The Placebo Effect: Is it real, is it ethical?

[pantsorama]

It is a narrow definition, and that's part of the reason there's confusion among laypersons... the definition in the OP is more what I call a 'placebo expectation effect' but unfortunately the body of literature does contain a mix of definitions for the term 'placebo effect'.

This is part of the elegance of Hróbjartsson's approach: their reasoning was that if a person being billed for a placebo with expectations and the whole ball of wax had as much improvement as a patient put on the waiting list, then whatever's happening, it's not being caused by anything associated with the placebo.

Cool - thanks for pointing this out!

 

[angrysoba]

From my memory of Ben Goldacre's book, the placebo effect has sometimes been studied by having one group take four sugar pills and one group take two sugar pills to see who reports feeling better.

However, Goldacre also points out that discovering whether or not placebos have any effect is of much less interest to medicine than finding out what works better.

 

[Meed]

Couldn't one split people into two groups: those who get no medicine at all, and those who get a sugar pill?

Yes, but that won't account for the issue of whether people think that their symptoms have improved vs whether their symptoms have actually improved. That's why I suggested physiological measures being a better way of testing it.

Of course, people thinking their symptoms have improved can also be deemed to fall under "the placebo effect" and in that sense of the term, one can simply test a sugar pill vs nothing.

 

[Iamme]

If I were running that test I would have said, "Here, this is an inert pill that you're body can't absorb. There's no physiological reason to believe it will do anything for you. Please take it for the duration of the test and give us the results."

Really. Why if people were told this and still some said this helped them? Why you`d learn something ELSE, besides just the placebo effect. You`d also learn these people would have to be plumb ! Then, study them for THAT.

 

[Pup]

Yes, but that won't account for the issue of whether people think that their symptoms have improved vs whether their symptoms have actually improved. That's why I suggested physiological measures being a better way of testing it.

If the symptom is depression, for example, or pain where there's no loss of range of motion, what physiological measures would you use to test it? Some things are difficult to objectively test, yet are important for quality of life.

I'm having trouble seeing the difference between someone thinking their pain, depression, craving for alcohol, anxiety, etc. has improved, and it actually improving.

 

[Meed]

If the symptom is depression, for example, or pain where there's no loss of range of motion, what physiological measures would you use to test it? Some things are difficult to objectively test, yet are important for quality of life.

I'm not sure that you could use physiological measures for either of those, but you probably could for anxiety.

I'm having trouble seeing the difference between someone thinking their pain, depression, craving for alcohol, anxiety, etc. has improved, and it actually improving.

Why?

 

[Pup]

Why?

Because of what you said in the first paragraph: "I'm not sure that you could use physiological measures for either of those."

If there's no physiological measure, then there's no test other than the patient's report, which is based on what they think. Therefore, there's no way to define "actual" improvement, apart from what the person thinks.

 

[Libra]

. . .
Thats not to discount patient perception's part in the placebo effect. It is probably the largest part, truth be told. Although I have no idea how one would measure the size of each effect. But to say that it is all patient perception is too narrow a definition, IMHO.

Yes that part of patient's perception is important.

I periodically allow myself to think about religious folk as persons receiving a 'placebo' for their psychological and sometimes physiological challenges and that their altered perception allows them some relief / success in some of these matters they attack armed with religion.

Apologies for not staying 100% on topic there, but in short, Yes it's a real effect and the use of it can be both ethical and unethical, depending on the circumstances, IMHO.

 

[Meed]

Because of what you said in the first paragraph: "I'm not sure that you could use physiological measures for either of those."

If there's no physiological measure, then there's no test other than the patient's report, which is based on what they think. Therefore, there's no way to define "actual" improvement, apart from what the person thinks.

Do you acknowledge that it's possible for a statement such as "I've been less depressed this year than last year" or "the time I had a charlie horse caused me more pain than the time I had a kidney stone" to be incorrect? I think the accuracy of such judgments is often dubious. It is easy to see that a statement such as "I drank more coffee this week than last week" could be wrong. And further complicating things, a judgment such as "I was in more pain yesterday than today" is much less straightforward since we don't typically associate psychological experiences with numerical quantities. When you take into account the power of pervasive human biases such as confirmation bias, selective memory and so on, it becomes even more difficult to trust these types of comparative judgments.

It is important to distinguish defining something from operationalizing it. We might operationalize how much coffee someone drinks per week by asking them how much they think they drank, but we would not define how much coffee they drink to mean "how much coffee they think they drink".

 

[Belz...]

The placebo effect is not a change in the medical condition but in the patient's reporting.

 

[MikeG]

......2. Assuming that there is evidence in support of the placebo effect, evidence that indicates it works and that people believing in treatments is an effective way to help them, is it ethical? .......

Sorry to be a bit picky, but the effect, if one accepts that it is real, is neither ethical nor unethical. It simply is. You may as well ask if a snowflake or a brick wall is ethical. What you mean, and what you go on to imply, is the question of whether using or exploiting the effect is ethical. This is a rather different thing.

Both your headline and your OP make this mistake.

 

[Pup]

Do you acknowledge that it's possible for a statement such as "I've been less depressed this year than last year" or "the time I had a charlie horse caused me more pain than the time I had a kidney stone" to be incorrect? I think the accuracy of such judgments is often dubious.

Compared to what standard? How are you going to objectively prove that their pain was greater or lesser at a particular time?

Even with actual drugs, how does one test, for example, the objective degree of pain relief that any given amount of morphine produces?

We can assume that there will be human biases in reporting pain, depression, etc., but if minimizing the symptoms is the goal, the person's perception is the standard.

A doctor can tell someone: I know you said you feel as if the cancer is gone, but the biopsy shows it's still there, so we need to continue treatment.

But it doesn't work to say: I know you said you're not feeling any pain, but the test shows you're still in pain, so you need to continue pain medication.

It is important to distinguish defining something from operationalizing it. We might operationalize how much coffee someone drinks per week by asking them how much they think they drank, but we would not define how much coffee they drink to mean "how much coffee they think they drink".

How are you going to define the amount of pain a person is in, or the amount of depression they feel, or the strength of their desire to drink alcohol, other than by their subjective experience, in order to compare the real amount with what they report?

Edited to add: There was a long thread on objectively measuring pain a while ago: http://www.internationalskeptics.com/forums/showthread.php?t=152854

 

[Meed]

Compared to what standard? How are you going to objectively prove that their pain was greater or lesser at a particular time?

"Such judgments are dubious" need not be compared to any standard to be true. And I didn't say that such a thing (i.e. whether one's pain was greater or lesser at a particular time) could necessarily be proven.

Even with actual drugs, how does one test, for example, the objective degree of pain relief that any given amount of morphine produces?

A double blind experiment in which a morphine group is tested against a placebo group is the best way I know of. It would rely on self reports, but the placebo group would control for the any possible biases in judgment resulting from the knowledge that one is undergoing a treatment and sample size / statistical inference would account for random variation. These experiments don't definitively prove anything, but they can potentially provide very compelling evidence.

We can assume that there will be human biases in reporting pain, depression, etc., but if minimizing the symptoms is the goal, the person's perception is the standard.

A doctor can tell someone: I know you said you feel as if the cancer is gone, but the biopsy shows it's still there, so we need to continue treatment.

But it doesn't work to say: I know you said you're not feeling any pain, but the test shows you're still in pain, so you need to continue pain medication.

Agreed, that doesn't make sense. I'm not suggesting we should doubt self report judgments to the point where it we disbelieve statements such as "I'm in pain" or "I'm not in pain". I'm suggesting we should sometimes be skeptical of statements such as "I was in more/less pain during week A than week B" when there are possible judgment biases at play. For example, let's say John just read a book on homeopathy and became a believer. He begins taking a homeopathic remedy to help with his pain. A week later you ask him whether his pain has improved since he started the treatment. Now, the problem is that confirmation bias is very strong. If he believed that it would help, then he will (whether he is aware of it or not) tend to seek and remember evidence that supports that belief and disregard and not remember evidence that goes against the belief. For instance, he might have had a more painful than usual Tuesday morning and a less painful than usual Wednesday evening. According to confirmation bias, the latter is more likely to stick out in his mind. It's not like we make these judgments by remembering exactly how much pain we were in at every minute during the week and mathematically averaging them out. We rely on heuristic shortcuts and heuristic shortcuts are susceptible to bias.

 

[Pup]

A double blind experiment in which a morphine group is tested against a placebo group is the best way I know of.

But that's completely subjective from the patients' point of view. I thought you were wanting something that eliminates patients' bias.

It would be an accurate test of how the placebo worked compared to morphine, and I think it would be a perfectly adequate test. But according to what you've been saying, I thought you'd consider it inadequate, because the patients' reports about their pain could not be objectively verified, the way an X-ray could verify a broken bone or a blood test could verify blood sugar levels.

I'm suggesting we should sometimes be skeptical of statements such as "I was in more/less pain during week A than week B" when there are possible judgment biases at play.

Those problems are easily solved by double-blinding, as you noted, and also by asking the patients how they feel today, writing it down and then comparing it to how they answer the same question when asked a week later. But I'd certainly expect any medical trials to be double blind, and to ask questions of the patients and write down the answers at the time of assessment and not rely on their memory of how they felt previously.

I'm not sure what we're disagreeing on, if subjective assessments from patients are suitable, like "My level of pain/depression/intrusive thoughts/etc. is at 2 out of 10 today."

 

[Iamme]

Wouldn`t pain show up in a brain monitor? Wouldnt that settle the argument?

 

[Meed]

But that's completely subjective from the patients' point of view. I thought you were wanting something that eliminates patients' bias.

It's subjective, but the control group and the double blinding seek to prevent systematic biases from affecting the results. That is, the control group and the treatment group should be equally biased on average. Then it would seem to be fair to conclude that if the treatment group reports significantly less pain it is probably because their pain has actually been reduced.

When I brought up physiological measures it was in the context of how one might test the placebo effect itself. More specifically, whether it actually reduces symptoms or merely biases judgment of symptoms. A placebo can't be tested against another placebo and if you test a placebo group against a no-treatment/no-placebo group then you can't account for judgment bias confounds. That is why I suggested physiological measures as a possible solution. I do agree, though, that their viability for assessing psychological symptoms is questionable.

Those problems are easily solved by double-blinding, as you noted, and also by asking the patients how they feel today, writing it down and then comparing it to how they answer the same question when asked a week later. But I'd certainly expect any medical trials to be double blind, and to ask questions of the patients and write down the answers at the time of assessment and not rely on their memory of how they felt previously.

I'm not sure what we're disagreeing on, if subjective assessments from patients are suitable, like "My level of pain/depression/intrusive thoughts/etc. is at 2 out of 10 today."

It seemed you were saying that how much one thinks they've improved IS how much they've improved, whereas I view it as an imperfect measure that may be inaccurate at times. Otherwise we wouldn't need carefully controlled experiments.

I may have given off the impression that I think self reports are useless, but that was not my intention. I think generally speaking, they are the best measure we have when it comes to most psychological symptoms. But we still have to treat them with a level of skepticism, otherwise we open the door to all sorts of claims about alternative medicines, alternative diet plans, and so on "working wonders". It's easy to find perfectly sincere glowing testimonials of all sorts of snake oil products. I suspect a lot of that can be attributed to confirmation bias.

 

[Meed]

Wouldn`t pain show up in a brain monitor? Wouldnt that settle the argument?

FMRIs are very expensive and thus not always practical, but yeah I believe they can be used to assess pain. I'm not sure how accurate such measures are, though.

 

[Arcade22]

An increase in people reporting they feel better but not necessarily an increase in people actually being better?

Depends on what you mean by "being better". Someone could feel perfectly healthy and healed because they were given sugar pills although that cancer is still eating away at their brain.

 

[blutoski]

FMRIs are very expensive and thus not always practical, but yeah I believe they can be used to assess pain. I'm not sure how accurate such measures are, though.

Pain is too qualitative to measure with any type of brain scan such as fMRI. The scan would only show that there was sensation, not whether it was painful versus a tickle.

We don't necessarily need physiological measurements for pain or mood - there are binary outcomes such as 'ability to resume daily activities' or 'suicide attempts'.

The findings of Hrobjaartsen et al were that over a thousand studies have been done to investigate whether placebos can improve all sorts of conditions, and while patients report improvement in linear subjective surveying for a couple of conditions, the objective binary outcomes show a different story.

This has important implications for healthfraud, as we see the same story over and over again: the patient lauds the superior results of the quack who 'saved' them, right up to the day they die prematurely of the condition that was 'improving dramatically'.

If the argument 'it works via the placebo effect' is as unsupported as it seems, a lot of the moral argument for legalizing treatments that have no scientific rationale is undermined.

 

[Reality Check]

So the researchers told participants that it could improve their symptoms ... that IS the placebo effect.

No- that is called priming the subjects.
The placebo effect is that telling people something will work can make them perceive results even if that something has no effect, StevenLeonCooper!

The study was not about whether people can know from the effect that a drug is a placebo. It was about whether the subjects even when they knew that they were taking a placebo would show the placebo effect.

Ted Kaptchuk does tend to overemphasize the results of the placebo trials he has done which even so have had interesting results.
The latest is Altered Placebo and Drug Labeling Changes the Outcome of Episodic Migraine Attacks which is analyzed at Science-Based Medicine