Skip to main content

Clinical Trials

The Rising Placebo Effect

The placebo effect has many interesting and annoying features, among them the way that it varies so much among different therapeutic areas. There is no placebo effect for a broken leg, or for pancreatic cancer: these things are going to play out the way that they do, regardless of what you think about them. But for the pain of a broken leg, there you might have something. In general, the things tied most closely to self-awareness (pain, mood, etc.) show the strongest placebo effects.

Now a new analysis of clinical trials for pain medication shows that the placebo effect in this area has been getting stronger. (Here’s a Nature News article on the subject). The same also seems to be true for antipsychotics and antidepressants, but this effect seems to be mainly (or only) visible in large-scale US trials:

“We were absolutely floored when we found out,” says Jeffrey Mogil, who directs the pain-genetics lab at McGill University in Montreal and led the analysis. Simply being in a US trial and receiving sham treatment now seems to relieve pain almost as effectively as many promising new drugs. Mogil thinks that as US trials get longer, larger and more expensive, they may be enhancing participants’ expectations of their effectiveness.

That certainly could be it. People may also be raising their expectations of what modern medical science is capable of, and figure that in these days of modern times (as the Firesign Theater used to put it), new drugs must be quite a bit better than the old stuff. In a way, they’re right – that Jack Scannell article I linked to yesterday talks about what he’s called the “Better Than the Beatles” problem in drug discovery. That’s the competition we always face against our greatest hits, effective medications that have gone generic and are now cheap and well-proven. For simple pain relief, it’s not easy to beat the likes of aspirin, ibuprofen, naproxen et al., so if you’re going into that market, you’d better have something good. Unfortunately, the harder-to-treat forms of pain are really, really hard to treat. And if you’d like to treat them without the possibility of addiction, then that’s just too bad. There’s not a heck of a lot between ibuprofen and opiates, although not for lack of trying.

So people may be expecting more out of pain medication trials, but they’re not getting it. This new study found that against neuropathic pain (a large and terribly underserved market), the drugs studied over the last 23 years have all been about the same: not too effective. But the placebo responses in these trials have been creeping up the whole time – well, the placebo responses in the US. Asia and Europe have been flat.

That suggests placebo and drug responses may not always be strictly additive. This isn’t entirely unexpected, Mogil argues, because both placebos and pharmaceutical painkillers tap into similar biological mechanisms — such as the release of endorphins in the brain. But if true, it suggests that growing placebo responses are masking real painkilling effects. “There are a lot of people in the pain field who believe the drugs that are failing clinical trials actually work, it’s just that the trials can’t show it,” he says.

Hmm. I see where he’s coming from, but I think that’s a dangerous line of thought. If they worked really well, we wouldn’t be having this argument at all. If you start throwing out clinical trial results because the drug “actually works” anyway, where is there an end to it? I suppose one could draw some sort of placebo-control cutoff and say that if the placebo group responds to some high level, then the drug. . .no, actually, I still can’t see it. I’m going to remain an absolutist for now: if your drug can’t beat placebo, then there’s no way to say that your drug does anything for patients.


34 comments on “The Rising Placebo Effect”

  1. Hap says:

    1) If you think it works, and it doesn’t matter what the data say, what was the point of the trials?

    2) If not working better than placebo is success, then why use something expensive that works no better than placebo? Why not use something cheap? Of course, someone else might think differently (

    1. Jeff says:

      I’d actually love to try this. My daughter has been having mild stomach aches for a couple months. Doctors think it’s diet related. Wife thinks its Gluten (because… you know… Gluten).

      I’d love to have a little placebo pill I could give her “from the doctor”. It would be worth it if it “cures” what I already think is largely in her head.

  2. dearieme says:

    Some cruel and heartless person will propose that this is evidence that Americans are uniquely hysterical and suggestible. Whereas a cooler inspection suggests only that they are becoming increasingly hysterical and suggestible, which is a plain different thing.

  3. Michael says:

    Is there any chance that regression to the mean plays a role? ie, patients tend to enroll in trials when their pain is at its worst (and thus likely to get better)? Or have trials been done comparing placebo to “no intervention”?

  4. Morten G says:

    Oh, me me! I have theories galore.

    1) Prescription strength painkillers are much more easily available in the US than Europe/Asia. And the people in the placebo group are likely to go take some. Patients lie.
    This hypothesis is of course shot down if the clinical trial includes blood tests to ensure that the patients don’t take any other painkillers.

    2) US Americans are better entertainers and more charismatic. The US school system includes a bunch of show-and-tell and debate teams and it’s generally in the culture to charm people. The US doctors might just be better at convincing the patients that the placebo is going to work.

    3) European / Asian doctors might be better at explaining that while the patient enrols in a trial for an experimental drug, there is a good likelihood that the patient does not receive any of the drug. Or US American patients might have a harder time wrapping their heads around the concept.


    Test patients for drugs and medications that might interfere with the outcome.
    Perform double-blind head-to-head trials instead of placebo controlled.
    Recruit patients on the autism spectrum since they lie less.
    Crossover trials so that all patients are both on placebo (or the competition to beat) and the experimental drug.

    And it is not completely dumb that a strong placebo effect obscures efficacy. If you get 100% response in both placebo and intervention group then you can’t say if the drug did anything or not. Doesn’t mean that it should be approved. If anything the placebo should be. If you still believe in the drug then you should redesign, resubmit, and redo the trial.

  5. DCRogers says:

    Maybe it’s because American doctors *underprescribe* placebos?

    “Half of all German doctors prescribe placebos, new study shows”

    “Placebo cures shown to help with depression and stomach complaints – in Bavaria, 88% of doctors have prescribed them”

    And a second question, asked quite seriously: since placebos have biological effects, shouldn’t they also be studied to identify negative side-effects? For example, perhaps people on placebos for depression are at a higher risk of suicidal ideation?

  6. anonymous says:

    Ariely figured this out a long time ago, in a controlled study. The phenomenon is not limited to medicines, though it was demonstrated in a study using pain meds.

    If Ariely is right, the more expensive a placebo, the more effective it will be. Perhaps instead of controlling trials with placebo we should control them with aspirin or ibuprofen. If the prescription drug can beat the inexpensive OTC pain killer, it works. Based on Ariely’s work, the inability to beat a placebo that is perceived to be high value isn’t the same as not treating a patient at all and is superior to treating with an inexpensive placebo.

    Perhaps the driving force for efficacy here is that we have trained our society to expect that new drugs will be astronomically expensive.

  7. Me says:



  8. Vader says:

    “Simply being in a US trial and receiving sham treatment now seems to relieve pain almost as effectively as many promising new drugs.”

    This could mean that the promising new drugs aren’t actually all that promising, no?

  9. anon says:

    Did you see that Valeant has been subpoenaed for price gauging?

  10. Scott says:

    Isn’t the obvious answer just to relocate clinical trials outside the US if you’re worried about not beating a strong placebo effect?

  11. J.S. Mill says:

    Aside from academics, who cares whether the drug hits it’s target as long as the placebo has the exact same phenotypic response? We are in the business of maximizing utility of the patients and, if placebo proves to be the opiate of the masses, let them eat sugar pills.

  12. Mark Thorson says:

    It’s been known for a long time that the color of a pill affects its placebo response, which is why placebo-controlled trials are always performed with placebos that look exactly like the active medication. I’ve wondered whether the aroma when you take the cap off the bottle also affects the response. If we really understood all of the factors, we might be able to make much more powerful placebos.

  13. old man says:

    Just run a pre-trial using only placebo, then exclude all the responders from the real trial which would still be placebo controlled. This gets rid of the highly suggestible patients.

  14. Sci-Fi Geek says:

    This could present interesting ethical dilemmas, as explored in another Nature publication:

  15. Magrinho says:

    Would extortionately expensive, generic placebos be effective?

  16. Anon says:

    No, the placebo effect is not getting stronger; we are just testing drugs with smaller benefits that are barely better than placebo. Diminishing returns. That’s exactly what you would expect with incremental innovation.

  17. Sara Rae says:

    What if you test three “drugs” in pain trials? If you test the placebo, OTC meds, and the new drug blindly, perhaps you would remove some of the mental strength of the placebo.

  18. Nelson says:

    Placebo is doing better because American research sites provide food, cabs, and entertainment to participants in order to encourage them to show up for visits and not complain. The more pleasant the experience of participating in a trial, the stronger the placebo effect. European clinical trial sites for psychiatry and pain are far more likely to be academic or hospital-based and not allow these extras.

    Another aspect of this is selection bias: CROs routinely exclude countries from studies if placebo is doing too well there. CROs never exclude U.S. trial sites based on placebo differentiation.

  19. a. nonymaus says:

    Re: Anon of 15.10.15, 11:19 a.m.

    If true, these new medications would fall below the standard-of-care, which presumably beats placebo. Unless you are referring to conditions where there is no existing medication, in which case it’s not incremental innovation but going after what’s left once the easy problems are solved. As an example, neuropathic pain versus the acute traumatic pain of surgery.

  20. JFT says:

    If the best we can do is the same as a placebo, then it is not a useful intervention. Yes, comparing to tylenol might be a good idea-and should perhaps be used as an additional control group.

    Pain is a bitch. Both for the people who suffer from chronic pain, and those who try and target it.

  21. Vader says:

    @Matt Thorson:

    I have sometimes speculated, half-seriously, that almost all antidepressants are are really placebos that are hard to properly double-blind because they have nastier side effects than a sugar pill.

  22. Mark Thorson says:

    There was a trial of St. John’s Wort in major depression a few years ago that found it didn’t beat placebo, but it also had an active control arm using setraline (Zoloft) and that didn’t beat placebo either.

    This trial elicted a lot of comment at the time.

  23. NJBiologist says:

    @Michael: “Is there any chance that regression to the mean plays a role? ie, patients tend to enroll in trials when their pain is at its worst (and thus likely to get better)?”

    It’s actually worse than that in a lot of pain trials. Many trials have a cutoff: patients rating their pain below the cutoff are excluded, on the reasoning that higher scores will better show efficacy. However, the enhancement of regression to the mean seems at least as likely.

    It will be interesting to read the article and see if any attention was paid to the changes in trial design seen over the past few years.

  24. MoMo says:

    Great. Another reason physicians will deny pain Meds to cancer patients. I can hear it now “You don’t get actual pain Meds- Placebos work just as well- Suffer!”

  25. Anon says:

    @a.nonymaus: Both, in fact.

  26. Ben T. says:

    Anyone know what happens if you run a trial where you do your best to convince both groups that they’re taking sugar pills? You’d have to make a good show of it because people could still suspect that it’s a ploy and they’re getting the drug. But I wonder if you’d see more spread between an effective medication and “placebo” when they’re both fighting against patient skepticism instead of patient optimism.

  27. Anon says:

    @Ben T. – Good idea in principle, as it could reduce baseline noise, though recruitment and retainment rates may be rather low…

  28. Rock says:

    This is why people we consider quacks pushing homeopathic medicine, herbal supplements, acupuncture and the like can claim their treatments work because often they due entirely through the placebo effect. The effect is not purely psychological either, since they have demonstrated measured increases in endorphins in pain studies using “sham” acupuncture. Did not know European doctors prescribe placebos. Interesting. Wonder what they charge.

  29. placenta says:

    If you try to make a drug placebo proof then they’ll just build a better placebo

  30. Kaleberg says:

    I don’t think OTC placebos would work for the same reasons that you can’t tickle or startle yourself. I’m sure have to be prescribed by someone with a suitable level of authority. Does this work through the neocortex?

  31. tangent says:

    “standard-of-care, which presumably beats placebo”

    Well, presumably *did* beat placebo when it was approved. Does it still?

    If placebo is really strengthening globally, maybe it’s caught up. Or, maybe the placebo effect weakens as the drug in the trial gets old and boring.

    1. Oliver H says:

      Well, it the placebo effect increased, you’d normally expect it to add to the standard of care as well, at least when there IS a placebo effect to begin with. Often overlooked is its evil twin, the nocebo effect. Wonderful thing, the mind: If you believe you’re getting pretty strong chemotherapeutic drugs, you might just develop nausea, vertigo and blood pressure problems even if you’re just getting saline solution…

  32. Thomas says:

    My attempt would be: suggest everybody they are on OTC medicine for the first week, and then to the test medicine the next week.
    In reality, give – for the full 2 weeks of the trial – half the population the OTC/placebo, and the other half the new stuff.

Comments are closed.