Friday, November 21, 2008

"Middle-aged women drive rise in U.S. suicides." SSRI Rx?

The headline:

"Middle-aged women drive rise in U.S. suicides; Overall rate grew 0.7 percent between 1999 and 2005, researchers report"

Link to the story (if link gets old, just google the headline):

" "The biggest increase that we have seen between 1999 and 2005 was the increase in poisoning suicide in women — that went up by 57 percent," said Susan Baker, a professor in injury prevention with a special expertise in suicide. "

This adds a second prominent group currently with notably high suicide rates, and possibly also with notably high rates of Rx of SSRIs. gulf war vets are the one group with suicidality being well-documented, but with major media and VA researchers as yet unable to put their finger on the connection with SSRI Rx.

Vet stories:
"Soldier suicide rate may set record again" [sept 2008 story]

I would consider whether the prescription rates in these women have had a similar increase. Suicidality problems had been increasing in adolescents, then the 'black box' warnings came out for adol and ssri and sucide risk, and rx of these meds to adol decreased.

Tuesday, November 18, 2008

Conflict of Interest: Resolved? What does that mean?

I just posted on the reently published guidelines for antidepressants in the treatment of depression - carefully phrased to indicate the qualifier IF you are gonna use meds, while not bringing too much attention to the IF issue (successful, as headliens seem to indicate).

The article is: Qaseem et al. Using Second Generation Antidepressants ... Annals of Internal Medicine 2008;149:725-733.

Wondering (OK, actually, "suspecting") a conflict of interest, I looked for a COI statement. One conflict is noted: One author, Snow, has received pharmaceutical funding. Fair enough. Just let us know so we can include a grain of salt when flavoring the findings.

But that statement is followed by the curious statement:
"Any conflict of interest of the group members was declared, discussed, and resolved."


There is no such thing.

One of the challenges of science is to minimize any and all sources of bias, so as to test, as purely as possible, the specific hypothesis at hand.

There are many kinds of bias. One area of bias is "experimenter bias." Simply put: as long as an experimenter has some vested interest in an outcome, SCIENCE itself declares that, whether any bias is involved or not, that the result is to be considered suspect to some degree. The theoretical possibility of experimenter bias is enough to taint the study. That is all that is necessary to have a theoretical conflict of interest: to have any degre of care regarding the outcome. Done. Degree of nobility has nothing to do with it.

If any author has in recent history had any financial relation with any of the involved pharma companies, the taint is there.

It cannot be "resolved." It just is.

So: I have no idea what "resolved" means.

Something is fishy.

Being skeptical, as we scientists should be, I start thinking abt who has bankrolled this endeavor: it has apparently been paid for by the "American College of Physicians' operating budget."

which should seem to put the study "above" the spectre of bias or COI.

But let's think abt thia bit more.

Who bankrolls the ACP?

So, I google.
ACP publishes several journals, including Annals of Internal Medicine, that make money by selling ads for: first on their list of acceptable ads: FDA-approved prescription meds.

OK, so the journal itself, exercising its editorial control, has a vested interest in the livelihood of pharmaceutical companies.

In the words of Thurgood Marshall, "Those companies willing to pay for advertising space got it."

Lead author, Qaseem: google: OK he has been funded by Endo, a company testing SSRIs for pain. As luck would have it, this was one of the sub-topics of the article!

So, now lead author Qaseem, and the Annals itself, have COI / have a vested interest in the results but with no disclosure. Snow has "resolved" conflict, whatever that means.

Well, that's enough for me to see the spectre of conflict and bias, "resolved" or not.

"Therapy Wins: Odds Ratio 2.25." Or "Read Between The Lines."

Meds Versus Therapy for depression, and Therapy Wins!!

Well, ya gotta kind of read between the lines to come up with this result.

First: let me put a confidence interval around that odds ratio of 2.25, and explain what that odds ratio is comparing.

The article I am looking at is the newly published clinical guidelines for treating depression, brought to you by the American College of Physicians, as published in the Annals of Internal Medicine. Nov 18, 2008, v. 149, i 10, pages 725-733.

Well, not quite a guide for treating depression. Here is where the 'read between the lines' part comes in.

Without being all that clear about it, the article states, basically: to paraphrase:

"If you are gonna treat depression with pharmaceuticals at all, here is a guide to the ins and outs of which to use."

That is quite different from: "if you have a depressed patient in front of you, here is an evidence-based guideline for what you should do."

Why not the 2nd question? Isn't that what we want to know? How to treat depression?

I assume that most of us would appreciate a timely, well-done review of how to treat depression. But the article clearly states:

"Various treatment approaches can be used to manage depression, such as pharmacotherapy, psychotherapy, and cognitive behavioral therapy. However, the scope of this guideline is limited to pharmacotherapy with second-generation antidepressants (selective serotonin reuptake inhibitors [SSRIs], serotonin norepinephrine reuptake inhibitors [SNRIs], and selective serotonin norepinephrine reuptake inhibitors [SSNRIs])."

And that is the end of the issue of: how to treat depression.

No statement abt meds versus therapy, or whether combo meds-and-therapy is better, the same, or worse.

Why not?

My guess: if and when someone does a similar comprehensive review of evidence for depression treatment, the conclusion will be: if you only can do one modality, do psychotherapy. If you can do anything additionally, do a combo. If meds are the only thing available - say you are on a deserted island with only a pharmacy and no hairdresser, bartender, or minister, then do meds.

This article is startign to get some news coverage.

The headline message, the take-home message, has not been:

"meds might help, if you are the 40% of patients who are lucky to get some relief, and if you don't mind the risk of suicidality (2.25 times the odds compared to placebo, which happens to be moderately less effective than SSRI)."

The headlines have been:

"SSRIs are equally effective for depression. So, just take your pick."

No headline is gonna say: "Why bother with SSRIs?"

Although that is what I wonder, looking at a range of evidence: placebo response in SSRI trials, efficacy of psychotherapy (numbers bounce around but can be 70%, not 40%).

This article side-steps the issue of: how to treat depression. But I am afraid it will be consumed that way.

I haven't been able to read through everything really closely, but since the issue of SSRIs and suicide is incrasingly being documented [hint: google "SSRI" and "suicide"], I thought I would see what this treatment guideline has to say abt suicide.

First: the review concludes that the risk of suicide is not greater versus placebo. Given existing evidence, I am surprised to see this bold statement. I will have ta look into the strategy for making this findign go away.

BUT - the article does acknowledge that suicide atempts are greater for SSRI vs. antidepressant; odds ratio 2.25.

Curiously, though, they have an untenable confidence interval, so I know something is fishy in Denmark. The confidence interval, the "error band", around the 2.25 odds ratio is:
3.3 to 4.6.

This is either: sloppy editing, i.e., a typo, or it indicates that someone does not understand the statistics they are reporting.

If you don't understand a confidence interval, I am gonna explain by analogy:
A hurricane is approaching New Orleans. Forecasters say they project it will hit New Orleans, but they know the hurricane may stray from this path, one way or the other. So, they give the range where it might hit: from Baton Rouge to the wet of New Orleans, to Mobile which is east of New Orleans. That is the "error band" that "surrounds" the "point estimate."

The best-guess, specific point estimate is New Orleans.

Same deal. So, a point estimate of 2.25 with a best-guess range from 3.3 to 4.6 is like saying the hurricane is going to Baton Rouge, but it might hit anywhere from New Orleans to Mobile. That just doesn't make sense.

They may publish a correction fairly soon. Or, we will just have to dig into the original data to find the source of this statistic.

Either way:

Why not skip the risk of a suicidal act, and go do some talk therapy?

Well, there ya have it. Ask a stupid question, and get a stupid answer.

Here was their primary question:
"Key question 1: For adults with MDD or dysthymia, do commonly used medications for depression differ in efficacy or effectiveness in treating depressive symptoms?"

Now ya have your answer. The article is quite skillfully engineered to stick to this question while side-stepping the question of: why bother with meds at all, considering the side effects?

UPDATE: here come the headlines:
from "HealthDay" and "US News" (actually supplied by "scout news service"):
"2nd Generation Antidepressants Prove Effective."