This is the surprising finding of a paper just published, Adherence to placebo and mortality in the Beta Blocker Evaluation of Survival Trial (BEST).
BEST was a clinical trial of beta blockers, drugs used in certain kinds of heart disease. The patients were aged about 60 and they all suffered from heart failure. Everyone was randomly assigned to get a beta blocker or placebo, then followed up for 3 years to see how they did.
Here's the big finding: in the placebo group of 1174 patients, the people who took all of their placebo pills on time (the good adherers), were significantly less likely to die than the patients who missed lots of doses. People who took over 75% as directed were 40% less likely to die than those with less than 75% adherence:
That's pretty interesting. The pills were placebos - they can't have had any benefit. So what's going on?
It gets even better. You might be tempted to write off these results as obvious: "Clearly, people who follow the study instructions are just 'healthy' people in other ways - maybe they take more exercise, eat better, etc. and that's what protects them."
Certainly, that's what I'd have said.
But what's remarkable is that when the authors corrected the statistics for all the confounding variables they measured - including things like age, gender, ethnicity, smoking, body mass index and blood pressure - it barely changed the effect. Some of the factors did correlate with adherence, but not in a way that it could explain the adherence effect on mortality.
This isn't the first study to find this effect. The authors themselves have already reported it, as have other researchers going back decades (many of which also tried, and failed, to explain it through confounding factors.) They say that it's unlikely to be a case of publication bias.
So what we have is a large effect, which cannot be causal, yet which can't be explained by any obvious confounds. Logically then, it must be the result of a confound (or more than one) that aren't obvious.
This is an important lesson. It's common for someone to do a study and find an interesting / scary / controversial correlation between two things. Often one is some kind of lifestyle factor, diet, environmental exposure, or whatever, and the other is some nasty disease. "And it wasn't explained by confounds!", such studies often conclude.
What the placebo adherence effect demonstrates is that there may be confounds no-one has thought of. They might even be impossible to measure. And if these mystery confounds can literally kill you, they can probably cause all kinds of other effects too.
In other words this illustrates the truism that correlation is not causation - not even when you're really sure it is...
Pressman, A., Avins, A., Neuhaus, J., Ackerson, L., and Rudd, P. (2012). Adherence to placebo and mortality in the Beta Blocker Evaluation of Survival Trial (BEST) Contemporary Clinical Trials DOI: 10.1016/j.cct.2011.12.003
38 comments:
So what we have is a large effect, which cannot be causal
Why could this not be causal?
While I agree there may be cryptic confounding, I don't think you can exclude a causal effect of "swallowing a pill" just because there's no known mechanism.
Sorry for the double-post, but... One possible confounder is psychological state associated with severity of illness. More severe illness -> greater apathy/lethargy/motivation -> poorer adherence.
It could also be causal in another way: Good adherers to their placebo regimen are probably people who follow the instructions of their physician in other matters too. For example: Do they take a flu shot every winter, do they take their other medicines as prescribed, do they consult often, etc.
For elderly patients, a lot can happen during three years even taking into account the admission criteria in the clinical study (presumably no other serious condition at study onset) and general health measured at the beginning.
***So what we have is a large effect, which cannot be causal***
A second what "red" said above. What about the placebo effect is not causal? A *belief* in treatment caused an improvement in health.
I'm curious, do you think that beliefs have no causal power? It seems to me that clearly they do, in all sorts of ways.
Sorry, that "not causal" thing was a bit unclear.
What I meant was, it cannot be that actually taking more placebos, prevents death. Because placebos are inert.
Maybe people who believe in the placebo more strongly a) get a benefit from that belief and b) also tend to take the placebo more faithfully, compared to doubters.
Although I'd be amazed if that had such a strong effect on mortality.
But I would say that's not causal. It's still a confound.
There's actually a theory that explains this - Michael Hyland's motvational concordance. He suggests that the ritual of taking the pill, especially when its congruent with your system of beliefs, has healing power. Now, obviosuly its not the sugar or lactose having the impact, but the act of taking your medicine has psychological significance, which could easily activate the body's immune responses and healing systems.
Your point about confounds is a good one, but I'm more inclined to go for the ritual placebo theory, as at least its testable.
***What I meant was, it cannot be that actually taking more placebos, prevents death. Because placebos are inert.***
If taking two sugar pills results in twice the improvement compared to taking one sugar pill (an actual finding) and taking a placebo is known to result in actual neurophysiological changes (e.g. taking a placebo thought to be an opiate results in an attenuated physiological response analogous to taking an actual opiate; another actual result) then taking more placebos might actually have a causal relation to health improvement (i.e., death prevention).
There was a recent New Yorker article on the subject of research on placebos that I would recommend...very interesting stuff.
p.s.
it still seems as though you are suggesting that beliefs do not have causal power.
There is only one explanation for this: the researchers used homoeopathy pills as their so-called "placebos", and what we are witnessing is the amazing healing power of homoeopathy.
As a medical oncologist, Ive seen any number of things i'll admit i can't explain. In this case, however, it may simply be that adherence with placebo is a surrogate marker for adherence to other meds. Many CHF pts are on several meds for their disease and if they don't take meds properly, they won't work and results (and patients) will suffer.
What I would be interested in is a comparison between the magnitude of the "adherence effect" between the placebo and the active drug groups.
Anonymous: I'm not saying that beliefs don't have any causal power but I don't deny I'm skeptical that beliefs alone could cause such a change in mortality rate.
Ultimately because this is not a RCT, we can't tell. It comes down to a balance of probabilities: is it more likely that this is due to confounding, or due to a placebo effect?
Neil: That's quite possible; placebo adherence could be a proxy for other adherence. But if so, it's unlikely to stop with medicines. People who adhere to prescribed drugs probably also have different lifestyles, personality traits, psychosocial variables, from people who are sloppy about that kind of thing.
I wonder if you could come up with a cheaper way of measuring "placebo adherence".
The problem with placebos is that you need a whole clinical trial with sugar pills being dispensed by pharmacists... very expensive and time consuming.
But suppose you set up a computer system or whatever, that people had to log on to every week, at a certain time, and fill out a questionnaire or something.
The questionnaire itself could be anything, the outcome measure would be whether people did it on time as instructed.
I bet that would correlate with placebo adherence. And if so, it could be an important thing to measure in long-term health studies.
Call it "compliance".
As Neil points out, CHF is very tricky to manage. Patients have to monitor themselves closely. It's probably a surrogate for good self-care and attention. I suppose it's possible that placebo reduces anxiety, though, and leads to a slower resting heart rate. Interesting study. Thanks.
I remember reading about a study once several years ago that showed that if doctors handed actual pills to a patient in a brown paper bag as opposed to giving them a script, then the pills had a more powerful effect. To me, this showed the powerful effect of simply creating a trusting and personal atmosphere to facilitate belief in the effectiveness of a medical intervention.
It is the belief that has an effect and anything that promotes the belief maximizes the effect.
Diligence in taking pills could be seen as a marker for the belief that a pill is effective.
I'm interested in looking more closely at two things:
The patients' perceptions of the physician/patient relationship: what role did the patients perceive themselves to play and what role did they perceive their physicians to play relative to the decision to prescribe the studied intervention
And secondly, what expectations did the patients have about the optimal plan of treatment relative to the prescribed/studied intervention?
I hypothesize that the more empowered patients perceived themselves to be and the more congruent their expectations of treatment were with the studied intervention, the lower their stress responses to the illness and so predict lower stress markers and lower mortality rates.
I was very surprised when I first heard about this study:
http://www.sciencemag.org/content/293/5532/1164.abstract
where a placebo actually induces dopamine release in the brain (it is an L-dopa placebo for Parkinson's patients). I don't know how this would translate to heart disease medication, or whatever the placebos were replacing in this study, but it is a powerful example of 'belief' resulting in relatively drastic physiological changes.
"Take your Placebo and Diet" Veeeeery interesting this post thanks Neuroskepic to let us know this paper. I have a question, although much is already said. Seems like that a positive attitude toward the cure can be the or part of the explanation, as you said who adhere to the therapy, may be more motivated to have also life style changes and as far as I know patients who has congestive heart disease need to change diet and life style. So how big is this component I don’t know but THIS may have for sure an effect on mortality.
I wonder if this effect can be seen in study in which NO life style changes are involved.
We need to keep in mind the unrealistic assumptions in "control of confounds": that we have specified all of the confounds and measured them perfectly. before we run off to some occult explanation, consider imperfect control of confounds, the most likely suspect.
Optimists simply live longer. They see the "point" in doing all kinds of things. Like cleaning the house, socializing, having sex, meeting new people, engaging in new activities, and taking meds - all of which are known to correlate with health but which weren't taken into account by this particular study.
But maybe their optimism is but an effect of an underlying confound ;-)
What if it takes time to become a good adherer?
Is the study reporting an average adherence number? It looks like they are but then I have not read the study. So if it takes time to become a good adherer and you die before that time then you appear to be a bad adherer but what actually put you in the bad adherer group was a random event of death.
Yes i thought this was interesting. There's a good podcast on the physiology of the placebo effect.
http://www.brainsciencepodcast.com/bsp/neurobiology-of-placebos-with-fabrizio-benedetti-bsp-77.html
Neuroskeptic points out that even if the ritual of taking a placebo had a physiological impact this would still be a confound, since the placebo itself is inert. But it's interesting that potentially the actual act itself of taking a sugar pill could induce physiological changes with a positive impact.
I suppose once you get over the whole cartesian mind-body divide it becomes less suprising. It is well documented psychological factors can cause physiological changes both in the brain and body.
DS: Yeah, they call that idea "time effects" and they discuss it in quite a lot of detail in the paper.
Red, Disgruntled PhD, Manchester psychiatry society,Richard,
Thanks for your comments.
neuroskeptic,
Correlation is not causation meant only for my masters in stats at paris VI University that you cannot take a correlation for a causation
but not that it was enough to think that a causation like a placebo or a homeopathy good effect is impossible to make it so.
Beware that antidepressants are in many cases not more useful than placebos but that their side effects are no placebo but then when the side effects gets withdrawal symptoms
big pharma and academic psychiatrists on their payroll teach GPs, gerontologists and psychiatrists that temporal correlation is causation.
Like if you were made to believe that because one alcohol dependant person had suffer a delirium tremens episode from stopping alcohol intake by himself that person should resume her heavy daily alcohol intake because the delirium tremens was absolute proof that alcohol was good for that person's health; and that one has to accept an alcohol health risk because without alcohol the brain of that client will get damage as it had been proved without the shadow of a doubt by the delirium tremens episode following alcohol withdrawal. cirrhotic and cancer
I saw last week a women who had been beautiful for her age. By that I mean mostly a strikingly beautiful figure at 50 something and an interesting lively mind and sexually active plus physically active (black karate belt still frequenting the tatamis ) and had become overweight in an ungraceful manner and without any libido in a temporal correlation with an antidepressant treatment.
She had been prescribed an antidepressants following the treatment of a brain meningioma (not the worst between the brain tumors who can be cured when the surgeon can take it off completely) by neurosurgery. She the antidepressant treatment didn't work and had made her figure heavy and formless and suppressed her libido when she was afraid that her change in figure might have act negatively on her husband sexual interest in her without suppressing his libido in general.
She had stopped her antidepressant by herself in one go.
She felt bad following that brutal interruption of antidepressant regimen -as expected- insisting on vertigo being her main complain
and she got a prescription for two antidepressants instead of one and her weight gain is in progress.
Of course, I could do nothing since she has a psychiatrist who told her that the withdrawal symptoms were an absolute proof that she was depressed and in need not only of her former antidepessant but of another one in addition since it is known-said that psychiatrist and not being alone in believing so- that when you stop an antidepressant it ceases to be active when you resume taking it.
Of course, it is an anecdotal story but a sad one who makes me think that prescribing placebo or homeopathy in difficult situations when the social pressure for psychiatrizing the fact of being human (and loosing dear ones by death or otherwise, getting a tumor to be operated with a neurosurgeaon who sees you only every 6months for the follow up , emptynest etc...)is so strong is a very good thing.
I think the disconnect between causality and correlation has much more profound implications. It means that research "evidence" alone can never be sufficient basis for making epistemological decisions. I've suggested elsewhere that that ultimately all decisions are ethical - i.e. consistent with real world actions. See http://metaphorhacker.net/2011/03/epistemology-as-ethics-decisions-and-judgments-not-methods-and-solutions-for-evidence-based-practice.
That's not to say that going by correlationist research evidence is not a good idea for a lot of the time. But it cannot possibly ever be the immutable point of arbitration of truth it is commonly mistaken for.
I propose a complex Rube Goldberg Sequence:
1. Person takes all placebos as prescribed.
2. The pills that were taken out of the bottle reduce the weight of the bottle very slightly.
3. The miniscule weight shift causes the earth to develop an infinitesimal wobble in its rotation.
4. The slight change in rotation makes a butterfly in China flap its wings a little differently.
5. You know the rest ... the flapping of the butterfly wings eventually causes a thunderstorm in Pasadena.
6. The rain leaks through the ceiling of the research offices.
7. The water soaks the researchers' computers, scrambling up their data.
8. The researchers use the garbled data to come up with their erroneous conclusions.
Dominik,
Actually the problem has been think upon usefully by Italian historian and semiotician Carlo Ginzburg-from which Thomas SZASZ Psychiatry the Science of lies" has taken much in an honest way ussing comas of sort: Freud was wrong pretending that psychoanalysis could belong to the (Naturwissenschaften) hard sciences and not be part of the (Geisteswissenschaften) sciences of the human spirit and behavior and it is the same for medicine.
Ginzburg argues that not only psychoanalysis but all of the medicine field research is conjectural and would never meet the criteria of scientific inference essential to the galilean approach.
In medicine and clinical research we have to deal with a collection of individuals living their lives
when the scholastic philosophers and galilean types scientists are lucky because they can state :
"individuum est ineffable" (we are not concerned with individuals and nothing can be said about it in a free translation).
Neuroskeptic,
Actually I refrain -when commenting your post on free will- to remind the fact that Sigmund Freud insisted on negating free will to the human kind in a fraudulent attempts to pass his imaginary cases of cure by psychoanalysis for proofs that he invented a new hard science. sciences...
Saying "what we have is a large effect, which cannot be causal" is just refusing to accept that we only know a tiny fraction of what there is to know about how human bodies work.
As we all know, the phrase "placebo effect" is a placeholder for cascades of physiological mechanisms we have not yet identified.
What happens in a brain and body when a person takes a sugar pill or is hit on the forehead by a preacher or shaman or goes through some other non-medical experience to cure an illness? How can such events set off trains of other events that do cure illnesses?
We don't know. Not a clue, really. Hence the placeholder phrase.
Eventually we will have the tools, time and people to trace those cascades of physiological mechanisms. And then, of course, the placebo effect will disappear, replaced by a very different and more detailed understanding of physiology and medicine than we have today.
So no, it's not at all clear that there are hidden confounders in this study. There are unexplained results - but we should expect that, and work to understand them, not assume that the observations are false or incomplete.
All of our sciences are in their infancies, no matter how puffed up we get about the marvels we have discovered. Our successors will tell anecdotes about our primitive understanding and our absurdly wrong (from their perspective) assumptions.
Maybe the placebo effect is only seen for diseases or illnesses with a strong psychological component. (Stress and heart disease can go together.) For those diseases in which your state of mind has very little bearing, perhaps there would be no effect from the placebo and only from the medication.
"People who take their medication as directed are less likely to die - even when that "medication" is just a sugar pill."
except that this wasn't really a trial of sugar pills vs beta blockers. it really looks more like something along the lines of "what happens when you ask people who are already taking scads of drugs to add either a beta blocker or a sugar pill to their regimen?"
if you scroll down to the bottom of Table 1 (Base-Line Clinical Characteristics of the Patients According to Treatment Group), you see that >90% are taking digitalis, a diuretic and ace inhibitor; nearly half are taking aspirin, a vasodilator, and an anticoagulant; and roughly one third of them are diabetic. my first guess would be that people who aren't taking their placebos regularly might not be taking their other drugs regularly either.
It would seem to me that there are many factors to taking medication. It doesn't really matter if it is a placebo or the real medicine. First, we have to start with the person's willingness to see a doctor, is it by their own choice? Is there a level of trust in the doctor's prescription of said medication, is there a cost factor to the person getting the medicine, and most importantly do they have a mind set to get better? We have to realize humans are very complex and that one correlation or theory may not apply to all in a blanket judgement or statement. People say laughter goes a long way, but we don't go out of our way to add laughter to our lives because we are all too busy trying to live our lives on a daily basis.
Hi Neuroskeptic
You wrote:
"DS: Yeah, they call that idea "time effects" and they discuss it in quite a lot of detail in the paper."
OK they discussed something called time effects but did they eliminate the effect in some way? If so how?
hipparchia: That's a good point, and could well explain it, I would class that as an unmeasured confound.
Sorry to be only a clinician looking to improve prescriptions and health of clients but why should so many comments be trying to find bias in this study in order to refute that a doctor can help a client by prescribing a placebo?
In medicine,to my mind, it is not extraordinary claims like placebos are useful or homeopathy is a good tool for physicians trained in it- that should require extraordinary evidence;
Extraordinary evidences are needed for the efficacy and benefit considering the side-effects of new (and old)treatments with dangerous side-effects like Agomelatine who made a French "real doctor" (I mean by "real doctor" not a psychiatrist ) wrote against a blatant promotion by Australians academics on the Big Pharma payroll to The Lancet editors (see the NS on the Trojan horse recent post) .
Idem for the packing of children if Pr Delion 's controlled clinical trial proves a benefit:
The Lancet, Volume 370, Issue 9588, Pages 645 - 646, 25 August 2007doi:10.1016/S0140-6736(07)61322-1
Neuroskeptic
Hipparchia? I am honored ... I think.
DS: Oh. I was replying to user "hipparchia". I still need to look up the paper again to check those time effects... working on it.
I don't suppose the authors considered Locus of Control as a variable to control for...
http://en.wikipedia.org/wiki/Locus_of_control
Post a Comment