Antidepressants Only Helpful for Severly Depressed

[Malerin]

Antidepressants no help in milder cases
Study: Meds no better than placebos for all but most severely depressed

http://www.msnbc.msn.com/id/34712755/ns/health-mental_health/

[Uncayimmy]

That's depressing.

Perhaps this helps differentiate between chemical issues and emotional issues.

[Cuddles]

So if the drugs work, then you're no longer depressed, while if they don't work it means you weren't all that depressed in the first place and therefore have nothing to be depressed about. Everyone's a winner!:)

[T.A.M.]

Antidepressants no help in milder cases
Study: Meds no better than placebos for all but most severely depressed

http://www.msnbc.msn.com/id/34712755/ns/health-mental_health/

Too bad your link, so far as I can see, does not link to the actual metanalysis. Not finding anything on an initial google scholar search in terms of a recent metanalysis.

Anyone?

TAM:)

[Ivor the Engineer]

Too bad your link, so far as I can see, does not link to the actual metanalysis. Not finding anything on an initial google scholar search in terms of a recent metanalysis.

Anyone?

TAM:)

http://jama.ama-assn.org/cgi/content/short/303/1/47?rss=1

[Professor Yaffle]

A meta-analysis in 2008 (kirsch et al)came to a similar conclusion. This blog did some interesting re-analyses of the data (earlier posts are at the bottom) with lots of good links:

http://pyjamasinbananas.blogspot.com/search?q=kirsch

[fls]

Antidepressants no help in milder cases
Study: Meds no better than placebos for all but most severely depressed

http://www.msnbc.msn.com/id/34712755/ns/health-mental_health/

http://jama.ama-assn.org/cgi/content/abstract/303/1/47

That's not actually what the meta-analysis found. What it found was that the benefit from anti-depressants increased as the baseline severity of the depression increased. That is, the effect of anti-depressants was small with mild to moderate depression and larger with more severe depression.

Linda

[Professor Yaffle]

http://jama.ama-assn.org/cgi/content/short/303/1/47?rss=1

I see that it is based on only 6 trials which surprises me. Could someone with access to the full paper check see if they can find out which of their exclusion criteria seem to be responsible for such a small selection of the available data? Also which drugs were included in the 6 trials - were they mainly SSRIs, or were some of them on newer drugs such as venlafaxine and mirtazepine?

[fls]

I see that it is based on only 6 trials which surprises me. Could someone with access to the full paper check see if they can find out which of their exclusion criteria seem to be responsible for such a small selection of the available data? Also which drugs were included in the 6 trials - were they mainly SSRIs, or were some of them on newer drugs such as venlafaxine and mirtazepine?

The bulk of the studies were excluded because they lacked patient level data on baseline severity or because patients were excluded on the basis of a placebo washout (i.e. in some trials, patients were given placebo for a few weeks prior to the start of the trial and those patients who showed improvement were excluded in order to increase the power of the study). This meta-analysis did not want to exclude people who responded well to placebo.

Three trials used Imipramine and three used paroxetine.

Linda

[Moochie]

Antidepressants no help in milder cases
Study: Meds no better than placebos for all but most severely depressed

http://www.msnbc.msn.com/id/34712755/ns/health-mental_health/


ETA: I've been on Pristiq (Effexor 2.0) for two weeks now. I'm listening to a lot more Leonard Cohen.



M.

[Professor Yaffle]

The bulk of the studies were excluded because they lacked patient level data on baseline severity or because patients were excluded on the basis of a placebo washout (i.e. in some trials, patients were given placebo for a few weeks prior to the start of the trial and those patients who showed improvement were excluded in order to increase the power of the study). This meta-analysis did not want to exclude people who responded well to placebo.

Three trials used Imipramine and three used paroxetine.

Linda

Thanks Linda. The kirsch study I mentioned above showed a non-significant trend for the newer ADs to be more effective and a later Lancet paper found differences in efficacy among anti depressants: http://www.thelancet.com/journals/lancet/article/PIIS0140-6736%2809%2960046-5/fulltext

[Professor Yaffle]

A Maudsley debate on this subject from 2008:

http://www.iop.kcl.ac.uk/podcast/?id=238&type=item

[Thunder]

so, the less depressed you are, the less effective the anti-depressants are.

sounds like a good thing to me.

:)

[Ivor the Engineer]

The bulk of the studies were excluded because they lacked patient level data on baseline severity or because patients were excluded on the basis of a placebo washout (i.e. in some trials, patients were given placebo for a few weeks prior to the start of the trial and those patients who showed improvement were excluded in order to increase the power of the study). This meta-analysis did not want to exclude people who responded well to placebo.

Three trials used Imipramine and three used paroxetine.

Linda

If you believe that you'll believe anything.

Clinical trials which are planned and executed in such a way that their results are less generalizable to real-world clinical practice is rather bizarre, don't you think?

[Moochie]

And this (http://link.email.slate.com/r/BZYORR/SI4D/2618DS/NFZXU/KEVZ9/RF/h).


M.

[Ivor the Engineer]

so, the less depressed you are, the less effective the anti-depressants are.

sounds like a good thing to me.

:)

Unfortunately, many of the side effects are just as potent no matter how depressed a person is.

[Professor Yaffle]

Unfortunately, many of the side effects are just as potent no matter how depressed a person is.

I use this as a measure of how well I am. Once the side effects start bugging me more than the symptoms of depression, I know I am pretty much better.

[fls]

I'm listening to a lot more Leonard Cohen.

M.

:)

[fls]

Thanks Linda. The kirsch study I mentioned above showed a non-significant trend for the newer ADs to be more effective and a later Lancet paper found differences in efficacy among anti depressants: http://www.thelancet.com/journals/lancet/article/PIIS0140-6736%2809%2960046-5/fulltext

The JAMA study appears to have been undertaken specifically to address the issues raised by the Kirsch study. The bulk of the background section discusses the limitations of the Kirsch study and what additional detail would be useful. But it wasn't designed as a comparison of different types of anti-depressants.

Linda

[Elaedith]

If you believe that you'll believe anything.

Clinical trials which are planned and executed in such a way that their results are less generalizable to real-world clinical practice is rather bizarre, don't you think?

'Increase the power' means to increase the statistical power to detect an effect if it exists. What does that have to do with generalizability? The effect is there whether it is detectable or obscured.

[Dancing David]

Well it looks as though they are using the BDI (Beck depression Inventory) which is not really a good scale for judging response to treatment

Using a scoring system for depression where a diagnosis of 24 or above indicates a very severe case, the researchers said patients treated with drugs saw their scores drop by 13 points, compared to a drop of 9 points for those given a placebo.


There is also the charming belief in placebos


The so-called placebo effect is powerful in treating depression, where people believe they are helped even though they are taking an inactive sugar pill, DeRubeis said.


So without looking at the measures which most likely were not a very fine tuned scale, then it is hard to say what the effect for moderate depression is.


ETA:
It was not the BDI but the HAMILTON DEPRESSION RATING SCALE - 24 item
http://www.medafile.com/cln/HDRS.html

So the media report got it wrong on the scale as well, it is a scale of 24 items some of which are scored 0-4 and others which are scored 0-2

and it is a strange scale because it tops out at 15! So even though someone might score the maximum of 83 point, they are scored as a 15. So this is crappy scale for telling anything about response to treatement. It should be scored from 0-83
instead
0 - 4 normal, depending on age, education, complaints
5 - 8 mild
8 - 11 moderate
12 - 15 severe


So this is almost meaningless if it is the version of the scale that they used. Especially if we are looking athe low end of the scale IE the moderate to mild, if someone scored that I would probably be reluctant to give them more than an adjustment disorder, when I did assesments.

(It might not be.)

this is the Palo Alto VA version

[Ivor the Engineer]

'Increase the power' means to increase the statistical power to detect an effect if it exists. What does that have to do with generalizability? The effect is there whether it is detectable or obscured.

The purpose of using a placebo washout in a clinical trial is to make the effect size appear much larger than it would be in clinical practice for marketing purposes.

[fls]

'Increase the power' means to increase the statistical power to detect an effect if it exists. What does that have to do with generalizability? The effect is there whether it is detectable or obscured.

Increasing internal validity can increase the power as you describe. Studies tend to progress from establishing internal validity to establishing external validity (generalizability). Of course, I only think that because I am a clueless moron. ;)

Linda

[Ivor the Engineer]

Increasing internal validity can increase the power as you describe. Studies tend to progress from establishing internal validity to establishing external validity (generalizability). Of course, I only think that because I am a clueless moron. ;)

Linda

Nonsense. Most people who have not got money to burn design their experiments so they have reasonable internal and external validity.

[fls]

Nonsense. Most people who have not got money to burn design their experiments so they have reasonable internal and external validity.

Right. I don't know what I am talking about.

Linda

[Ivor the Engineer]

Right. I don't know what I am talking about.

Linda

In this case, yes.

http://cjasn.asnjournals.org/cgi/content/full/1/6/1360

Internal and External Validity

Ideally, trials are designed and conducted both to minimize bias (i.e., have high internal validity) and to be relevant to a wide but defined population (i.e., have high external validity, also termed generalizability). In reality, a tradeoff between internal and external validity is commonplace. The determinants of internal validity are easily appreciated and include appropriately implemented randomization procedures, complete follow-up with few missing data, and high adherence with low drop-in and dropout rates. In contrast, the determinants of external validity are less intuitive and rely on a clinical judgment that is related to the therapy under study and the types of candidate patients (15). Trials with high external validity tend to be inclusive and enroll people who might not adhere to the trial protocol (e.g., refusing follow-up, dropping out of intervention). As such, trials with high external validity may sacrifice internal validity. In general, because a biased or poorly executed trial is unlikely to influence policy or practice, even if it has apparent external validity, investigators place a greater emphasis on internal validity than external validity (Figure 6).

But please feel free to continue to hide behind your authority.:)

[Dancing David]

Idealy and what happened may be different especially in a meta-analysis.

[Elaedith]

In this case, yes.

http://cjasn.asnjournals.org/cgi/content/full/1/6/1360



But please feel free to continue to hide behind your authority.:)

Did you read what you highlighted? It says that ideally trials maximise both internal and external validity, but in reality, internal and external validity trade off.

That is the reality. You cannot consider the generalizability of an effect until it has been reliabilty identified. Detecting the effect often requires procedures that maximise statistical power. These procedures often require conditions that depart from what would occur in 'real life'.

This is pretty basic stuff and it applies to all fields that use experiments.

[fls]

In this case, yes.

http://cjasn.asnjournals.org/cgi/content/full/1/6/1360

But please feel free to continue to hide behind your authority.:)

What's your problem? I agreed with you. Like I said, I have no idea what I'm talking about. Internal and external validity and the design of clinical trials are not something I have any experience with, nor have I taken any graduate level classes on these subjects. I have no authority in this regard.

Linda

[Ivor the Engineer]

Did you read what you highlighted? It says that ideally trials maximise both internal and external validity, but in reality, internal and external validity trade off.

That is the reality. You cannot consider the generalizability of an effect until it has been reliabilty identified. Detecting the effect often requires procedures that maximise statistical power. These procedures often require conditions that depart from what would occur in 'real life'.

This is pretty basic stuff and it applies to all fields that use experiments.

If an effect is not clinically significant it doesn't matter that it's been identified. In this case all that's been done is a load of money has been wasted finding out the drug doesn't work very well. Well not quite. It makes the effect size look larger than it really is in practice. I wonder who might be interested in doing such a thing?

Clinical significance is determined by comparing against a placebo group made up of people similar to those who the drug is designed to treat in practice.

This is pretty basic stuff.

[Ivor the Engineer]

What's your problem? I agreed with you. Like I said, I have no idea what I'm talking about. Internal and external validity and the design of clinical trials are not something I have any experience with, nor have I taken any graduate level classes on these subjects. I have no authority in this regard.

Linda

Your so sexy when your in a huff.

Ding dong.

[fls]

Your so sexy when your in a huff.

Ding dong.

I've just decided to come clean. I made up all that other stuff before I realized how things work around here. I hate to break it to you, but I'm a 30 year old guy who's never had a girlfriend, living in my parent's basement. I stole a picture of my best friend's girlfriend to use as an avatar.

I feel much better now that that's off my chest.

Doug

[Ivor the Engineer]

I've just decided to come clean. I made up all that other stuff before I realized how things work around here. I hate to break it to you, but I'm a 30 year old guy who's never had a girlfriend, living in my parent's basement. I stole a picture of my best friend's girlfriend to use as an avatar.

I feel much better now that that's off my chest.

Doug

Apart from the avatar* sounds like we have a lot in common.

Robert


*I don't have any friends. Can I share yours?

[JoeTheJuggler]

So if the drugs work, then you're no longer depressed, while if they don't work it means you weren't all that depressed in the first place and therefore have nothing to be depressed about. Everyone's a winner!:)

That's pretty close to what I was thinking. Mild depression isn't really an illness, but major depression is. The drugs work for the real illness.

So. . . if you're not sick, you can take herbals or whatever and enjoy a placebo effect.

[JoeTheJuggler]

I've just decided to come clean. I made up all that other stuff before I realized how things work around here. I hate to break it to you, but I'm a 30 year old guy who's never had a girlfriend, living in my parent's basement. I stole a picture of my best friend's girlfriend to use as an avatar.

I feel much better now that that's off my chest.

Doug

And I bet you're not even a real doctor!
:)

[fls]

Apart from the avatar* sounds like we have a lot in common.

Robert


*I don't have any friends. Can I share yours?

He's pretty much out the door already. You know what they get like when they get a girlfriend.

Doug

[fls]

And I bet you're not even a real doctor!
:)

No, but I play one on WoW. And my mom's a radiology tech.

Doug

[Ivor the Engineer]

Wouldn't it be a more efficient use of humans to plan a trial with a larger N to increase the power rather than exclude placebo responders, which severely compromises the external validity of the trial?

[kellyb]

I agree with Linda. Just think about what the press releases must have said. The agenda (if there was one) was to prove that antidepressants don't work for moderate and mild depression.

Cynicism fail, Ivor. :)

[Ivor the Engineer]

I agree with Linda. Just think about what the press releases must have said. The agenda (if there was one) was to prove that antidepressants don't work for moderate and mild depression.

Cynicism fail, Ivor. :)

:confused:

ETA: The press releases of such trials would say “X effective for treating depression in placebo controlled trials.”

[Dancing David]

That's pretty close to what I was thinking. Mild depression isn't really an illness, but major depression is. The drugs work for the real illness.

So. . . if you're not sick, you can take herbals or whatever and enjoy a placebo effect.

It is a matter of degree of a spectrum of various factors. There are so many issues, what impacts one person's functioning is not what impacts another.

There is also what you can call the eleventh hour syndrome. People do not seek mental health treatment when the barn door is left open, they seek treatment after the horse has run away and the barn has burned down.

It is very important to use specific behavioral markers in assessments (which is a problem with most assessments and scale tools).

This is pretty typical of someone who comes in saying something like :"I think I may be a littler depressed."

:Interviewer
:Client

I-Do you have problems sleeping?
C: No, not really, some times I wake up early.

I-What time do you fall asleep?
C-That depends.

I-tell me more, please, what sort of patterns are there?
C-Sometimes I fall asleep at eleven, sometimes two.

I-How often do you fall asleep at eleven?
C-About once a week.

I-How often at two?
C-Most of the time, except when I can't.

I-How late do you fall asleep then?
C-Hard to say, sometimes three sometimes five.

I-In the morning?
C-Yeah.

I_How often do you think you fall asleep after two am?
C-Every other day.

I-What time do you wake up, usually?
C-Eight, I have to be at work at ten.

I-So sometimes you only get three hours of sleep?
C-Yeah.

I-Out of seven days, how many days do you get three hours of sleep?
C-I don't know, three or four.

And that is a very standard pattern for assessment, someone say that "No, not really." about having problems sleeping when they sleep three hours 3-4 nights a week. now that may not be an issue, but is typical of the way these things go, that also does not include early walking, waking during sleep, disturbed sleep, etc.

And it goes this way for mood, social interaction, people will almost always not know how to asses their own patterns, they minimize and get used to living that way.

[kellyb]

:confused:

ETA: The press releases of such trials would say “X effective for treating depression in placebo controlled trials.”

What the media report about trials reflects what the press releases say, and is reflective of what the people who designed the trial want the results to say. If the people who designed the trial and analysed the data were trying to promote antidepressant use in mild and moderate depression, the media would be telling everyone "New study says antidepressants are somewhat effective in mild and moderate depression after all!"

[Ivor the Engineer]

What the media report about trials reflects what the press releases say, and is reflective of what the people who designed the trial want the results to say. If the people who designed the trial and analysed the data were trying to promote antidepressant use in mild and moderate depression, the media would be telling everyone "New study says antidepressants are somewhat effective in mild and moderate depression after all!"

I think see where you are confused. I wasn't criticising the study in the OP, but studies performed with a placebo washout period.

I should apologise to Linda/Doug because this is apparently quite common in trials of neuroleptic drugs.

Linda, on this point you were correct and I was wrong, so I apologise.

However, I still think it is a bizarre practice.

[JoeTheJuggler]

It is a matter of degree of a spectrum of various factors. There are so many issues, what impacts one person's functioning is not what impacts another.

There is also what you can call the eleventh hour syndrome. People do not seek mental health treatment when the barn door is left open, they seek treatment after the horse has run away and the barn has burned down.

It is very important to use specific behavioral markers in assessments (which is a problem with most assessments and scale tools).

This is pretty typical of someone who comes in saying something like :"I think I may be a littler depressed."

:Interviewer
:Client

I-Do you have problems sleeping?
C: No, not really, some times I wake up early.

I-What time do you fall asleep?
C-That depends.

I-tell me more, please, what sort of patterns are there?
C-Sometimes I fall asleep at eleven, sometimes two.

I-How often do you fall asleep at eleven?
C-About once a week.

I-How often at two?
C-Most of the time, except when I can't.

I-How late do you fall asleep then?
C-Hard to say, sometimes three sometimes five.

I-In the morning?
C-Yeah.

I_How often do you think you fall asleep after two am?
C-Every other day.

I-What time do you wake up, usually?
C-Eight, I have to be at work at ten.

I-So sometimes you only get three hours of sleep?
C-Yeah.

I-Out of seven days, how many days do you get three hours of sleep?
C-I don't know, three or four.

And that is a very standard pattern for assessment, someone say that "No, not really." about having problems sleeping when they sleep three hours 3-4 nights a week. now that may not be an issue, but is typical of the way these things go, that also does not include early walking, waking during sleep, disturbed sleep, etc.

And it goes this way for mood, social interaction, people will almost always not know how to asses their own patterns, they minimize and get used to living that way.

I don't understand what the problem is.

Are you suggesting that what distinguishes a serious depression from something not severe in this study is different from how it's done clinically?

My point is that just feeling sad (or having some symptoms that don't rise to the level of a major depression diagnosis)is not the same thing as depression, and this study seems to support that position. I would think the one is an actual illness with some kind of biological/medical origin, and the other is just part of the human condition--life's ups and downs.

[paximperium]

However, I still think it is a bizarre practice.
The reason it is done is to detect the small changes that are not purely due to placebo. The researchers are aware that placebo can have a huge affect(often more than the medication itself) and can heavily bias the study

To power the study, the sample size would have to be larger to find the smaller changes hidden in the placebo noise. This would lead to more cost and difficulty in the studies. Like Linda said, it may be internally valid in the study but once pushed into the general use, it often times loses this effect.

[Dancing David]

I don't understand what the problem is.

Are you suggesting that what distinguishes a serious depression from something not severe in this study is different from how it's done clinically?

My point is that just feeling sad (or having some symptoms that don't rise to the level of a major depression diagnosis)is not the same thing as depression, and this study seems to support that position. I would think the one is an actual illness with some kind of biological/medical origin, and the other is just part of the human condition--life's ups and downs.

I was commenting on the self perception of depression and yes, if they are using something similar to the Hamilton Depression rating scale I linked to earlier, I would say that it is not a good scale at all for measuring the level ofsymptoms of depression.

Here is another one:
http://healthnet.umassmed.edu/mhealth/HAMD.pdf

I will point out that it is an inconsistent scale, that some items are scored 0-2 and other 0-4, now this one does have 21 items and the total score can be 0-67, but you will note that it is a very gross scale on most of the items. T
Take Item 12 where you score 0-2 on
Somatic Symptoms (Gastro Intestinal)
There is no real grade here , you either eat or you don't. There is no discussion of amounts and frequency, weight gain or loss, and if you report problems eating you get a 1, if you need urging from others you get a 2.


To sum up
That is what I mean that this is a gross level scale, just like the BDI (Beck Depression Inventory), it is a good tool for measuring the pretense or absence of depression, but it is a terrible tool for grading levels of response and levels of symptoms.

So I question is we have a scale that says

0-4 no depression
4-8 mild depression
8-12 moderate depression
12-15 severe depression

and we have an effect size of 9 for antidepressant treatment compared to an effect size of 13 for placebo, then we have a real issue is saying anything at all because the effect size is greater than the difference between the measure of mild to severe.

This means that we are using a scale to measure response to treatment that is so gross that is does not measure an effect that the scale can deal with.

They are not talking about a gross scale effect IE presence of absence of depression they are talking about a scale effect that separates mild, moderate and severe depression but the effect measure is larger than the scale they are using.

So it can, could and may be true, but the scale used is not capable of such fine tuning and measure. This is the problem with Talking to Prozac as well, the measure of treatment used in a gross level scale (the BDI) and so we don’t really know if it is significant or not.

My comments in response to your statements where i discuss an interview point out the difficulty of measuring a 'mild' vs. a 'severe' depression, most people will tend to way under report.

[calebprime]

I was commenting on the self perception of depression and yes, if they are using something similar to the Hamilton Depression rating scale I linked to earlier, I would say that it is not a good scale at all for measuring the level ofsymptoms of depression.

Here is another one:
http://healthnet.umassmed.edu/mhealth/HAMD.pdf

I will point out that it is an inconsistent scale, that some items are scored 0-2 and other 0-4, now this one does have 21 items and the total score can be 0-67, but you will note that it is a very gross scale on most of the items. T
Take Item 12 where you score 0-2 on
Somatic Symptoms (Gastro Intestinal)
There is no real grade here , you either eat or you don't. There is no discussion of amounts and frequency, weight gain or loss, and if you report problems eating you get a 1, if you need urging from others you get a 2.


To sum up
That is what I mean that this is a gross level scale, just like the BDI (Beck Depression Inventory), it is a good tool for measuring the pretense or absence of depression, but it is a terrible tool for grading levels of response and levels of symptoms.

So I question is we have a scale that says

0-4 no depression
4-8 mild depression
8-12 moderate depression
12-15 severe depression

and we have an effect size of 9 for antidepressant treatment compared to an effect size of 13 for placebo, then we have a real issue is saying anything at all because the effect size is greater than the difference between the measure of mild to severe.

This means that we are using a scale to measure response to treatment that is so gross that is does not measure an effect that the scale can deal with.

They are not talking about a gross scale effect IE presence of absence of depression they are talking about a scale effect that separates mild, moderate and severe depression but the effect measure is larger than the scale they are using.

So it can, could and may be true, but the scale used is not capable of such fine tuning and measure. This is the problem with Talking to Prozac as well, the measure of treatment used in a gross level scale (the BDI) and so we don’t really know if it is significant or not.

My comments in response to your statements where i discuss an interview point out the difficulty of measuring a 'mild' vs. a 'severe' depression, most people will tend to way under report.


you know i luv you david, in the way that a man's man like Roy loves another man, but

measuring the pretense or absence of depression

is just too good to pass up.

sorry:D

actually i read the rest with interest.

[Dancing David]

you know i luv you david, in the way that a man's man like Roy loves another man, but

measuring the pretense or absence of depression

is just too good to pass up.

sorry:D

actually i read the rest with interest.

Ah yes, the PRETENSE of the symptoms of depression.

Curse you speel check!

:dl: on myself

[kellyb]

I think see where you are confused. I wasn't criticising the study in the OP, but studies performed with a placebo washout period.



Durp. Sorry. I wasn't paying attention.
I think a clinical trial could "honestly" be performed with a placebo washout period just for curiosity's sake. In theory.