Skip to main content

Clinical Trials

Expensive Shams Are the Way to Go, Apparently

Prepare to be weirded out a bit. We’re going to talk about the placebo effect again – actually, we’re going to talk about its evil twin, the nocebo effect. In the same way that a placebo is an inactive/nonexistent agent that people think is doing them good, a nocebo is one that people believe is doing them harm. For example, in every double-blinding trial of a new drug, there will be people who drop out of the study because of side effects. But when these patients are unblinded, some of them will turn out not to have been taking the drug. You can see this in the figures at the end of the trial, when the major side effects are reported, and it’s then mentioned whether these were significantly different than the control group. Classic nocebo symptoms are headache, tiredness, digestive/stomach problems, and hard-to-localize pain, but there are others. To complicate matters, any of these can also be a real side effect of a real drug as well.

It’s already known that the placebo effect is stronger when people believe that they’re taking a more expensive treatment. This is exactly the same effect, you’d have to think, as the tendency to rate an unknown wine as more enjoyable when it’s presented with a higher price tag (and if you think that wine merchants – and restaurants – are unaware of that latter effect, you need to adjust your priors, as the Bayesians say). So you have to wonder if the nocebo effect also strengthens as a result of perceived price, and this new paper says that it does.

The authors (from Hamburg, Colorado, and Cambridge) have developed an NMR imaging technique that lets them monitor the cortex, brainstem, and spinal cord simultaneously, which covers a lot of the pain-sensing machinery. They then took a skin cream with no pharmaceutical activity at all and packaged it two different ways – one batch went into a cheaper-looking box with generic orange printing, and the other went into a fancier-looking blue box. Preliminary studies showed that people (not in the later study) estimated that the former definitely cost less than the latter, and in the actual study, they reinforced that by references to a “cheap” and an “expensive” cream for the two test groups. Each group also had a control skin cream, but in reality, all three of them were the same.

The subjects were told that they were testing a cream that was prescribed for dermatitis, but which had been reported to make people more sensitive to heat pain sensation, and that the purpose of the study was to follow up on reports that the two creams produced different levels of this effect (but not in which direction). They then applied the “cheap” or the “expensive” cream to a patch on the patients’ arms, along with another patch of the “control” cream, and then used a heat-generating skin testing device to establish each patient’s scale of pain sensation (low, medium, and high). The patients were then told that they would receive that same amount of heat on each patch of skin, test and control, but in reality they got more heat on the “test” patch, to reinforce the idea that the creams could produce heat sensitivity. In the next phase of the experiment, on a later day, the same test-and-control patch method was used, and the patients went into the NMR scanner. The heat-sensation scale was checked again while no imaging data was collected, then they went in for imaging and (supposedly) comparisons between the skin patches. In reality, they were getting exactly the same medium level of heat in each case.

And sure enough, the people who thought that they were getting the expensive cream showed significantly more pain sensation in the “treated” patches than the people who thought that they were getting the cheap stuff. (The belief is that people tend to associate the expensive one with newer, more advanced, and more potent drugs, while the cheaper substance are older and generic). The fMRI data were quite interesting. They identified a number of regions of activity, which corresponded well with previous imaging studies on pain sensation. The periaqueductal gray (PAG) showed a significant difference in the “expensive” cream treated patients as compared to the “cheap” stuff. Some regions in the frontal cortex and amygdala also showed differences. It turns out that the regions involved in the nocebo effect are quite similar to the ones involved in the placebo effect in earlier studies, suggesting that the same mechanisms are at work, in the end. The PAG, for example, is activated whether people think their pain is getting worse or getting better under the influence of a nonexistent treatment. The rostral anterior cingulate cortex shows a similar pattern – it’s involved in a continuum of activity whether there’s a perceived positive or perceived negative effect going on. And the level of that activation apparently correlates with the perceived expectation of strength or value, so if you’re looking for where in the frontal cortex this gets calibrated, that’s a strong candidate.

I’m reminded of studies of consumer preference. When I was doing my post-doc in Germany, I noticed that (at least in that era) that a surprising number (at least to me) of German shampoos were clear, or at most lightly colored while still transparent. That seemed to be marketing to a preference for purity – this clear, pure substance is just what you need to achieve cleanliness. That was as opposed to the many opaque, opalescent shampoos on the market in the US, which seemed to be holding out the promise of “Yeah, there’s something in there, this is some distinctive and powerful stuff”. Perhaps that also had to do with promises of cleaning power versus other effects on the hair, allegedly “feeding” it, adding “body”, and so on. I don’t expect to see any fMRI studies of shampoo any time soon, but if an easier technique develops to monitor brain activity, I wouldn’t rule it out, either.

That’s because this sort of study is still only the beginning. Neuroscience, as many have predicted (with varying levels of anticipation and fear) is going to continue to mess with our expectations and beliefs about consciousness, decision-making, and free will, and the better the tools we come up with to study and understand brain activity, the more thoroughly, I predict, will we be, well. . .unnerved.



44 comments on “Expensive Shams Are the Way to Go, Apparently”

  1. bhip says:

    People in Oncology complain about drug R&D. Try Neuroscience- you will feel considerably better about your lot in life….

    1. James says:

      Funny – a few years back, I switched from NS to Oncology. Just mixing up the nearly impossible challenges!

    2. Dr. NP says:


  2. johnnyboy says:

    Amusingly, ‘clear’ shampoos contain much additional chemical stuff to clarify them, compared to the opaque ones. And your basic shampoo contains essentially the same detergent concentration as a mild dishwashing liquid, while costing about 10 times more. Which saved me substantially as a student, though now I would never try to convince my wife to switch from her 20$/bottle stuff…

    1. GutDecipher says:

      How much money does one truly shave by switching from shampoo to dishwashing detergent…?

      1. johnnyboy says:

        I guess I’m old enough that a few bucks a week made a difference back then. Nowadays of course students know they’ll be 200 grand in debt when they get their diploma, so might as well pile it all on…

  3. bannonymous says:

    Were they sure that the fancier-looking blue box didn’t leach any potent Zinc complexes from its blue paint?

  4. Canman says:

    How do you qualify this as a nocebo effect and not a placebo effect as the people are expecting a positive response from the cream, and expecting a larger response from the more ‘expensive’ cream? True, the response is a ‘bad’ effect rather than some ‘cure’ or beneficial effect, but it’s what the subjects were told to expect. Thus would actually still be a placebo effect?

    1. tlp says:

      I also got confused with that part, expecting that heat would be a ‘side’ effect and so ‘expensive’ cream should clearly have much less of it. Instead, if the purpose of the study was formulated as stated here, subjects could be expecting direct correlation between heat sensation and anti-dermatitis activity. But then it’s not the true ‘side’ effect, is it? Does that mean that people associate _any_ effect with the price? Probably will have to read the full story.

      1. Another Guy named Dan says:

        more expensive = better
        better = more powerful
        more powerful = greater effect

  5. pete says:

    I didn’t read the paper (paywall) but it made me wonder to what extent some subgroups might begin to stratify (==> highly-suggestible vs less-so) if you extended the treatment BUT applied the same heat for each cream. That is, I’d expect the nocebo might “decay” with time so that some subjects might come to feel & realize that there’s actually no difference – whereas a select subgroup (perhaps?) might continue to experience a real difference (pain) over extended time. This hypothetical “high-suggestibles” subgroup of people would be an interesting cohort for further study, methinks.

    1. NJBiologist says:

      There is some literature suggesting that placebo/nocebo effects can be extinguished over time.

      My brain went in exactly the opposite direction: I would like to see this demonstrated in an experimental design that *doesn’t* allow for conditioning to happen–for example, on the first application of heat.

  6. steve says:

    It’s not just drugs. There was a NY Times article a while back about a guy who opened a store down the street from the famous Zabar’s Deli. He got his lox (smoked salmon for the non-congniscenti) from the same supplier as Zarbar’s but sold it at a much cheaper price. No one would buy it because they suspected it wasn’t good quality. He raised his price and voila! it flew off the shelves.

    1. Old Timer says:

      That’s NYC Effect! 🙂

    2. Anonymous says:

      I was in a shop where an optician was helping someone select new spectacles. He recommended the “guh-VINchies” (give-INCHies) which were on sale. I wonder if the frames by Givenchy (French: jhi-von-shi or so) were more expensive than the Give-inchies.

      In the 1970s, when the HP-35 was the cutting edge, cool, must-have calculator, a major fancy store in NYC had a radio ad promoting the “Hugh-lay Pack-arrrr” HP-35, which is the French fried version of Hewlett-Packard. It no doubt justified their selling at a non-discounted price, unlike other shops and school bookstores which were selling the slightly discounted Hewlett-Packard HP-35.

  7. Earl Boebert says:

    The lox anecdote is an example of negative price elasticity. Perfume is the classic example; up to a point, the more it costs the better it sells. Other “luxury” goods act the same.

    1. Another Guy named Dan says:

      Veblen or status goods – object becomes more desirable as the price increases, due to the positive wealth signaling effect. You’re not paying more for the good, you’re paying more for a greater signal.

    2. Hap says:

      With luxury goods the price could be part of the lure – you’re not really buying the item but social status from having something that other people don’t or having the money to spend on the item, and if the price is lower, more people can get it or its purchase says less about your financial status. With drugs or cosmetics, though, one wouldn’t expect that reasoning to work, because social status wouldn’t be the aim.

  8. DCRogers says:

    I can imagine these results being used to pre-filter clinical trial candidates.

    If your drug has no physically-measurable effect (like anti-depressants make you “feel better”) you may wish to have big placebo responders in your trial.

    If whether your drug works has a physical test (such as weight loss), then you may wish to have low placebo responders, to reduce the number of reported “side-effects”.

    1. Emjeff says:

      As a drug developer for more than 25 years, I can tell you categorically that a large placebo effect is always a problem. There is never an instance where this is desirable.

      1. Chrispy says:

        Really? I mean, it seems like if patients could tell that they were on the drug arm rather than the placebo arm then they’d do better. Consider the VEGF antagonists, which cause such a rash that it is really not possible to blind the studies. I have heard anecdotally that patients with bigger rashes tend to have better responses. While that may be because in these patients the drug is having a bigger effect biologically, it might also be because the patients expect a bigger effect.

        If I was really cynical I’d roll out niacin for depression. Niacin causes noticeable flushing, so the treatment arm would know they were getting the “real deal.”

        1. NJBiologist says:

          Some (typically small, academic) CNS drug trials have used diphenhydramine as a non-inert placebo. Depending on doses, the patients may decide that diphenhydramine was the active treatment… the patients definitely definitely feel it, and many of the older CNS meds have H1 antagonism as an off-target effect.

      2. DCRogers says:

        So is anyone aware of active efforts to change the placebo response rate in clinical trials, either directly (eliminating individuals with a high placebo response) or via change of country (some countries have a higher placebo response rate than others)?

        1. drsnowboard says:

          I think you are forgetting the other important adjective in RANDOMISED, placebo-controlled, clinical trial. A high placebo response is never a good thing because it will happen in both active and control subjects.

          1. MTK says:


            A high placebo rate in both arms is going to minimize your signal window (or whatever it’s called in clinical terms) and lower the chance of seeing a statistically significant response.

        2. Someone says:

          Laboratoire BOIRON,
          not sure if clinical, but definitely HUGE success.
          the most expensive sugar in the world and people love it

        3. Janex says:

          I’ve always though that the placebo response could be reduced by 2-step clinical trials. Step I give the patients in both arms placebo. Step II drop all patients who respond (positive or negative) to step 1 from the trial, the switch the remaining patients to the standard drug vs placebo arms.

          1. tnr says:

            Yes, this is called a placebo washout trial. The placebo washout duration needs to be blinded to both patients and physicians. Used often in anti-depressant trials although European regulators dislike it.

        4. r says:

          Yes. Mostly via the country route, which curiously is disease specific. Depends both on local prejudices and local clinical practice. Makes you really wonder about clinical trial results.

  9. Earl Boebert says:

    I think I share with a lot of “ordinary” people a lack of concern about whether something is a placebo effect or not. If you give me a pill that isn’t toxic and after I take it the pain goes away, I could care less what the mechanism was.

  10. Anon says:

    Why dont we just hook people to lie detectors ?

    1. fajensen says:

      Voodoo like lie detectors only work because of the placebo effect

  11. oliver says:

    Dear Derek
    I’m suprised that no one harped on the real elephant in the room yet, fMRI!!!
    This is total BS and rank pseudo science.
    They could even get a dead salmon to respond to fMRI, FFS!!

  12. Kling says:

    Hey thought experiment:
    Witch doctor prescribes homeopathy potion with fMRI readout. Control would be a glass of water administered by self serve cafeteria beverage dispenser, no smart water allowed.

  13. JimM says:

    Jeremiads against herbs and supplements mostly ignore the benefits people who use them are likely to receive from the placebo effect, and generally decry the fact that many herbs especially can be actively toxic, but this research seems to imply that such toxicity would significantly magnify any benefits to be had from the placebo effect.

  14. ap says:

    Waber, Shiv, Carmon, and Ariely won the 2008 Ig Nobel Prize in medicine for demonstrating that expensive placebos are more effective than inexpensive placebos.

  15. milkshaken says:

    I have always a suspicion of psychology studies with pre-determined conclusions (=that people can be fooled by the expensive packaging or by something that was suggested to them). First, the people who administered this “preliminary study” were fully aware what was going on. So there goes the double blinded study out of the window. Second, if there was a null or inconclusive result, you would not be able to publish the result of the study – which is the apparent goal of the study. Which means the researchers would have to justify how they used their funding, and they wouldn’t be able to get the next grant.

    In other words, I am afraid the quality of this study may not be quite up to the drug development standards

  16. Jbosch says:

    Is it too expensive to have study subjects get exposed to both arms? Control and treatment? In a double blind study so nobody knows when a study subject hets control or drug? The measured effect should be maximized as each individual serves as their own control.

    1. Pennpenn says:

      What, at the same time? Kind of missing the point of the trial then…

      I mean, in theory you should be getting some data from the first set so running the second would be redundant (maybe).

      1. jbosch says:

        Of course not at the same time. Say treatment with X for 3 months then wash out 1 month followed by treatment with Y. Basically two studies with the same subject but one is a control the other is the drug.

        1. Janex says:

          That’s called a crossover clinical trial design. And it is used, depending on the indication.

          Also, one indication where patients can be exposed to both drug and placebo simultaneously is ocular. Drug eye drops in one eye, placebo in the other. Don’t tell the patient which eye is getting which (does require careful compliance on the part of the patient not to get the two bottles of eye drops mixed up – either accidentally or on purpose [Hey if one eye was working and the other wasn’t I’d be awfully tempted to put that drop in both eyes too!]).

  17. Anonymous says:

    Not a placebo, but a sham report via Retraction Watch on Thurs, 5-Oct:

    Acerta Pharm “has admitted that one of its former researchers falsified early data on a compound that’s designed to fight cancer, now in human trials.” The report was published in Aug 2015; Astra-Zeneca paid $4 B for 55% of Acerta in Feb 2016. The falsified data apparently does not bear on the continuing clinical trials. But it probably helped to bring in $4 B deal!

  18. Project Osprey says:

    Regarding the shampoo situation – you’re absolutely right. I used to work for a big-name in personal care and a lot of what they do comes from those sorts customer perceptions. Any kind of ‘hypoallergenic’ laundry liquid is always white. Beyond that greens and blues tend to dominate because they are ‘clean’ colours (blue skies and grass I guess?). You’ll never see orange or red laundry liquid, even though its all just dye anyway. Shampoos are advertised schizophrenically: ‘we’ll wash your hair clean’ and ‘we’ll wash all these benefit agents onto your hair’. Don’t even get me started on foaming… sort term for shampoo because you want it to wash out easily but ages and ages for hand dish washing because people put a lot of stock into foam. From a practical stand-point you want you surfactant in the water rather that suspended in bubbles above it but I guess that’s people for you.

Comments are closed.