hypothetical question regarding placebo effect

Assume ABC disease--

1. Treatment X is tested on ABC disease, and in double blinded placebo controlled trials the result is as follows--

The placebo for Treatment X causes people to be absolutely cured 10% of the time.

Treatment X causes people to be absolutely cured 20% of the time.

No treatment group is associated with nobody to be cured.

The difference between Treatment X and its placebo is highly statistically significant in multiple studies.


2. Treatment Y is tested on ABC disease, and in double blinded placebo controlled trials the result is as follows--

The placebo for Treatment Y causes people to be absolutely cured 90% of the time.

Treatment Y causes people to be cured 92% of the time.

No treatment group is associated with nobody to be cured.

The difference between Treatment Y and its placebo is not statistically significant in multiple studies.


Which treatment would you rather have for ABC disease, Treatment X or threatment Y?

Which one should be covered by insurance?

I will state emphatically that I would rather have treatment Y, notwithstanding the fact that the treatment has not been shown to be statistically different from its own placebo effect.

Does anyone here have a different viewpoint?

based on the studies you cited.

ABC disease sounds like a psychological problem.

Don't SSRIs treat psychological problems? Aren't they covered by insurarance (generally)?

although I may be latching onto the terminology in your post.

Remember my post was a hypothetical, and I gave an extreme example to make a point.

If the placebos have different curative powers, they are not placebos!

As an example, sham surgery seems to have a much higher placebo effect than sham pills. Everyone insists on placebo controlled trials. I understand that. When studying acupuncture, for instance, and if it is effective or not, do you think it would be appropriate to test real acupuncture against a sham pill, for example? Most scientists would say NO to that. The studies *would* be easier, though.

If that is the case then the placebo effect in the studies cited becomes irrelevant. You have to adjust the figures given for the variations in efficacy for the different placebos.

If placebo X is less effective than placebo Y then the results are invalid as stated. (apples and oranges)

Even though the first treatment cured only 20% of the patients, and the second treatment cured 92% of the people, only the first treatment was better than the placebo. The second treatment, because of the very high placebo effect, was not statistically different from its placebo. Thus, that study failed, while the first one succeeded.

Given those results, I would prefer the second treatment anyway. Maybe I should have set this up as a poll.

This is an extreme example of course. I just like the idea of comparing two different treatments against each other, and both against a no treatment group. I really don't care much about the size of each placebo group. As you stated, it becomes "irrelevant" when comparing one treatment against another. In real life, outside the scientific world, most people make decisions based on logic. There would probably be a lot of anecdotal reports about the second treatment working, even though it didn't pass scientific muster.

You seem to be talking about a poorly designed experiment with TWO variables. And you seem to be evaluating one variable on the basis of the variation in the other variable. At any rate, you can't get useful conclusions until you isolate the variables.

I am talking about two separate studies done independently, but on the exact same condition. They each test a particular treatment for that condition. Other than that there is no connection between the two studies. Is that clearer?

The placebo is variable, and the treatment is variable. You cannot make any valid conclusions from a comparison of the two studies.

It is as if you were saying that cars with big engines get better mileage on ethanol than cars with small engines get on jet fuel.

To arrive at a valid conclusion, you have to isolate the variables. You might run the same fuel in both cars (to compare fuels), or run the same engine in both cars(to compare engines).

You just can't compare the efficacy of one treatment under one condition to the efficacy of another treatment under another condition.

It is not a valid comparison.

If you just want to compare the efficacy of treatment, then one is 20% effective and the other is 92% effective. But no relevant conclusion can be drawn from the inclusion of the placebo data.

To show how the 92% effective treatment could be tossed aside as being "not significantly better than placebo", and we would only have the 20% effective treatment as "scientifically valid". If the 92% treatment were some sort of pill, I don't think it could be FDA approved, for instance.

to point out that the placebo data is a red herring to distract you from the more important statistical difference in the efficacy of treatment X compared to the efficacy of treatment Y.

That is the only conclusion of statistical significance in the data presented. The use of red herring arguments should create immediate suspicion.

I take it you would pick the second treatment? That is all I asked, really.

A Mother Jones article about this a couple years back. Psychiatrist in NY used the FOIA to look at the results of the clinical tests done to certify the 4 or 5 major antidepressants sold in the US.

He found that the results showed the anti-depressants outperformed the placebos but only by a small amount which could be explained in other ways. For example, the way the effect of the drugs was scored was by self-report, things like the Beck Depression Inventory. So happens that the drugs (as opposed to the placebo) generally have been found to cause people to sleep more and when people sleep more, they report that they are less depressed. The slight difference between the placebo and the drug could be explained by that fact alone probably. Still the FDA certified the drugs. Lot of money involved.

So do these anti-depressants work, or are they themselves having a kind of placebo effect?

Author of the article interviewed one of the people who took part in the study. Turned out that she was in the placebo group but experienced a great improvement in her condition. After the study, like all the participants, she was offered a free supply of the drug for several years if she wanted it. She wanted it even tho she had been given all placebo. Reason? She said if the placebo helped her that much the drug itself would help her even more.

To me, all cure involves the placebo effect to some degree. Doctors should be trained to increase the placebo effect, to increase the faith of the patient in the treatment used. If you believe your doctor is a quack and/or that his treatment is bogus, you're not likely to get better. If you trust your doctor and/or believe in his treatment, you're much more likely to get better.

The SSRIs are only *barely* effective, given the placebo effect. It took very large and expensive trials to show any effect at all. Of course I know people who claim their lives have been changed by them. So was it the placebo effect (which is very large) or the *real* effect (very small)? It is impossible to tell in any individual case, but using the studies, the odds may be in favor of it being the placebo effect.

I will say that getting enough sleep makes a big difference in how someone feels. So I can understand the conjecture re: that.

I would *love* to see the SSRIs head to head against acupuncture twice a week for six weeks.....on moderate depression. Just leave the placebo effect *in* for both, don't control for it, and see how they do against each other!! They both have a large placebo effect, it seems. And, like you, I would love to see every ounce of placebo effect squeezed out of every treatment.

So happens that the drugs (as opposed to the placebo) generally have been found to cause people to sleep more and when people sleep more, they report that they are less depressed.

It isn't the amount of sleep that matters - it's REM sleep (Rapid Eye Movement sleep - when we dream). Specifically, many anti-depressants suppress REM sleep. Whether or not people tend to sleep more on anti-depressants, I don't know. But I do know that anti-depressants reduce the amount of REM that we gent. Anti-depressant effects are also seen when people are intentionally deprived of REM sleep specifically and allowed to go through the rest of their normal sleep architecture (i.e. Stages 1, 2, 3, and 4). In addition, anti-depressant effects are also seen when subjects are completely sleep deprived (though their cognitive function certainly suffers).

we know the mind can heal. I'd like to see a mind technique as powerful as placebo that is proactive: one that the subject is aware of. self-hypnosis, maybe? meditation? I wonder if any studies have been done on that?

it can be used along with conventional treatments.

what difference does it make if the healing comes in a pill or in your head? and if it IS in your head, we shouldn't dismiss it. as far as I'm concerned, it's a medical breakthrough - as dissapointing as the news may be to pharma.

and both your study results would be tossed on the basis of that.

The studies were badly designed or the populations were badly picked, or somebody with an agenda has diddled the results.

when you see two different results from two studies evaluating treatments (or placebo) for seemingly the same disease state, the first thing to check is to see how the study populations in the two studies differ. For instance, does one study exclude patients with several comorbidities (ie, exclude sick patients with lots of problems and only taking relatively healthy patients) while the other study includes all-comers.

A perfect example is two different intra-abdominal studies comparing antibiotic A vs antibiotic B. If study X ends up having a predominance of otherwise healthy 35 year olds with appendectomies, its going to find a high efficacy even if either antibiotic is as good as tap water. If study Y has the same drugs, but it predominantly is made up of 70 year olds with colon perforations, the efficacy rate is going to be lower.