The Placebo effect - real or imaginary?

[Pebble]

The placebo effect has never been demonstrated in a DBRCT as superior to the null hypothesis - so does it exist?

We know that those receiving placebo in a clinical trial seem to do better than similar patients in a registry - but this could be due to the more rigorous inclusion & exclusion criteria applied to those who enter the trial, augmented by superior background care for those ostensibly actively treated.

We know that some ineffective treatments are superior to other comparators that are also ineffective - such as sham surgery versus 'best medical therapy'

So the question is - does the degree of patient 'buy in' to the intervention being offered materially affect their response to therapy? If yes, this means that Placebo therapy is not inactive and there are different levels of placebo response.

Are all placebos equally ineffective? in which case observed differences between trials are due to natural variation of event rates. If this is so where differences are observed this may be because the comparator used actually harm the patient producing measurably worse outcomes. Alternately, the bulk of differences due to inadequate concealment are due to researcher bias. If so (assuming the patient accepts both treatments as potentially valid) any observed differences between ineffective therapies is consiously or unconsiously 'manufactured' by those carrying out the studies.

If the latter position is correct, then where hard endpoints are chosen (death, biochemical response assessed independently) concealment is unnecessary. It also follows that surgical interventions can never be demonstrated as superior to medical therapy except in respect of hard endpoints, and meta-analyses of inadequately concealed comparators belong in the bin.

[chaggle]

The placebo effect has never been demonstrated in a DBRCT as superior to the null hypothesis - so does it exist?

We know that those receiving placebo in a clinical trial seem to do better than similar patients in a registry - but this could be due to the more rigorous inclusion & exclusion criteria applied to those who enter the trial, augmented by superior background care for those ostensibly actively treated.

We know that some ineffective treatments are superior to other comparators that are also ineffective - such as sham surgery versus 'best medical therapy'

So the question is - does the degree of patient 'buy in' to the intervention being offered materially affect their response to therapy? If yes, this means that Placebo therapy is not inactive and there are different levels of placebo response.

Are all placebos equally ineffective? in which case observed differences between trials are due to natural variation of event rates. If this is so where differences are observed this may be because the comparator used actually harm the patient producing measurably worse outcomes. Alternately, the bulk of differences due to inadequate concealment are due to researcher bias. If so (assuming the patient accepts both treatments as potentially valid) any observed differences between ineffective therapies is consiously or unconsiously 'manufactured' by those carrying out the studies.

If the latter position is correct, then where hard endpoints are chosen (death, biochemical response assessed independently) concealment is unnecessary. It also follows that surgical interventions can never be demonstrated as superior to medical therapy except in respect of hard endpoints, and meta-analyses of inadequately concealed comparators belong in the bin.

Sorry Pebble, it might be me but it could be your terminology.

How can "some ineffective treatments [be] superior to other comparators that are also ineffective"? Surely if a treatment is ineffective (i.e. it has NO effect) then it cannot be superior (or inferior) to another treatment that also has no effect.

In the same vein "Are all placebos equally ineffective?" Are there degrees of ineffectivity? And are placebo's ineffective? It is called the placebo effect.

Have I misunderstood or do you need to find a word other than effective?

[Pebble]

The question in a trial is whether the active ingredients confer measurable benefit. So the comparator must to the investigator and recipient be indistinguishable from each other to eliminate bias.

This does not mean that the comparator is itself inactive.

We assume the 'placebo' effect is real when we partly ascribe the benefits of CAM to interpersonal interactions and the social contract. We know that our response in given situations affect the probability of success, and it is believed that hope plays a role in recovery.

If all placebos are equal then the only role of the placebo arm is concealment. But then you have to explain why surgical procedures are almost always associated with a temporary improvement in symptoms, where as tablet therapy is generally less effective unless the ingredients have a provable benefit.

It also means that differences that have been consistently demonstrated between CBT and less obviously reassuring therapies are due to researcher bias, but that the same researchers fail to exhibit the same bias when comparing CBT to other similarly reassuring psychological interventions - without concealment. There are other examples where the researcher has the potential to know which group the patient is in, but only adequate concealment of the patient has been sufficient to neutralise the apparent benefit (sham PFO closure). So it is unclear to me that the patients response is an entirely inactive agent in the findings of these studies.

[wheels5894]

Was not a the trails of Prozac a demonstration of this? The placebo patients did nearly as well as the people taking the drug. This would seem to imply that either the placebo effect is a useful one in psychological conditions or that Prozac is ineffective, more or less. That the medical community started to prescribe Prozac suggests the former is the more likely explanation so I would suggest that there is a demonstrated placebo effect shown in these trials.

[Pebble]

We seem to forget that fluoxetine made its name in the treatment of major depression, where efficacy versus placebo and tolerability versus the old TCAs was established. The problem arose because of attempts to replicated these findings in mild derpression, OCD, etc. It has subsequently been replaced in major depression by more effective agents.

J Clin Psychiatry. 1985 Mar;46(3 Pt 2):26-31.
A comparison of fluoxetine, imipramine, and placebo in patients with major depressive disorder.

Cohn JB, Wilcox C.
Abstract

The efficacy and safety of fluoxetine were compared with those of imipramine and of placebo in a 6-week randomized double-blind parallel study of patients with major depressive illness. Mean values for all efficacy measurements were improved over baseline with fluoxetine and imipramine treatment (p less than .001). More fluoxetine patients completed the study than did imipramine or placebo patients. Predominant adverse experiences reported by imipramine patients were dry mouth and dizziness/lightheadedness. Predominant adverse experiences reported by fluoxetine patients were drowsiness/sedation and excessive sweating. In a subsequent 48-week open-label study, the predominant adverse experience in the fluoxetine group was excessive sweating and in the imipramine group was still dry mouth. In this study, fluoxetine relieved the symptoms of major depressive illness effectively and significantly better than placebo and was better tolerated than imipramine.