Placebos work, even when patients are in the know, study finds

[blutoski]

There has been a small study like this before, which also came to the conclusion that placebos can be more effective than no treatment. I can't post links yet, but if you google for "An Exploration of Neurotic Patients' Responses to Placebo When Its Inert Content Is Disclosed" you should find it.

Yes, one study in 1965. Still unreplicated AFAIK. If it was a Homeopathy study, I'm pretty sure we'd dismiss it with a chuckle. Same with the study in the OP: are there any plans to replicate it? How are they distinguishing between patient reporting bias and therepeutic benefit?

It's important to know that there are 200+ other studies that show no benefit over nontreatment.

This was the point of Hrobjaartsen's Cochrane Collaboration literature review (last updated about a year ago with all available studies at that time): what does the body of literature say?



ETA: link for literature review: [Placebo interventions for all clinical conditions]

 

[Squeegee Beckenheim]

Yes, one study in 1965. Still unreplicated AFAIK. If it was a Homeopathy study, I'm pretty sure we'd dismiss it with a chuckle. Same with the study in the OP: are there any plans to replicate it? How are they distinguishing between patient reporting bias and therepeutic benefit?

It's important to know that there are 200+ other studies that show no benefit over nontreatment.

Yes, I understand all this and, indeed, outlined some of the same issues that you bring up in the paragraph immediately after the one you quoted.

 

[Uncayimmy]

What I meant was that I keep seeing this as a thowaway recommendation whenever there's a benefit in a placebo group. One of the purposes of a control group is to identify imagined effects. If the effects are influencable by expectation, this should really be categorized as a probable imagined effect, and manipulating expectations should not be advised.

Why should we should ignore the value of expectancy?

Confounding^WP is the name for patients reverse-engineering the blinding. I put this in the wiki tag so you can review the entry on wikipedia.

I know what confounders are. Nothing in the Wiki article says anything about patients reverse-engineering placebos. So, I disagree that it's, as you put it, called confounding. And what do you mean by patients reverse-engineering the blinding? They can make good guesses about what they are on, but that's about it.

Good question. Active placebo^WPs are probably not highlighted in the protocols as such. It's actually unusual for there to be a detail about placebos at all. Some people consider it an important deficiency.

What you are saying? That some placebo controlled studies are not bothering to mention that they are using active placebos?

If we're talking about (in my example) severe depression and they're feeling suicidal, the standard treatment would almost certainly be for them to be on breakthrough meds (that's why they're candidates for a clinical trial for an antidepressant medication in the first place!). Putting them in a nontreatment group contrary to best treatment guideline when there's elevated risk of suicide is totally unethical. It's not even my opinion - this is the ruling by ethics boards throughout the West.

That's irrelevant to no-treatment vs sham treatment distinctions, which is what we're discussing here.

 

[Uncayimmy]

I was saying that by Phase III we should know, because the previous Phase 0 and Phase I were done with real patients for the purpose of identifying side effects. There would be multiple controls, depending on the condition.

Your understanding is different than mine. Phases 0 and 1 are unblinded and uncontrolled. They usually have small sample sizes and only last about a month. They are looking for adverse events more so than what we call side-effects. Phase 2 is when the blinded studies with controls start.

OK: I misinterpreted your paragraph. You said:

I know what I said. I expect people to follow the thread, and we're talking about placebos.

To address the question again: yes, the goal of a placebo is to create a control that does not affect the target outcomes (symptomatic or therapeutic) but if it can be identified as a control and this defeats blinding, there's no point.

It depends entirely on the nature of the study. Subjective reporting can be affected, but objective measurements are virtually unaffected. It really depends on the situation.

 

[blutoski]

Why should we should ignore the value of expectancy?

Because it's jumping to conclusions. If the benefit is reporting bias, there's no real improvement and it may represent a distorted patient-physician relationship, and it's inappropriate to advise doing this.

I know what confounders are. Nothing in the Wiki article says anything about patients reverse-engineering placebos. So, I disagree that it's, as you put it, called confounding.

I find this akward. I work in clinical trials. We call it confounding. It is a phrase used to describe all sorts of extraneous variables, one of which is reverse-engineering to defeat blinding.

The confounding subject at Wikipedia may not have the examples you're looking for because it's not a term specifically about clinical trials: it is a phrase used to describe reverse-engineering other tools, like surveys or psychological inventories. (In another thread a few years ago, there was a poster who was upset that it's hard for laypersons to get hold of MMPI, Stanford/Binet, or Rorschach tests. The reason is that this would confound these tests.)

Another example of confounding that we attempt to mitigate in the protocol is pooling. It's the practice of many patients getting together to combine their medications and mix them up so that they're on average getting a partial dose. This is obviously more common when the condition being treated is life-threatening and the standard treatments are not very good. ie: HIV med trials in the 1980s were frequently confounded by pooling. The protocol now is to administer single doses under supervision.

And what do you mean by patients reverse-engineering the blinding? They can make good guesses about what they are on, but that's about it.

Yes, that's what I meant about reverse-engineering. Making 'good' guesses. It will confound the results.

What you are saying? That some placebo controlled studies are not bothering to mention that they are using active placebos?

Basically. Or they say it's an active placebo but don't go into much detail. Placebo descriptions can lack detail for brevity.

That's irrelevant to no-treatment vs sham treatment distinctions, which is what we're discussing here.

You said:

Years ago some researchers argued that anti-depressants should include no-treatment groups as well as sham treatment groups. Unfortunately, I don't think that's happened in but a few studies.

I pointed out that there was an ethical reason this is rare and becoming rarer.

You asked:

What is the ethical issues? Some people are getting placebos while others do not participate in the study. What is unethical about tracking the no-treatment group? Obviously if you are tracking them, they are seeing doctors. If they seem suicidal, which can happen at any time to any group, the doctor can deal with it.

I answered your question in good faith.

 

[blutoski]

Your understanding is different than mine. Phases 0 and 1 are unblinded and uncontrolled. They usually have small sample sizes and only last about a month. They are looking for adverse events more so than what we call side-effects. Phase 2 is when the blinded studies with controls start.

Yes, I'd say that's true. I was trying to get more specific about when we'd have better information about side effects and would be able to prepare active placebos.

As you can imagine, it would not be possible to propose an active placebo if there was no information about side effects. When there is information about side effects, an active placebo could be employed to reduce risk of confounding.

 

[Uncayimmy]

Because it's jumping to conclusions. If the benefit is reporting bias, there's no real improvement and it may represent a distorted patient-physician relationship, and it's inappropriate to advise doing this.

Well, the researchers disagree. The study I was talking about used the same medication for nausea and tested for benefits of expectancy. I see no ethical concern whatsoever with selling a med to create some expectancy if it improves their quality of life.

I find this akward. I work in clinical trials. We call it confounding. It is a phrase used to describe all sorts of extraneous variables, one of which is reverse-engineering to defeat blinding.

I've never seen the term "reverse-engineer" used in reference to a patient making a guess about which medication he is on. Perhaps it's a language thing.

Basically. Or they say it's an active placebo but don't go into much detail. Placebo descriptions can lack detail for brevity.

Can you point me to some studies with active placebos?

I pointed out that there was an ethical reason this is rare and becoming rarer.

Actually, it's become more common since that recommendation was made.

I answered your question in good faith.

You still haven't demonstrated that it's unethical to have a third group - no treatment - while it's ethical to have a sham treatment group. Any argument against the former applies to the latter, at least that I can dream up.

 

[Alan]

http://www.nejm.org/doi/full/10.1056/NEJMoa042580

In this randomized, double-blind, active placebo–controlled, four-period crossover trial, patients received daily active placebo (lorazepam), sustained-release morphine, gabapentin, and a combination of gabapentin and morphine — each given orally for five weeks. The primary outcome measure was mean daily pain intensity in patients receiving a maximal tolerated dose; secondary outcomes included pain (rated according to the Short-Form McGill Pain Questionnaire), adverse effects, maximal tolerated doses, mood, and quality of life.

 

[Uncayimmy]

http://www.nejm.org/doi/full/10.1056/NEJMoa042580

Yes, I've seen it in reference to pain studies before. They use lorazepam because it's known not to have an analgesic effect, but, it does provide a bit of sedation. And in this care we're looking at comparing drug combinations, so I'm actually a bit surprised they even bothered with a control.

 

[blutoski]

Well, the researchers disagree. The study I was talking about used the same medication for nausea and tested for benefits of expectancy. I see no ethical concern whatsoever with selling a med to create some expectancy if it improves their quality of life.


I've never seen the term "reverse-engineer" used in reference to a patient making a guess about which medication he is on. Perhaps it's a language thing.

Probably a language thing. "confound" is my go-to word for this problem.

The most outrageous example of 'reverse engineering' that I've seen was a patient who was a lab tech at a local college and ran his scrip through a mass spec. He knew exactly what was in it, down to the dosage. I only heard through a total coincidence because one of his coworkers was a mutual friend.

Can you point me to some studies with active placebos?

I'm interested myself, now. I'll tackle this when I get home. It's probably one of those things that peaked in the 90s, because most hiv and psych medications don't have short-term side effects that are so severe anymore.

You still haven't demonstrated that it's unethical to have a third group - no treatment - while it's ethical to have a sham treatment group. Any argument against the former applies to the latter, at least that I can dream up.

Gotcha: that's my sloppy wording. Both groups might be hard to justify for the condition you suggested (depression). The comparison of these two groups would be rare because each of the arms would be frowned upon, much less both.

Having said that, it's possibly OK for mild depression, dysthymia. Maybe bipolar II if the last major depressive was years ago and pt is consistently cycling between hypomanic and mild depression. Nontreatment and sham treatment could possibly pass an ethics committee depending on the selection critieria and a plan for managing dropouts in the analysis.

I think one of the challenges with this is the dropout rate. An unmanaged minor depressed group will have patients 'drop out' when they have a major depression episode. They will be either treated with therapy or meds (or both) and the question is: are they to be treated as dropouts now that they're no longer really in a non-treatment control group? Maybe binary endpoints? But the trend is toward spectrum nowadays.

 

[Percent]

Because it's jumping to conclusions. If the benefit is reporting bias, there's no real improvement and it may represent a distorted patient-physician relationship, and it's inappropriate to advise doing this.

Well, the researchers disagree. The study I was talking about used the same medication for nausea and tested for benefits of expectancy. I see no ethical concern whatsoever with selling a med to create some expectancy if it improves their quality of life.

I thought I'd toss in a few more links about the power of demand characteristics on experimental outcomes. Link. Link. To me, it is obvious that much of the effect of placebo can be attributed to expectancy. It isn't at all uncommon for experimenter's biases to corrupt dependent measures, either. Both experimenter/doctor and subject/patient can be impacted by expectancies. Gee, both the doctors and my cousin believed she only had an ovarian cyst, not a ruptured appendix, and both were willing to say she was getting better with no treatment .

I certainly don’t think it is unethical for qualified providers to use expectancy to improve clinical outcomes. As I reported, my brother often does just that, usually to avoid negative aspects of invasive treatment and to wait for regression to the mean when he knows there is a reasonable chance things can resolve themselves in a short time.

I have concerns about a couple other things. What happens when physicians are misled to believe a treatment is clinically effective, beyond placebo, when it actually isn’t? Could good doctors be led to take unnecessary risks with the care of their patients by placing too much confidence in some effect? Of even greater concern, what happens when placebo is leveraged by unscrupulous (or even well-meaning) sorts who don’t have the medical training and/or resources to handle situations that could arise, such as misdiagnosis of a life-threatening condition?

Yeah, I’d love to see more research as well. I can find studies on dopamine that include control groups separate from placebo treatment showing an expectancy effect within the dopamainergic system but there aren’t as many studies to be found with transmitters targeted by common anti-depressasnts

Ftfm. Those typos have been driving me BGC. I have a profound ophthalmological disability and minutiae often escape me. So, often, do boulders.

I did find a no-treatment and placebo-controlled study of anti-depressant use (Lexipro) that employed a fairly large sample size. I believe the study was presented at an annual meeting of The European College of Neuropsychopharmocology. Here’s a link to a summary. The study, conducted in Israel, was designed to assess the impact of CBT, prolonged exposure, and anti-depressant therapies on PTSD.

“Of those who were assessed, 398 had qualifying symptoms; 298 of those subjects were randomized into the trial, with treatment beginning within 20 days of the traumatic event. Treatment arms included 12 weekly sessions of cognitive therapy, 12 weekly sessions of prolonged exposure, 12 weeks of blinded treatment with placebo or 20 mg of the SSR1 escitalopram (Lexapro), or 12 weeks on a waiting list. The group on the waiting list received only weekly telephone calls.

At the end of the treatment period, all groups showed significantly reduced rates of PTSD. The cognitive therapy and prolonged exposure groups had the lowest incidence of PTSD (18% and 21%, respectively).

Although escitalopram reduced the incidence of PTSD to 61%, it was not any more effective than taking a placebo or being placed on the waiting list; the incidence of PTSD was reduced to 59% with placebo and to 57% for those on the waiting list.”

Thus, for this study of PTSD, Lexipro was not effective beyond placebo or no treatment.

I've always been fascinated by placebo, but I didn't think I could get this many hours of supposed entertainment or knowledge from it. Whee.


Anne

 

[Uncayimmy]

I certainly don’t think it is unethical for qualified providers to use expectancy to improve clinical outcomes. As I reported, my brother often does just that, usually to avoid negative aspects of invasive treatment and to wait for regression to the mean when he knows there is a reasonable chance things can resolve themselves in a short time.

I have concerns about a couple other things. What happens when physicians are misled to believe a treatment is clinically effective, beyond placebo, when it actually isn’t? Could good doctors be led to take unnecessary risks with the care of their patients by placing too much confidence in some effect? Of even greater concern, what happens when placebo is leveraged by unscrupulous (or even well-meaning) sorts who don’t have the medical training and/or resources to handle situations that could arise, such as misdiagnosis of a life-threatening condition?

It all needs to be evidence based. Interestingly, I've seen studies where negative expectancy makes things better. In other words when told about the difficulties they can expect to face, they fare better in regards to nausea.

http://www.ncbi.nlm.nih.gov/pubmed/16738082
METHOD: Seventy-five participants were assigned to one of three groups. Positive-expectancy group participants were given placebo pills that would allegedly protect them against the development of nausea and motion sickness. Negative-expectancy group participants were given the same pills as nocebos; they were led to believe there was a tendency for them to make nausea somewhat worse. Placebo-control group participants were told the pills were indeed placebos that would have no effect whatsoever.

RESULTS: Subjective symptoms of motion sickness were significantly lower among negative-expectancy group participants than positive-expectancy and placebo-control group participants (p<0.05). Gastric tachyarrhythmia, the abnormal stomach activity that frequently accompanies nausea, was also significantly lower among negative-expectancy group participants than positive-expectancy and Placebo-Control Group participants during drum rotation (p<.05) [corrected]

CONCLUSIONS: Inducing negative expectations through nocebo administration reduced nausea and gastric dysrhythmia during exposure to provocative motion, whereas positive placebos were ineffective for preventing symptom development. That manipulation of expectation affected gastric physiological responses as well as reports of symptoms, suggests an unspecified psychophysiological mechanism was responsible for the observed group differences. These results also suggest that patients preparing for difficult medical procedures may benefit most from being provided with detailed information about how unpleasant their condition may become.

 

[blutoski]

I'm interested myself, now. I'll tackle this when I get home. It's probably one of those things that peaked in the 90s, because most hiv and psych medications don't have short-term side effects that are so severe anymore.

OK: so, yes, active placebo peaked for HIV meds in the '90s and is rarer today.

OTOH, it's becoming more standard for psych clinical trials and research.

The first example I came across was that whenever a tricyclic antidepressant is compared against other types, an antimuscarinic is introduced to the other type to improve blinding. eg: Atropine, to create dry mouth &c. It's the same receptor, so it's not just simulating the effect, which is the preferred way for blinding.



I also touched base with my wife about non-treatment groups for depression, and she confirmed that nontreatment for major depression would be a no-go from an ethics committee. It's witholding the standard of care.

However, it's possiblle to get approval for non-treatment of mild depression in some situations. But it's probably rare.

As a result, a pure non-treatment group for depression studies would be rare. There may be a "grouptherapy / CBT only" group or something called "real-world" protocol where the patient may or may not be on another med, but just not on the experimental med. The intention with the latter is to acknowledge that patients often self-medicate with recreational drugs anyway. (My wife hates real-world studies, because the results are consistently irreproduceable. She feels they are a fad that is in decline.)

 

[quarky]

Meanwhile, back at the ranch:

If depression was a disease, we are losing the battle.
If depression was a gold-mine, we are winning the battle.

 

[Uncayimmy]

I also touched base with my wife about non-treatment groups for depression, and she confirmed that nontreatment for major depression would be a no-go from an ethics committee. It's witholding the standard of care.

As far as I can tell they have always used placebos and continue to do so. Just go to PubMed and search Antidepressants Placebo. In fact there's a lot of controversy over whether antidepressants work better than placebos. Only in the most severe cases does there seem to be any clinical benefit. One of the issues is that the drug companies try to get as many severely depressed subjects as possible because it makes the numbers look better.

 

[blutoski]

As far as I can tell they have always used placebos and continue to do so. Just go to PubMed and search Antidepressants Placebo. In fact there's a lot of controversy over whether antidepressants work better than placebos. Only in the most severe cases does there seem to be any clinical benefit.

Well, I think we're saying the same thing. Basically, that ship has sailed, and the new insight into MDD vs other Axis I mood disorders has updated the ethics committee's decisons vis a vis MDD clinical trials.

Major Depressive Disorder should not treated the same as "depression" here at the end of 2010. Knowing that meds work for MDD, but probably not for the other significant mood diagnoses on Axis I (Dysthymia, DD-NOS) contributes to their being considered the standard of care for MDD.

One of the issues is that the drug companies try to get as many severely depressed subjects as possible because it makes the numbers look better

Yeah, that's what makes a bad study, though. Psychiatrists would be interested in whether drug X treats condition Y. MDD is not the same as 'a patient who reports feeling depressed' and an MD would consider this when evaluating the study.

This sould be outlined in the patient selection criteria. If I were reviewing literature for Dysthymia, I should search for studies selecting subjects with Dysthymia.

 

[blutoski]

It all needs to be evidence based.

Also science-based, though. One challenge with EBM is that this gets gamed a bit by the alties and big pharma who seem to have a machine that spits out studies all day.

That's part of the reason Drs. Hall and Novella try very hard to promote SBM over mere EBM.

Interestingly, I've seen studies where negative expectancy makes things better. In other words when told about the difficulties they can expect to face, they fare better in regards to nausea.

To me, this means these studies are reporting abundant study noise. I always ask for independent replication and hope to throw them into funnel plots to see if any of this is real. There's not a lot of good research out there, unfortunately.

 

[Percent]

It all needs to be evidence based. Interestingly, I've seen studies where negative expectancy makes things better. In other words when told about the difficulties they can expect to face, they fare better in regards to nausea.

So, "hang on it's going to be a bumpy ride." Well, that makes sense, too, and could beat being told “this is only going to sting a little” when in reality a doctor is about to inflict massive pain. With the former, your brain can respond to the call (a la Pavlov's bell) with some anticipatory endorphins, or whatever else might help combat the situation.

One interesting (and unfortunate) thing for some is that it appears not everyone is a placebo responder to the same degree. So, some of us can't muster up the goods when we need to.

As for the evidenced-based needs, it does look like the research community is more open to formal investigation of placebo these days. This, I assume, means more scientific investigation, perhaps using new tools.

For example, The New York Times did a piece on Tor Wager's work last June. Dr. Wager has applied a variety of techniques to look at placebo for pain, though his work has clear applications beyond pain. He collaborates with many others, in particular Fabrizio Benedetti (placebo expert), Helen Mayberg (heavy hitter in depression), and Kevin Ochsner (major contributor in social cognitive neuroscience and emotion regulation).


Anne

 

[Uncayimmy]

To me, this means these studies are reporting abundant study noise. I always ask for independent replication and hope to throw them into funnel plots to see if any of this is real. There's not a lot of good research out there, unfortunately.

Well, yeh, replication is always key. It's an avenue worth exploring, but ultimately there's no money in it.

Speaking about nausea specifically, the studies I've seen looked at two things. One is the expectancy of the person coming in for treatment. Much of that is based on past experience, but not necessarily with chemotherapy. You know, some people are more likely to have "weak" stomachs. It's also based on what they heard about the treatment. The other thing they have looked at is the expectancy from the perspective what the clinician tells people both in terms of the drug prescribed and what the person can expect.

It would very interesting to me to see what patterns there are, if any, in the above. There's always time for a brief questionnaire in the waiting room. If a little bit of info could help a doc say the right things to improve how a patient feels during treatment, I'd be all for it.

Some of us know ourselves pretty well. I want to hear it straight, no punches pulled. Other people just don't like to hear it.