  21. #21 David
    September 7, 2009

    Pal wrote: “It is not possible to create strong or weak placebos,…”

    But it is possible. The people who make the best placebos are the homeopaths, chiropractors and accupuncturists. The bigger the mask the witch doctor wears, the better the placebo response. Remember that the patient’s response isn’t just to the pill but to the total intervention.

    And Silberman got it mostly right: for trials with soft endpoints, particularly for things like pain and depression (areas I work in), placebo response in large trials does seem to be increasing with time. Part of this is due to enhanced expectation: now that everybody accepts that antidepressants work, the next trial out enrolls people who from the start think the new treatment will be an improvement.

    And part of it is due to the industry’s behavior. As pressure increases on drug companies to complete trials early, trial conduct suffers. The factors that lead to rapid trial completion and lower budgets also contribute to placebo response. See Irizarry et al, 2009 Clini J Pain25(6):469.

    Pal’s conclusion: “If drugs in testing cannot outperform placebo, then the researches have done a good job of testing the drugs honestly. If the researchers are failing to develop drugs that beat placebo and the company’s bottom line is suffering, it’s not the fault of the sugar pill.”

    Here I strongly disagree. Consider a pain trial. The typical endpoint is the patient’s subjective rating of pain, on a 0-10 scale. If the placebo arm performs well, there’s less room for differentiation of the active arm. Even effective drugs will inevitably fail in a trial with a high placebo response. It is indeed a crisis affecting the industry.

  22. #22 Mr. Pedantic
    September 7, 2009

    Begs the question is a logical fallacy. It does not mean raises the question. You put it in quotes so perhaps you already knew that. However, you went ahead and abused it anyways. From one pedant to another, please stop.

  23. #23 Brian
    September 7, 2009

    You seem to be confusing the ‘placebo effect’ with placebos.

    The placebo effect is highlighted by situations like if you give a bunch of undergraduates non-alcoholic koolaid and tell them it’s spiked, they’ll begin to get drunk as if it were the real thing.

    I mean, you’re right in that we use placebos to try to isolate effects we’re not controling for, but what this article is saying is that the placebo effect seems to be getting stronger. IE people’s bodies are reacting more strongly to the perception that they’re getting the ‘real’ drug.

  24. #24 BAReFOOt
    September 7, 2009

    No shit, sherlock? Guess what: We already knew that, and yet STILL understood the article about the “getting better“. Because we knew what was meant.

    But as a typical white coat, you expect that everyone around you is an idiot, and you’re the all-knowing god!

  25. #25 sam
    September 7, 2009

    David,

    if the placebo arm performs so well that there isn’t room for effective drugs to differentiate then clearly we don’t need those drugs.

    We can treat the pain by doing all the things that the trial was doing with the subjects. Heck give them a sugar pill too if you feel like it, they’re ever so slightly cheaper than whatever the drug being trialled was.

  26. #26 al
    September 7, 2009

    Do you have autism or something? I can’t really see why a person with normal language skills would whine about the statistically verifiable effects of medicines becoming less pronounced relative to placebo (that is, new drugs are getting weaker in a relative sense) being phrased as “placebo is getting stronger.”

    And, as a matter of principle, the placebo response can get stronger. There is absolutely no a priori reason why a placebo can’t be found to be 100% effective at “treating a symptom” in a clinical trial. If it did, what drug could possibly compete with that? Obviously, if there is a real medical condition, the placebo isn’t going to fix it. That doesn’t mean that the drug undergoing tests is ineffective. It means that people’s expectations of how well drugs will work has become an article of faith. This is obviously a “limiting case”, but it ought to get my point across.

  27. #27 Geoffrey A. Landis
    September 7, 2009

    Duh. Yes, of course placebos themselves are not getting more powerful. By definition, placebos have no power. This does not indicate that the people writing the article are idiots and don’t understand what a placebo is; it indicates that the person commenting is failing to understand that despite the fact that a placebo is a drug that does nothing, placebo treatment does have an effect, and the effectiveness of placebo treatment is (according to the article) increasing.

    People who take a placebo think that they are getting a treatment. Look up “placebo effect” in a dictionary.

    The question, given that placebos are pretense-drugs that by design do nothing, is why is placebo treatment getting more effective?

    There are dozens of possible explanations. Ignoring the question is not one of them, at least, not one that has real explanatory power.

  28. #28 Alan
    September 7, 2009

    PalMD, you seem to be failing to distinguish between placebo conditions and no-treatment conditions. The critical difference between placebo and no-treatment is not the sugar pill itself, but the belief on the part of the placebo group that they are getting a treatment. The placebo effect is the difference between the placebo group and the no-treatment group. This excludes effects due to the natural progression of the disease, just being in a study, etc…

    If the difference between the no-treatment group and the placebo group is greater than it used to be, and the outcomes for the no-treatment group have not changed from in the past, then, in fact, the placebo effect HAS gotten stronger. This is entirely possible, since it depends on the belief of the participants, which may very well be different now than in the past.

  29. #29 Mr. Critical Thought
    September 7, 2009

    @Mr. Pedantic

    I think you need to re-read the section you are referring to. PalMD is correctly using the phrase “Begs the Question”.

    His previous sentence:

    …the insanely nonsensical claim that “placebos are getting better”

    PalMD is asserting that to claim “placebos are getting better” you must assume that placebos are effective in the first place. Which is a point he feels has not been proven. This is clearly a proper use of the phrase “Begs the Question”.

  30. #30 Ivan
    September 7, 2009

    Gotta agree with BAReFOOt. Thanks for pointing out the blindingly fucking obvious.

  31. #31 David
    September 7, 2009

    Sam(#25) – you’re trying to directly translate what happens in a trial, as measured on artificial rating scales, into what happens in real life, where the measurement is on softer, multidimensional perceptions. That’s a very flawed approach.

    A short-term trial can be done in by placebo effect but that doesn’t mean the condition is adequately treated in real life. Diabetic neuropathic pain is an excellent example. It’s a major clinical problem. In the paper I referenced above, 3 trials showed placebo effect that would translate into satisfactory treatment response in the real world. The problem is, it’s only maintained for the short term of the trial.

  32. #32 Mr. Pedantic
    September 7, 2009

    @Mr. Critical Thought

    Yes, I was wrong. It seems he was using the phrase correctly. I just misunderstood since there is not any actual question begging going on. The placebo effect implies an effect. Anything that has effects is effective. QED.

  33. #33 William Johnson
    September 7, 2009

    Although I’m not a big fan of Wired, I thought the article was reasonably grounded and that PalMD’s critique was pedantic/semantic nitpicking. It’s common for people to get hypercritical if they feel that somebody else is invading “their” field of expertise.

  34. #34 Dave
    September 7, 2009

    Maybe it’s a conspiracy. Merk pays people to infiltrate Pfizer trials and make false reports, and vice versa.

    Can double-blind tests make this impossible?

  35. #35 RM
    September 7, 2009

    It is not possible to create strong or weak placebos

    Actually, it is. There have been studies (sorry don’t have them at hand) which showed things like patients report better outcomes when given a red sugar pill vs. an otherwise identical blue sugar pill. Or a “treatment” is more effective when administered by a person in a white lab coat than one without.

  36. #36 Steve Silberman
    September 7, 2009

    Yes, RM, that’s mentioned in my article too. See “Study on the Effects of Tablet Colour in the Treatment of Anxiety States,” Schapira and McClelland, as well as “Commercial features of placebo and therapeutic efficacy,” Waber.

  37. #37 Alex
    September 7, 2009

    There are people here saying that PalMD is “pointing out the blindingly fucking obvious” and others saying that he is just plain wrong. IANAD, so which is it?

    I can’t see how PalMD is “pointing out the blindingly fucking obvious”. This isn’t a medicinal journal, this is a science blog read by many a layman such as myself. None of this is “obvious” to me. Maybe you guys need to go take a non-placebogenic chill pill and remember that not everyone understands stuff that is obvious to you. Besides, no-one made you read this blog post, and this is the internet, people can write whatever they want (mostly), are you going to complain every time someone writes something “obvious” to you?

    I assume even you guys had to learn about the placebo effect before you understood it anyway. You weren’t born with this knowledge were you?

    And as for the guy who tried his best to insult anyone autistic who comes here to read… wtf? That’s a bit uncalled for.

    Even if PalMD is correct and he was “pointing out the blindingly fucking obvious”, so what? It’s called debunking, as clearly it wasn’t obvious to the writer of the Wired article. I’m fairly sure anytime someone debunks whatever stupid Michelle Bachman is spewing, they have to resort to “pointing out the blindingly fucking obvious”, but that doesn’t make it wrong to “pointing out the blindingly fucking obvious”, it’s what needed to be done to challenge the misinformation out there.

    Whether or not PalMD is correct, I dunno, as as I said, IANAD.

  38. #38 hibob
    September 7, 2009

    Steve – Has there been a serious analysis about how the requirement of public registration of clinical trials may have “increased” the placebo effect? I’m not saying that all trials are properly registered and monitored (less than half the time, apparently), but at least there are more rules in place to cut down on some of the more blatant data manipulation. If a trial is properly registered before data starts coming in, and is then properly monitored after data starts coming in so that if the original goals aren’t met they can’t just be swapped for softer ones, A lot more drugs will fail to beat placebos. It’s also harder to bury an inconvenient result now than it was in the past. Back when Prozac was at stage III, they dropped several trials from the application because they showed “anomalously” high placebo effects.

  39. #39 Steve Silberman
    September 7, 2009

    My article has been “debunked”? That’s news to me. I’ve found the comment thread of this post very interesting. Many of the other commenters provided pointers to research that in fact I had mentioned in the article, as would be obvious to anyone who read it. I would agree with Peter that I could have hit the reader over the head more forcefully with the crucial distinction between a “true” placebo response (not measured in clinical trials without a no-treatment arm, as I explained above) and the statistical noise derived from placebo control groups in trials. But Peter’s evident belief, when he made his original post, that there is no way of talking about a true placebo response — and that placebo response exists ONLY as an artifact of clinical trials — was not accurate either, as many people here pointed out, and the research of people like Benedetti, Wager, Zubieta, Leuchter, Moerman, et. al. shows. I appreciated Peter’s criticism, but this thread hasn’t been a huge gobsmack for me. I think we all learned things reading it.

  40. #40 Steve Silberman
    September 7, 2009

    > Has there been a serious analysis about how the requirement of public registration of clinical trials may have “increased” the placebo effect?

    Not that I know of, but that’s an interesting notion.

    I assume that’s one of the things the drug companies will be looking into during the Placebo Response Drug Trials Survey, which I talk about in the piece. Though I also suspect that one of the main goals of that project is to look for biomarkers of high placebo response, so that people who have them can be excluded from future trials — which opens up many other cans of worms.

  41. #41 Cyde Weys
    September 7, 2009

    I’d just like to throw my support behind the article of the Wired piece (hey Steve, and thanks for engaging with the online discussion over this, both here and on Slashdot). A lot of what he spends time talking about are psychoactive medications, which as he points out, are MOST likely to be affected by how well you think they’re going to work.

    And, even just over the past decade, psychoactive drugs have penetrated far further into the mainstream, both in aggregate number of people taking them versus now as well as general public awareness (such as all of those ads on television). A lot more people being knowledgeable of the drugs, and being convinced of their efficacy, could easily increase the level of placebo response beyond the meager effects of the first generation drugs.

    Sorry PalMD, but your response just comes off as too hair-splitting and over-reactionary, and frankly it looks like you didn’t read the article in its entirety, or at least not with the aim of trying to understand what it actually said versus with the aim of finding ways to attack what you seemed to think it said.

  42. #42 PalMD
    September 7, 2009

    Alex, I don’t think any of this is blindingly obvious, which is why Steve could write an article about it and I could write a critique of his piece. It’s also why there is a lively discussion going on between Steve, me, you, and the rest of the readers.

    These issues are complex, definitions are understood by different people differently, people have firmly held beliefs—there’s no end to the complexity here, so frustration is pretty much assured.

  43. #43 Steve Silberman
    September 7, 2009

    This has been a great discussion. Thank you Peter, and everyone who commented here.

  44. #44 anon
    September 7, 2009

    If what he says is true, then what is happening is that people are becoming more and more expectant of real results from taking drugs.

    They’re psychosomatically inducing themselves into feeling results from these trials because in this decade, people must feel they have a better understanding of what medications are out there and what they do.

    They don’t fear them or shy away from learning anything about such an area of knowledge that people in the past perhaps might have thought only a doctor would concern himself with.

    If Prozac isn’t showing clinical effectiveness against a placebo group, then perhaps the entire concept should be reconsidered. As we already know that Prozac does in fact work.

  45. #45 seriously...
    September 7, 2009

    Funny things, every serious long-term anti-depressant study shows that in long-term use theres no difference from placebo to the real drug…

    OK want analogy, everybody knows that on long-run Drinking Booze won’t help solve your issues… well guess what, what would another chemical’s do? yep bring the magic and fix your life… (Shees?)

    Medical companies are allowed to publish data, then buy from some poor pissed university with those bs quota’s for articles one has to publish in year to stay in job…

    so who the fuck is going to conduct real controlled experiment, when your actually just “conducting mind blowing research” by writing n+1 articles in year?

    And now they have come up with idea of Increased placebo effect in Double-Blind test? seriously, post these news to any hifist forums and they join physics forums and start spamming people with “FACTS”

    i suggest you take your thinking cap’s and place em on head…

    think about that certain “substance’s” might increase or decrease with some genetic/hormonal combination placebo effects, but then it can be ratified and analyzed.

    Other alternative just reading about placebo and believing it makes it work even BETTER!! thus more knowledge you acquire about it(as it is mental process, if this is the case) one can gain infinite effects from placebo so all the stories of people living 10years with just staring at sun must be true if they just believe they can do it, then they can do it…

    Face it: worst case scenario of what happens when money decides facts and really pathetic attempt to cover-up wide incompetence.

  46. #46 seriously...
    September 7, 2009

    Oh, yeah even this only affects like anxiety&depression… not living on sunlight, it’s still pretty good joke :D

  47. #47 hibob
    September 7, 2009

    >I also suspect that one of the main goals of that project is to look for biomarkers of high placebo response, so that people who have them can be excluded from future trials — which opens up many other cans of worms.

    No doubt. If a drug’s efficacy is only statistically significant by removing patients with placebo biomarkers from the clinical trial, I think the FDA would only be able to approve the drug for patients without the biomarkers. That should be fun to bring to market.

    While it makes sense for Pharma to look for ways to exclude placebo effects from its trials, it makes even more sense for both therapists and Pharma to try to ferret out the mechanisms and try to exploit them. If the placebo effect could be brought about reliably and without deception through therapy, a lot of people could get off a lot of meds and a lot of snake oil/woo. A drug that is an actual agonist for a placebo-mediating mechanism or could replace it outright would be pretty cool too.

  48. #48 rusl
    September 7, 2009

    “it’s not “placebo” that’s the problem”?

    From this article it seems like you fundamentally don’t get it?! Placebo is a real phenomena. That is why we name it. It is different every time. It isn’t an object. The mechanism is ultimately unknown by definition – otherwise we would name it differently.

    Placebo is the problem: inasmuch as there is one – which there is not. The point of the wired article is that this is a problem in drug tests. Problems are whatever we define them as. Your the one with the unscientifically biased value judgement that says placebos cannot be a problem. And of course from a scientific ethics perspective you are right to say so. It is wrong to view the placebo as the problem rather than to focus on improving the drugs being tested. Yet that is only true for a perfect placebo – something I’m sure you would point out to be impossible and absurd. Statistically at least, this is a widespread phenomenon so maybe there is something else going on here? Maybe it is something that has changed in testing methodology? Maybe it is something about this generation of drugs? Maybe it reflects a wider social change like people being more responsive to medical attention overall or something else entirely? But your overall conclusion that we not look at or consider this scientific observation isn’t enlightening to anyone – though it certainly opens up dialogue.

  49. #49 bluesixer
    September 8, 2009

    Hmmm…I read the Wired piece, but I took the “better placebo” rhetoric as low-key snark directed at not only the spectacular R&D failures of the last decade but also at the insufficiently researched and ultimately ineffective treatments that made the industry so ridiculously profitable.

    But now that I think about it – odds are that snark wasn’t the author’s intent… Pity.

  50. #50 Ellie
    September 8, 2009

    Just wanted to add my voice to the many saying I read the original article before reading this so called “dismantling” of it and found Steve’s writing to be of the highest quality and very interesting. I have found nothing here to convince me that he’s wrong and have been left feeling rather insulted by the implication that I can only have fallen for the woo because I’m stupid.

  51. #51 PalMD
    September 8, 2009

    Ellie, the problem with Steve’s piece is not that it isn’t well-researched or interesting—it’s that it perpetuates a fundamental misunderstanding.

    The bump in outcomes seen in placebo group vs no-intervention group vs. test drug group could perhaps be attributed to “stronger placebos” and what that might actually mean is an interesting question. But much more likely is that something else changed rather than the inert treatment.

  52. #52 Greg Laden
    September 8, 2009

    This is my overlong contribution to the discussion: What is the Placebo Effect, and it it getting stronger?

  53. #53 TTabetic
    September 8, 2009

    You, Elie, try this link

  54. #54 Steve Silberman
    September 9, 2009

    For the record, Ellie, I disagree with Peter’s summation of my piece as somehow perpetuating a “misunderstanding” that somehow sugar pills are getting “stronger.” He may have gotten that from the headline (which I didn’t write) and the mystery I talk about in the lead paragraphs, but if that was supposed to be my point, I sure wasted a lot of time talking about problems with the design of clinical trials, changing populations of trial subjects, inconsistencies with depression rating scales, the changing cultural dynamics of trials conducted offshore, a focus by drug developers on diseases that engage the higher centers in the cortex… and all the other possible contributing factors I mention. Peter may have his own ideas about what the “something else” was that has changed, but my article offers you the perspectives of scientists who have been studying precisely these questions for a very long time.

  55. #55 Steve Silberman
    September 9, 2009

    > odds are that snark wasn’t the author’s intent… Pity.

    Bluesixer, snark isn’t the only way to make the points you’re making. Snark is not my preferred mode as a journalist, though I realize it’s quite popular these days, particularly in Wired. But that doesn’t mean I don’t see the same dynamics you do.

  56. #56 Alex
    September 9, 2009

    My article has been “debunked”? That’s news to me.

    That’s not what I said. I said “if PalMD is correct” then this post is a debunking. As I also said, IANAD, so I don’t know if he’s correct or not. I was specifically addressing that part of my comment to those who thought PalMD was “pointing out the blindingly fucking obvious”. If those people are right, then your article was debunked Steve, however, as I say, I don’t know if they’re right, but there are some more comments to read so we’ll see if there’s any that make it clearer.

  57. #57 PalMD
    September 9, 2009

    I don’t think the article was in need of “debunking” but the impression left was (which was apparently not the intent) that placebos are getting stronger, which implies a common but erroneous misunderstanding of “placebo”. This thread runs through the piece and was perpetuated by the people quoted.

  58. #58 Steve Silberman
    September 9, 2009

    Not sure how the notion that sugar pills themselves are getting stronger is perpetuated by trial manager Arif Khan making an observation about the quality of care in control groups, or Fabrizio Benedetti discussing his experiments with Alzheimer’s disease and analgesia, or Tor Wager talking about areas of neural activation triggered by the act of taking a placebo, or William Potter admitting that his initial data search at Lilly could not determine the reason for so many drugs failing in trials. Those are the people I quoted, and they’re each leading scientists in their fields. But as I said at the beginning of this thread, readers can and should judge the article for themselves.

  59. #59 pariah
    September 9, 2009

    PalMD, not all placebo trial are double blind as implied in your arrogant rant.

  60. #60 RonK
    September 9, 2009

    PalMD, if you dismiss the placebo response as being artifactual, do you also dismiss the nocebo response in the same way? Assume there exists a hypothetical medical condition X from which 1% of the affected patients recover and the other 99% are condemned to suffer a horrible, painful death with no possibility of anesthesia. Would you inform a patient which you diagnose with condition X about the excruciating details of the suffering he is most likely to experience, in the name of making said patient better informed (assuming that the patient hasn’t explicitly asked for this information)?

  61. #61 WotWot
    September 13, 2009

    Little late to the party, but what the hell…

    That study [Hrobjartsson and Gotzche] was front-loaded against finding a placebo response.

    I do not agree with that at all. H & G did a pretty clean initial study, and follow ups. Not perfect, but pretty good. More than solid enough that they have to be taken seriously, no matter how “highly controversial” (read: ‘unwelcome’) the findings may be.

    If the placebo response was real and significant, it should have show up clearly in that meta-analysis. Not maybe, not a small possible effect for pain alone, but a definite effect for a range of conditions and symptoms.

    But it didn’t.

    That is a huge problem for those claiming the placebo effect is real and substantial.

    As it currently stands, at the very least the size and ubiquity of the placebo response is under serious question.

  62. #62 Stan Polanski
    September 14, 2009

    I’m one of those chronic lurkers who only comments when he just can’t help it. This excellent thread has nudged me, a placebo-effect skeptic, over the response threshhold. I have two thoughts I want to share about placebos.

    First, as a medical practioner, it mystifies me why any physician should care whether placebos have a substantial effect. Any benefit produced by a white coat, an empathetic attitude, or a sugar pill must be additive to the actual effect of a treatment. I only care about the difference. If all you have is the placebo effect, you’ve got nothing (or nothing that you can, in good conscience, charge for.)

    Second, I want to suggest a hypothesis, not previously mentioned in this thread, to explain the apparent increase in the placebo effect in clinical trials for conditions such as pain and depression that involve subjective reporting of the severity of symptoms. Subjects in clinical trials are “grading” their physicians much as teachers grade students, and I suspect that scores on pain scales and depression scales are similarly subject to grade inflation. Any of you who have had children in high school in the past decade or two will have noticed that a mere 4.0 GPA no longer separates the highest achievers from the also-rans. I suggest that patients in trials are subject to the same cultural forces that lead teachers to give A’s for work that would have rated a B twenty years ago. A previous comment in this thread noted that there is more room for improvement in the score of a placebo than for an effective treatment; therefore, any confounding factor (such as my hypothesized grade inflation) that increases scores will tend to narrow the gap between treatment and placebo.

    I know I’m poking my nose into this discussion long after most of the participants have moved on, but I would still appreciate a reponase or two. Hello?

  63. #63 PalMD
    September 14, 2009

    Someone’s still listening.

  64. #64 ktbug ladydid
    September 22, 2009

    Steve Silberman, I find your image for the wired article not only wrong, but offensive, and terrifying. Placebos are not as good as actual medication. If you really think that, there’s something wrong.

  65. #65 needs info
    September 22, 2009

    Wired reports an interesting phenomenon but it goes on too long and doesnt say anything interesting. Exactly how much more placebo is there. What would Prozac score on an RCT.

  66. #66 giantslor
    November 20, 2009

    Geez. I’m a layman but I understood Steve’s article perfectly, and I did not get any wrong impressions about placebos. I was going to pass the article along to my mom, but now I see a lot of commenters and the OP seem to have misunderstood Steve’s article, so I might end up with a confused mom.

    Commenter ktbug ladydid takes the cake. “Placebos are not as good as actual medication.” You didn’t read the article, did you? Pharmaceutical trials left and right are producing results where the placebo is as good as the “actual medication.” How, then, is the placebo not as good as the medication??? The article also explains exactly how the placebo effect is getting stronger.

    Reading Steve’s article and this thread actually make me more sympathetic toward alternative medicines, many (most?) of which work by the placebo effect. The more the patient trusts the particular alternative medicine, the better. I could actually get behind the idea that it’s ethical and deserving of payment to create an elaborate ruse that fools the body into healing itself. But I’d only support it for those neurologically-related ailments that the body can actually treat, not everything under the sun like homeopathy claims. And only in cases where no better traditional medicine exists.

  67. #67 Truman Green
    March 22, 2010

    The article in Wired by Steve Silberman would appear to be reasonable and PalMD’s nitpicking interesting, but I think what is going on here is that Silberman’s claim that people are more susceptible to the placebo response, has the subliminal effect of treating the drug manufacturer’s efficacy trials as though they have been based on real science and not just the junk science that should be the real description applied to them.

    Think about it, folks. If you can make billions of dollars by tweaking trial results and all your pals are running the governmental agencies, well….must I say more?

    Here’s the point–made by several commenters on several sites–but strangely never by one of the writers.

    So a candidate drug beats a placebo by 20%. So what? this beats an artifical P-value standard, maybe, but that comprises its totally usefulness. This means that 80% of the subjects received no benefit, because the placebo effect must be subtracted from the effect of the candidate drug to arrive at a net effect.

    Drugs can be approved if they out-perform a placebo by as little as 5%–which, for you more unbrainwashed people, will obviously mean that 95 percent of people will get no benefit.

    What else would you buy that you knew didn’t work for 95% of the population?

    A car? A dog? A gun? A hotdog? Condoms? I don’t think so.

    So this is the main scam that the pharmascorpians and their pals in the governing agencies and media are running:

    The truth is that very few drugs–besides antibiotics, insulin, immunesuppressors, birth control pills and some opoid-mediated painkillers like morphine, work anyway.

    And Steve Silberman is doing his job–which is to legitimize efficacy-testing of drugs (by taking it seriously), when it’s mostly a fantastic hoax.

    I bet the pharmas initiated all these stories about the placebos being more effective.

    Do any of you people really believe that human nature has changed that much in 25 years or so?

    For example, I just did a six-month review of the efficacy testing of influenza vaccines and was totally shocked that anyone would believe that these efficacy tests have anything to do with medical science, when the the drug companies have merely invented surrogate markers for efficacy such as how many antibodies they can count in the blood of subjects–not how many people actually get a clinical case diagnosis of influenza after they are vaccinated, and how many don’t. And even worse, they pick a scapegoat surface glycoprotein such as hemagglutinin, as the main infective antigen, when there’s no science to support this. But who’s going to complain? Which science guy is going to risk his Beamer and his mortgage to point this out?

    So, it’s all 100% junk science.

    Beating placebos by a few points is just another self-serving surrogate efficacy standard. In reality, it means nothing, except that the candidate drugs don’t work and would never be approved if the governing agencies hadn’t been infiltrated for the last 60 years or so.

    No, the placebo effect isn’t getting stronger–The pharmies have just changed their tactics. Let’s blame the placebos, guys.

    Besides, where’s all the metaanalysis studies that would be required to show this trend?

    I can’t find any on PUBMED. Where are all the metaanalyses published in medical journals?

    I think, while there may be rumours about the increase in the placebo effect, there is no actual science.

    The studies presented by Silberman are all about how the placebo response is not as simplistic as once thought–even different people in different countries experience it differently–as if that’s some kind of revelation.

    Come on, guys, is this really news. You mean it never occurred to the pharmies before that a New Yorker would experience the placebo differently than a Maori, an Eskimo or a Masai?

    Or that drugs which were merely winked into approval 30 years ago are being subjected to a bit of actual scrutiny today?

    Interesting also that there are no actual numbers given for failure rates of specific drugs such as Prozac. We wouldn’t want to let the cat out of the bag regarding how much such drugs out-performed placebos when they were first approved.

    And who approved acetaminophen, anyway? And Vioxx? Did Vioxx beat a placebo? I bet!

    Alix Spiegel of NPR begins a story entitled: “The Growing Power of Sugar Pills” like this:

    “The other day I came across a fake news story on the internet. It was a send-up of the pharmaceutical industsry which featured a bunch of drug industry executives wringing their hands in despair: placebo pills, the fake news reported, were getting stronger. What’s a drug executive to do?”

    Anyone know what story Siegel is referring to? Sadly, she doesn’t tell. And Google doesn’t seem to know.

    Truman Green
    Surrey BC