Placebo effect

[ponderingturtle]

You conflated.

If *I* think *I* am doing better, then by definition I am (at least for mental or experiential measures).

If *YOU* think *I* am doing better, then there might be observer bias involved that needs to be eliminated (eg via double blind).

Exactly, but both are part of the placebo effect, as you can not tell which was responceible for the improvement by any data analysis.

Edit: On rereading your statement, I think we were in agreement after all.

 

[joobz]

Becuase you stated that if I think you are feeling better that is an improvement irregardless of how you actual feel.

I never stated that, or implied that. I fully agree with saizai on his view of palcebo

Also dictionaries are not the best choice for technical meaning, unless they are focued on technical meanings

agreed, but even medical dictionaries agree with my statement, an online medical dictionary from the Department of Oncology at Newcastle University.
http://cancerweb.ncl.ac.uk/cgi-bin/omd?query=placebo+effect

And this is wrong, you can not easily differeniate a real outcome from a percieved outcome, especialy when the one perciving the outcome might not be the patient.

Untrue, Take temperature for example. You can state,"this pill will raise your temperature" and then measure quanitatively body temperature, a quantifiable outcome.

No is defines it as a real outcome, not a error of perception and observation, this it says it is real. End of discussion.

I think this is the crux of our disagreement here. The definition doesn't make it true. It just defines the term.

For example. Someone claims to Talk to the dead.
By it's inferred definition, talking to the dead is the ability to communicate with the deceased and relay that information back to the living realatives.

However, everyone here would agree that what is really happening is cold reading. The charlatan is very good at telling people what they want to hear.

"Cold reading" is the spoiler to "talking to the dead", but it does not define it.

You state a mechanism for the placebo effect but it is not the effect.
So, to be most accurate-single blind takes care of placebo effect AND patient fraud.
Double blind removes -placebo effect, patient fraud, Doctor Bias, Doctor Fraud.

Protection agenst fraud is not what it is for though, it is to protect against observer errors from expectation. There are some sorts of fraud that it would make more difficult, but that is not why it is done.

ok, agreed. but again, double blind helps control for many aspects that can't be predicted a priori.

 

[ponderingturtle]

I never stated that, or implied that. I fully agree with saizai on his view of palcebo

You implied that anyone who was recorded as doing better on a placebo, was actualy doing better or at least they thought they where doing better. And you brought up all of that with your "Placebo effect=Real outcome".

The point is that a placebo effect includes many things that are not real outcomes.

agreed, but even medical dictionaries agree with my statement, an online medical dictionary from the Department of Oncology at Newcastle University.
http://cancerweb.ncl.ac.uk/cgi-bin/omd?query=placebo+effect

So what is the term for the systematic unintentioal bias introduced by observers and not patients? That is part of the pacebo effect no matter how someone defines it

Untrue, Take temperature for example. You can state,"this pill will raise your temperature" and then measure quanitatively body temperature, a quantifiable outcome.

Ok now show a placebo effect on temperature.

Also if everying can be measured so readily, then there can be no observer bias, and that means homeopathy works! As you can not get a physcological effect in animals and young children and there are studies that use them and show that homeopathy works, and it can't be the placebo effect by your definition of placebo.

I think this is the crux of our disagreement here. The definition doesn't make it true. It just defines the term.

THe problem is that the placebo effect as it is measured includes many things other than the Real Placebo Effect that you are describing. The point is that just becuase you measure an effect doesn't mean anything has changed, as there are lots of things that can bias such results

For example. Someone claims to Talk to the dead.
By it's inferred definition, talking to the dead is the ability to communicate with the deceased and relay that information back to the living realatives.

However, everyone here would agree that what is really happening is cold reading. The charlatan is very good at telling people what they want to hear.

So what? As long as the poeple believe that someone is being convinced it is all you need right, you don't even have to convince anyone.

"Cold reading" is the spoiler to "talking to the dead", but it does not define it.

You state a mechanism for the placebo effect but it is not the effect.
So, to be most accurate-single blind takes care of placebo effect AND patient fraud.
Double blind removes -placebo effect, patient fraud, Doctor Bias, Doctor Fraud.

But what is refered to as the placebo effect includes all of those. Look at this from the same thing that you referenced, it shows that the placebo is used for bais as well as the so called placebo effect. The problem is that you can't seperate bais from the placebo effect, and as such you have no way of actualy showing that what you are labeling as the placebo effect actual exists, as it could just be bais.

ok, agreed. but again, double blind helps control for many aspects that can't be predicted a priori.

and at the end they are all lumped together into the placebo, and any changed noted are called the placebo effect.

 

[joobz]

You implied that anyone who was recorded as doing better on a placebo, was actualy doing better or at least they thought they where doing better. And you brought up all of that with your "Placebo effect=Real outcome".

I apologize if my phrasing was unclear. I was stating real outcome, meaning the real result as anticipated by the patient. The pain example was perfect for this. But it would also include situations where you can measure a real result

The point is that a placebo effect includes many things that are not real outcomes.

I now understand that you use palcebo as a pseudonym for experimental bias. Fine. If that is an easier definition for you, great.

So what is the term for the systematic unintentioal bias introduced by observers and not patients?

that would be observer bias. experimental bias.

Ok now show a placebo effect on temperature.

Don't need to. It was an example. But I would guess it would be easy to do. Just like increased heart rate, blood pressure. or any measurable physiological parameter that can be varied by metal state.

Also if everying can be measured so readily, then there can be no observer bias, and that means homeopathy works! As you can not get a physcological effect in animals and young children and there are studies that use them and show that homeopathy works, and it can't be the placebo effect by your definition of placebo.

I never stated everything can be measured so easily. I was just contridicting your statement that it's difficult to differentiate observed outcomes and real ones. There are many cases where placebo has actual outcomes that are easy to measure and don't need to take the patient's word for it.

I don't follow homeopathic research or studies at all. I'll assume it's poor experiments and observer bias. As for the controls in animal studies. You'll often give a fake treatment to an animal as control, to control for effects as a result of handling, the carrier vehicle that the drug is in, or surgical procedures. These controls are called the "shams" animals. Not placebos.

The problem is that the placebo effect as it is measured includes many things other than the Real Placebo Effect that you are describing. The point is that just becuase you measure an effect doesn't mean anything has changed, as there are lots of things that can bias such results

Yup, that's why you run controls. But I don't understand why you focus on observed change as not being change. Maybe i'm an empiricist, but if you measured something is differerent, then something is different. Maybe the probe is bad, instruments are wrongly set, maybe the patient's frame of mind is different, maybe buffers are spoiled. But if you measure a change, then a change exists. It's your job as a scientist to evaluate what the cause is for that change and prove it through controls and proper experimental design.

But what is refered to as the placebo effect includes all of those. Look at this from the same thing that you referenced, it shows that the placebo is used for bais as well as the so called placebo effect. The problem is that you can't seperate bais from the placebo effect, and as such you have no way of actualy showing that what you are labeling as the placebo effect actual exists, as it could just be bais.

Yes, the pill (which does nothing pharmacologically) that a is given patient but is told it will is a placebo. But that does not control for the bias the doctor may place into the study and that bias (which you include in the placebo effect) is not at all explained or substatiated by this definition.

 

[ponderingturtle]

I apologize if my phrasing was unclear. I was stating real outcome, meaning the real result as anticipated by the patient. The pain example was perfect for this. But it would also include situations where you can measure a real result

But the measurement can be corrupted by many things. Look how easy it is to create false memories, it would be easy to convince someone it was worse before, and to it unintentionaly if you where not ridgidly controling for it.

A percieved improvement does not nessacarily mean a real improvement, as you can mess with peoples memory and perception quite easily even if you don't mean to.

I now understand that you use palcebo as a pseudonym for experimental bias. Fine. If that is an easier definition for you, great.

No I am saying that procedural bais can not be seperated from the effect that is called the placebo effect.

Don't need to. It was an example. But I would guess it would be easy to do. Just like increased heart rate, blood pressure. or any measurable physiological parameter that can be varied by metal state.

Yes, hence white coat hypertension, but then again those can be effected by so many things. For example, you come into my office the first time and are nervious this raises your blood pressure, as you continue to show up regularly for monitoring you become less nervious and more at ease, this lowers the blood pressure that I measure. We both atribute this change to the drug I am testing on you, and it turns out you got the placebo.

There was no placebo effect in this whole thing as your higher blood pressure at the start opposed to the end was entirely an artifact of the experiment and not a real change in your average blood pressure. There was no placebo effect but my data shows one.

So we continue you on your placebo because it did you so much good, when all that changed is you got over you white coat syndrome.

I never stated everything can be measured so easily. I was just contridicting your statement that it's difficult to differentiate observed outcomes and real ones. There are many cases where placebo has actual outcomes that are easy to measure and don't need to take the patient's word for it.

And please show how many of those can not be becuase of other artifacts of the trial process?

I am not saying that there is no such changes, just that it is almost impossible to say if there was any sort of psychosymatic improvement vs other artifacts of the experimental procedure. So it is hard to say if something was what you are calling the placebo effect or was an error in measurement systemic to the experiment. One is an improvement the other is no change, but you can not tell them apart.

I don't follow homeopathic research or studies at all. I'll assume it's poor experiments and observer bias. As for the controls in animal studies. You'll often give a fake treatment to an animal as control, to control for effects as a result of handling, the carrier vehicle that the drug is in, or surgical procedures. These controls are called the "shams" animals. Not placebos.

Then they are only single blind studies as by the way you are describing it they are doing nothing to control for observer bias.

Yup, that's why you run controls. But I don't understand why you focus on observed change as not being change. Maybe i'm an empiricist, but if you measured something is differerent, then something is different. Maybe the probe is bad, instruments are wrongly set, maybe the patient's frame of mind is different, maybe buffers are spoiled. But if you measure a change, then a change exists. It's your job as a scientist to evaluate what the cause is for that change and prove it through controls and proper experimental design.

Yes, but you can not properly control to test a placebo as you can not remove observer bias in such situations. So any study that conclusively showed your definition of the placebo effect would still have all that bais on controled for.

Yes, the pill (which does nothing pharmacologically) that a is given patient but is told it will is a placebo. But that does not control for the bias the doctor may place into the study and that bias (which you include in the placebo effect) is not at all explained or substatiated by this definition.

And you are not properly removing all bais to see if it has any effect. You might be partialy controling for some observer bias but that would still not show that there was a real effect from the placebo.

 

[joobz]

No I am saying that procedural bais can not be seperated from the effect that is called the placebo effect.

Fair enough, I never claimed to know how to seperate this. I was just against the extrapolation that all double blind controls for is placebo when it is only a part of this issues that can creep up. but again, I do not see how the observer bias is equal to placebo. that's a different issue.


As for the "sham" experiments. I don't tell the in vivo experimenter which of my samples are sham and which are samples. We break the code once the numbers are back. So it still is double blind. kinda.

 

[Jeff Corey]

Experimenter or observer bias can and is quite clearly separated from the placbo effect in double blind studies. As is regression to the mean and other confounding variables in properly designed studies.

 

[saizai]

Question: Have you seen the study where they give people a sugar pill, alternately tell them that it's a stimulant or a depressant, and then have them rate how they feel on standard forms (and have a nurse take their vital stats)?

This design is immune to most forms of error, but nevertheless patients told they're on a stimulant feel stimulated (and their heartrate etc do go up) and patients told they're on a depressant feel depressed (and ditto in reverse).

Classic placebo effect.

 

[Dark Jaguar]

I was under the impression, and my apologies if this has been stated before, that all placebo affect actually does is make people ignore symptoms they have or stop thinking they have symptoms they don't. The disease itself is still as present as ever (if it ever was) and placebo doensn't actually heal them.

 

[blutoski]

I was under the impression, and my apologies if this has been stated before, that all placebo affect actually does is make people ignore symptoms they have or stop thinking they have symptoms they don't. The disease itself is still as present as ever (if it ever was) and placebo doensn't actually heal them.

There are probably several mechanisms that produce the 'effect'. Again: depends on which 'placebo effect' you're measuring: baseline or the effect specific to placebo?

In the former effect (baseline), it's probably mostly regression to the norm. This is a very real improvement that is caused by the fact that people don't volunteer for studies until their symptoms get really bad. Most conditions have oscillating symptoms, so the only way to go is 'better', whether the patient takes the medication or placebo, or nothing, he'll see improvement.

However, if we focus on why, exactly, the group who is given a placeo or sham treatment shows even *more* improvement than this baseline recovery, then three other mechanisms are suggested: the patient is exaggerating the improvement in order to please the doctor, or the patient is unconsciously biasing his response in favour of an expected outcome (just like scientists do when measuring unblinded samples) or thirdly, there's always the possibility that the metrics are really improving because of the placebo.

In order to get better resolution on this, we run experiments to compare placebo and non-treatment groups. This gives us the baseline, against which we may see some improvement due to placebo alone.

Within the placebo group, if we were trying to determine whether the improvent is based on expectation bias or reporting bias, we can blind the results so the patient's opinion is removed. An example of this would be to measure something the patient can't detect or manipulate (such as serum cholesterol levels).

 

[Jeff Corey]

There are probably several mechanisms that produce the 'effect'. Again: depends on which 'placebo effect' you're measuring: baseline or the effect specific to placebo?

In the former effect (baseline), it's probably mostly regression to the norm. This is a very real improvement that is caused by the fact that people don't volunteer for studies until their symptoms get really bad. Most conditions have oscillating symptoms, so the only way to go is 'better', whether the patient takes the medication or placebo, or nothing, he'll see improvement...

No. Statistical regression to the mean is not a confounding factor unless the scientist does not have an appropriate randomly assigned control group. Statistical regression to the mean is not a placebo effect.

 

[blutoski]

No. Statistical regression to the mean is not a confounding factor unless the scientist does not have an appropriate randomly assigned control group. Statistical regression to the mean is not a placebo effect.

Again... depends on what you mean by 'placebo effect': I'm talking about the non-treatment baseline.

Randomly assigning is irrelevant. If everybody who signs up for the study is at the peak of their current cycle, it doesn't matter whether you put them in the nontreatment, placebo, or experimental group... they'll all show improvement.

 

[Jeff Corey]

unlikely, but still is not a confounding variable.

 

[Dark Jaguar]

That's good info to have, thanks. What I get from that though is that placebo is still just a mental "outlook" on what's going on as opposed to somehow willing one's self better.

 

[ponderingturtle]

Fair enough, I never claimed to know how to seperate this. I was just against the extrapolation that all double blind controls for is placebo when it is only a part of this issues that can creep up. but again, I do not see how the observer bias is equal to placebo. that's a different issue.

They are equal because the placebo controls for both effects and you can not determine if the measured improvement was bais or a real improvement. So any given improvement can not be said to be either placebo of bias so any measured placebo effect included both in it.

So when you say "35% of depressed patients got better on placebo" you can't really determine if it was a placebo effect or observer bias that caused it. BUt you label the 35% cured by placebo as the placebo effect.

 

[Jeff Corey]

...So when you say "35% of depressed patients got better on placebo" you can't really determine if it was a placebo effect or observer bias that caused it. BUt you label the 35% cured by placebo as the placebo effect.

. Observer bias, also known as experimenter bias or the Rosenthal Effect, is completely different from the placebo effect. It is eliminated by insuring that the person performing the measurement of the dependent variable is unaware of the subjects' assignment to the placebo or treatment group.

 

[ponderingturtle]

. Observer bias, also known as experimenter bias or the Rosenthal Effect, is completely different from the placebo effect. It is eliminated by insuring that the person performing the measurement of the dependent variable is unaware of the subjects' assignment to the placebo or treatment group.

It is controled for that way, but is it eliminated? Do you remove expectation of patient improvement and the bias it leads to just by being unaware of which group they are in?

They are still looking for improvement in both groups of patient, so it is unsuprising that they would see improvement.

 

[Jeff Corey]

It is eliminated as a confounding variable. A confounding variable is, by definition, correlated with the independent variable (treatment). If it occurs equally in both groups, it is no longer a variable. It's a constant and cannot contribute to any difference between the groups, which is what you're looking for.

 

[ponderingturtle]

It is eliminated as a confounding variable. A confounding variable is, by definition, correlated with the independent variable (treatment). If it occurs equally in both groups, it is no longer a variable. It's a constant and cannot contribute to any difference between the groups, which is what you're looking for.

And all you have done is fold it into the placebo effect, and then treat both of those as a baseline. That works for evaluating the effect of a treatment, but means that you can not determine placebo effect vs observer bias, as both are present in the placebo group. It just removes its effect from being solely on those getting treatment, but it does not elliminate it.

So once again, you can not tell if a particular apparent effect on the placebo group was an actual effect or a result of bais. It does not matter which it was for the group recieving the treatment as they need to be comparied to both effects, but saying "there was a real reduction in X in this placebo group" becomes harder to say.

 

[Jeff Corey]

To answer that question you need a more complcated design.