Prozac doesn't work better than placebo

February 25, 2008 4:58 PM

A new peer-reviewed meta-analysis of clinical data demonstrates that four widely-prescribed SSRI anti-depressants, including Prozac and Effexor, are not more effective than placebos. Summary from the Guardian.

The review breaks new ground because Kirsch and his colleagues have obtained for the first time what they believe is a full set of trial data for four antidepressants.

They requested the full data under freedom of information rules from the Food and Drug Administration, which licenses medicines in the US and requires all data when it makes a decision.

The pattern they saw from the trial results of fluoxetine (Prozac), paroxetine (Seroxat), venlafaxine (Effexor) and nefazodone (Serzone) was consistent. "Using complete data sets (including unpublished data) and a substantially larger data set of this type than has been previously reported, we find the overall effect of new-generation antidepressant medication is below recommended criteria for clinical significance," they write.

posted by Rumple (86 comments total)

What about my jimmy legs?
posted by Max Power at 5:11 PM on February 25 [1 favorite]

For a list of effective antidepressants, see Schedule I.
posted by mullingitover at 5:16 PM on February 25 [8 favorites]

So does this make Tom Cruise and the Scientologists right?
Bow before the minions of Xenu
posted by Rubbstone at 5:20 PM on February 25

Careful, I got called into MetaTalk for making a similar statement.

Also, please note, placebos do, in fact, work, so this finding =/= "antidepressants don't work."
posted by OmieWise at 5:36 PM on February 25 [1 favorite]

How would you rate the efficacy of generic Mexican placebos versus namebrand?
posted by The Straightener at 5:38 PM on February 25 [4 favorites]

Conclusions

Drug–placebo differences in antidepressant efficacy increase as a function of baseline severity, but are relatively small even for severely depressed patients.

It took me 5 seconds of reading the link to find a direct contradiction of your text. At the very least, you're being intentionally unclear and weaselly with: "not more effective...."

I don't even TAKE anti-depressants and I am so sick of this shit. Would you tell cancer patients they aren't really sick and don't need medicine?
posted by drjimmy11 at 5:41 PM on February 25 [4 favorites]

Placebo or not, this shit works.

Previous studies suggest that while the first six weeks of happy pill and placebo are comparable, you see a sustained improvement in the people taking a real compound while the effect tapers off in those on the placebo pill.

Too depressed to find reference.
posted by bluenausea at 5:47 PM on February 25 [2 favorites]

reaching conventional criteria for clinical significance only for patients at the upper end of the very severely depressed category.

So your initial comment was, in fact, misspoken. But this post still sucks shit.
posted by docpops at 5:48 PM on February 25 [1 favorite]

bluenausea: epony- ...um?
posted by UbuRoivas at 5:50 PM on February 25

So, since I'm only mildly depressed, anti-depressants can't help me? That's depressing (but not severely depressing, perhaps paradoxically unfortunately).
posted by Mr. President Dr. Steve Elvis America at 6:00 PM on February 25 [3 favorites]

Well, I haven't read the study, but for starters, you can't get a placebo. If you know it's a placebo, you lose the placebo effect. The point is that these drugs are not more effective than nothing, which is important.
posted by Pastabagel at 6:04 PM on February 25

But this post still sucks shit.

And the problem is self-limiting. If the drug isn't having any impact, people just won't bother to take it -- particularly given the side effects. Consequently, the only people who will continue on the drug are likely to be those whose depression is severe enough to have it make a difference.

Also: I note that the authors are all psychologists who tend to have a vested interest in talking cures -- despite their extremely weak efficacy for most psychiatric illnesses. Go figure.
posted by PeterMcDermott at 6:05 PM on February 25

New business idea: Hotline for the mildly depressed. Call before you take your pills, we will make sure you become severely depressed.

Options include: Kick in the nuts, stealing all your money, Dawkins audiobooks for the religious, creationist audiobooks for the scientific, and a MetaTalk only subscription.
posted by Dr. Curare at 6:06 PM on February 25 [1 favorite]

Would you tell cancer patients they aren't really sick and don't need medicine?

Depends are they men with prostate cancer?
posted by Rubbstone at 6:06 PM on February 25

Everyone knows placebos are effective. You don't need to pay Pfizer for them, apparently. Unless parting with money is part of the placebo effect. I did not say antidepressants don't work. I have no comment on cancer and this post is not about cancer.

And this study purports to improve on "previous studies" so you may need to update your mental map, bluenausea.

drjimmy: read the article for more than 5 seconds yet?

How does a link to an open-access peer reviewed medical journal article about the major treatment option for a disease that affects 10s of millions of people, including myself, suck shit?

On preview: CBT has success rates comparable to pharmaceuticals for depression.
posted by Rumple at 6:06 PM on February 25 [2 favorites]

It sucks shit because whatever the intent of the study, tossing a single-link [a summary from the Guardian might as well be like asking the NY Post to explain the physics of rocketry] into a site like this is just inflammatory and ultimately does nothing to improve the debate about depression treatment. So I used the same amount of effort and thought to rebut as you obviously did when making the post.
posted by docpops at 6:17 PM on February 25 [1 favorite]

Placebo or not, this shit works.

[MD, PhD, taker of said placebo talking] I agree with bluenausea's sentiment. If it turns out that the effect of SSRIs is entirely, or largely, a placebo effect, I would be very surprised. For the following reasons. I tried many SSRIs, each of which had a mild effect, until I tried the granddaddy, Fluoxetine (Prozac). I experienced a (relatively) quick (two weeks) elevation of my mood, to the point where I was, in my self-diagnostic opinion, hypomanic, or at least feeling really good, and in fact made some very unwise decisions and, to be honest, engaged in risky and uncharacteristic behaviour. Now, for the past 8 years, I'm levelled out into (thanks God) being a normal human being instead of a depressed fearful slug.

I have no doubt as to the power of the mind over the body, but it would surprise me that my experience, including trying out higher and higher doses until I felt 'normal' and have been on that dose of the drug for years, represented a placebo effect. So I am inclined to doubt the study based on my own experience. But that is, of course, folly, and I will await further studies to confirm or deny the findings of this metaanalysis
posted by Turtles all the way down at 6:18 PM on February 25 [3 favorites]

I don't even TAKE anti-depressants and I am so sick of this shit. Would you tell cancer patients they aren't really sick and don't need medicine?

Do you mean, "If you're not with us, you're against us?" because it's easier to say and write and a more catchy slogan, and it's about as intellectually honest as what you wrote. Or, are you suggesting that if someone is ill it's more important to offer them a medicine, any medicine, whether it works or not, and then to suggest that they have "refractory" depression when the meds don't work?

Also: I note that the authors are all psychologists who tend to have a vested interest in talking cures -- despite their extremely weak efficacy for most psychiatric illnesses. Go figure.

This is disingenuous, and also wrong.

Disingenuous because allegiance to a treatment modality (which these psychologists may have to talk therapy) is not comparable to the "vested interest," in a very economic sense of the term, that psychiatrists and GPs have to psychotropic medications. No one gets rich talking to an individual patient for 50 minutes (even at $100/pop), but it's easier to get wealthy repping for pharmaceutical companies in between 15 minute "med management" meetings with four patients an hour (at $100+ a pop).

It's wrong because talk therapy has an effect size of 0.80 in treating real patients in real clinical encounters, and the type of therapy doesn't matter much. (Read Bruce Wampold and Michael Lambert for the stats.)
posted by OmieWise at 6:21 PM on February 25 [1 favorite]

OmieWise: Just because some people can make more money by prescribing the drug than engaging in the talking cure is not, logically, proof that the drug does not work.
posted by Turtles all the way down at 6:24 PM on February 25

Adding something about CBT to the main post might have avoided some of the reactions, then.

Additionally, won't this news make the drugs even less effective, once the mainstream media gets a hold of it? Known-placebo effect (is that a term?) or something.
posted by cobaltnine at 6:25 PM on February 25

The only exception is in the most severely depressed patients, according to the authors - Prof Irving Kirsch from the department of psychology at Hull University and colleagues in the US and Canada. But that is probably because the placebo stopped working so well, they say, rather than the drugs having worked better.

"Given these results, there seems little reason to prescribe antidepressant medication to any but the most severely depressed patients, unless alternative treatments have failed," says Kirsch. "This study raises serious issues that need to be addressed surrounding drug licensing and how drug trial data is reported."

Great point, Dr. Kirsch. Could you post your home number for my patients so they can dial your home collect for counseling that their insurance won't cover? What are these other treatments? A new minimum-wage job? A new childhood? Dental insurance? Light therapy? Vitamins? I'd love to see another study using pain medication instead of antidepressants. Yes, patients all say they work, but since the placebo worked as well, right up to when it stopped working (wtf?), we need to quit our sissified approach to pain and tell people to suck it up and walk it off.

All this mental masturbation, I hope they have a lot of dishtowels around.
posted by docpops at 6:25 PM on February 25 [5 favorites]

Known-placebo effect (is that a term?) or something

I can already feel my dosage becoming less effective.
posted by Turtles all the way down at 6:26 PM on February 25

Either these drugs work, or they don't. This study, which purports to be state of the art, finds that they do not, except in extreme conditions, work better than a placebo. While this does not mean that they don't work at all since placebos work, the standard is, one would think, higher than a sugar pill.

Turtles: how does one confirm or deny the results of a meta-analysis? Re-run the data? Wait until more clinical studies are done? Surely one more clinical study on its own cannot confirm or deny a meta-analysis? I am curious.

On preview: if the reason for continuing with these drugs is so the insurance companies can keep paying the pharmaceutical companies then I suggest you are shooting the messenger.
posted by Rumple at 6:30 PM on February 25 [2 favorites]

based on personal experience, the drugs really, really work. at least for me.
posted by brandz at 6:35 PM on February 25

I could be wrong, but I've perused this thing from top to bottom, and I come away with little more than head scratching -- and part of my job depends on critical analysis of the medical literature. This study, with its data pooling and it's dubious use of standardized mean differences seems a fine example of how over meta-analysis can washout real, statistically significant results from placebo-controlled, prospective studies. My gut tells me there's a fair bit of statistical slight of hand going on leaving me with only the observation that placebo effects in depression are quite strong -- which I already knew. If you look at what I think is the money figure here, even though the concept of scale is totally thrown off with standardized mean differences as the axis choice, it still looks like the SSRIs are having an effect beyond placebo.

Sorry but your or anyone else's editorial suggesting that this paper proves SSRIs are not more effective than placebos just seems off the mark.
posted by drpynchon at 6:37 PM on February 25 [1 favorite]

Rumple, every treatment carried out for non-life-threatening illness in this country to the tune of billions of dollars is, for the most part, pointless and untested. For example, nearly every visit for back pain, nearly every respiratory illness, most every complaint of nonspecific pain we see in primary care, of fatigue, of headache, of existential dread and hemorrhoids. We chew up billions of dollars in time and medicine because the cost of not treating the human condition would, in all likelihood, be far worse in terms of dollars and psychic morbidity - in essence, telling people, go away, you don't need any of this. Frankly, as a society we could really use a dose of 'suck-it-up', but to isolate depression for our microscope is such a cruel and pointless exercise.
posted by docpops at 6:38 PM on February 25 [1 favorite]

What I'm astonished by and nobody seems to be talking about is that they had to FOIA sue the FDA for the data supporting drug approval. That's criminal.
posted by a robot made out of meat at 6:41 PM on February 25 [6 favorites]

Turtles all the way down: this is actually a pretty robust finding, replicated many times in many different ways. Here's a link to a NEJM paper that, while different, had congruent findings.

There are plenty of other instances of this finding being replicated, although I don't have all my citations here. But, that does not mean that there are no effects for any patients. Certainly psychotropic medications have effects.

The big question is when proponents of a medical model/chemical imbalance hypothesis for depression, who, by their very stance, suggest that depression is an illness that should be approached scientifically, will pay attention to the science on this issue? You'll not catch me minimizing depression, or suggesting that it does not need or respond to treatment, but I'm also not going to ignore the actual science, which not only suggests that antidepressants aren't all that good at treating depression, but that talk therapies have an effect size of 0.80 for depressed patients. That very robust finding has been repeated in at least five meta-analyses comprising at least three decades of research.

On preview:
OmieWise: Just because some people can make more money by prescribing the drug than engaging in the talking cure is not, logically, proof that the drug does not work.

No, of course not, and I didn't mean to suggest that it did. I do, however, take exception to the suggestion that studies questioning the efficacy of medications from psychologists should be treated warily because of the "vested interests" of said psychologists, while there are very real vested interests among psychiatrists and GP who prescribe meds which far outweigh any profit to be had from talk therapy.

Great point, Dr. Kirsch. Could you post your home number for my patients so they can dial your home collect for counseling that their insurance won't cover? What are these other treatments? A new minimum-wage job? A new childhood? Dental insurance? Light therapy? Vitamins? I'd love to see another study using pain medication instead of antidepressants. Yes, patients all say they work, but since the placebo worked as well, right up to when it stopped working (wtf?), we need to quit our sissified approach to pain and tell people to suck it up and walk it off.

docpops, you've made some form of this argument in several threads on related topics, and I really sympathize with your obvious frustration at the lack of options for mental health treatment. Seriously, I think it's horrible and I understand why medications present a very real only chance for many folks suffering from depression, etc. I don't, however, think that that lack of treatment options invalidates studies like this, nor do I think researchers should exercise prior restraint because of it. I work in the mental health field, and I can say, without a doubt, that the limitations in treatment options, the way in which we view mental illness, and the ways in which we provide access to social services are all affected by an overly simplistic model of mental illness promulgated by precisely the parties that stand to profit most from that curtailed approach. While I understand that you're frustrated, I don't think that it's psychologists who are responsible for the mental health care landscape that you see before you.
posted by OmieWise at 6:43 PM on February 25 [3 favorites]

Well, I can tell you that at least Effexor is doing *something*, because when you try to come off the crap or miss a dose, you get a weird dizzy feeling some people call brain freeze or brain shivers, which is highly unpleasant and makes you feel like your blood pressure has gone haywire and your head is going to do that thing that Arnold Schwarzenegger's did in Total Recall when he fell out of the Mars base window.

My experience tells me that it does work, at least to some degree, to combat depression, although it don't work to cure serious depressive illness (because nothing really does). You have to make changes in your life yourself and become willing to stop being the way you are before that happens.

I actually believe, for most people without a history of mental illness, smoking pot is a lot better for you than any of these pharmaceutical magic pills. I certainly am very dubious about prozac, but I know someone with serious depressive illness who assures me that without it, she's just lying in bed miserable all the time.

So, ultimately, YMMV.
posted by Henry C. Mabuse at 6:46 PM on February 25

Omiewise - couldn't agree more. Thanks for your comment(s).
posted by docpops at 6:46 PM on February 25 [2 favorites]

Rumple: A study like this is more difficult, in my opinion, to judge because it relies not only on the experimental design and statistical results of the studies it has chosen to focus upon but additionally applies its own design methodology and statistical analysis to these results. A meta-analysis of clinical data is, in other words, one more step removed from what's acutally going on in the brain of the individual patient, were you able to measure it accurately. So, compared to my own (possibly placebo-biased) experience of my own condition, this kind of study is more rigorously controlled. But it includes all kinds of other errors that magnify those same errors that may have been introduced in each of the studies under study. Were the subject populations assigned randomly? Were they evaluated blindly? Were placebo effects unintentionally introduced into each of those study populations? Was everything perfect in each of the component studies, and the meta-analysis? I've spent a lot of time analyzing my own condition and the effects of various drugs on it, so am reluctant to relinquish my conclusions based on a study that may have introduced several more variables.

My comment, if you read it again, said that this study was interesting, but as kind of a scientist myself as regards my own treatment, I would be very surprised if, in the eyes of God and Mother Nature, Prozac has no objective effect on mood.
posted by Turtles all the way down at 6:49 PM on February 25

I'm more curious about the results of prozac vs prozac+therapy vs therapy vs placebo vs placebo + therapy.
posted by empath at 6:49 PM on February 25

Turtles: I don't see how it could be possible that prozac doesn't affect mood. It works on the same chemistry that ecstasy does and no one can doubt that THAT has an effect on mood.
posted by empath at 6:50 PM on February 25

Omiewise - couldn't agree more. Thanks for your comment(s). (docpops)

Seconded
posted by Turtles all the way down at 6:51 PM on February 25

Note: Kid Charlemagne makes drugs for a living - haven't you paid attention to the song!?!?! So clearly I'm a shill for the man. This disclaimer in place:

Reading the article carefully, the thesis seems to be that the placebo effect is extremely powerful in the case of depression but that it falls off over time, not that the drugs have no effect.

Also, meta-analyses is a red flag in my mind. Consider: "Inclusion of a drug type for which unsuccessful trials were excluded biases overall results in favor of that drug type, in a way that is akin to publication bias."

Imagine I'm a drug company doing work with an anti-cancer drug. We do three studies on small cell lung cancer, pancreatic cancer and melanoma and find that while our drug is great for pancreatic cancer and melanoma, it doesn't do diddly for lung cancer, so we go to the FDA with those two indications. Am I to understand from this that he's including the data for the indication or population segments and/or indications that did not get label inclusion? Does anyone see anything that would confirm or deny this suspicion?

Also, I'm not familiar with the Public Library of Science so I poked around in there. Check out this gem: "The study found that although annual health-care costs are highest for obese people earlier in life (until age 56 years), and are highest for smokers at older ages, the ultimate lifetime costs are highest for the healthy (nonsmoking, nonobese) people. Hence the authors argue that medical costs will not be saved by preventing obesity." Get that - lifetime cost. Not cost per year or anything like that. I'm sure you could lower lifetime costs even more by turning rabid dingos loose in the hospital nursery. Who edits this thing - the guys from Barr Pharmaceuticals?
posted by Kid Charlemagne at 7:01 PM on February 25 [1 favorite]

Ok, reading the whole paper, it appears that the FDA complied willingly with the FOIA request.

I wonder if standardized mean difference isn't an underestimate since one might expect a bimodal distribution with responders separated from non-responders rather than a pure continuum; I'm reading some of the papers to see if that's true. There are some other points to consider also. About 40% of people in each wing of the studies dropped out. For placebo, that's probably people with worsening depression. For drug it could also include people who need their dosage adjusted and those who suffered side effects. In any event, it's a pretty atrocious number.
posted by a robot made out of meat at 7:06 PM on February 25

Rumple, every treatment carried out for non-life-threatening illness in this country to the tune of billions of dollars is, for the most part, pointless and untested.

If I ever break my arm I think I'm going to look for a second opinion.
posted by Kid Charlemagne at 7:09 PM on February 25 [1 favorite]

I have to pitch in my anecdotal evidence too. During my fourth depression episode (and the worst one yet), I went through must have been four or five anti-depressants until I found one that worked. Effexor XR started working in about a week... and this is against a Nietzsche-level, staring into the abyss, six-month long depression. It was hugely effective for me, and I doubt it was a placebo effect, because I had no higher hopes of that drug working than all the previous ones that failed.
posted by wastelands at 7:25 PM on February 25 [2 favorites]

I had made a post on this topic in the past. The interesting thing to me was the bias in the results of double-blind studies on the efficacy of SSRI's based on their respective sources of funding. I believe i also got yelled at..
posted by dminor at 7:42 PM on February 25

Actually, placebo should be made an explicit and sustained part of every treatment.

If everytime you went for a prescription, you were forcefully told to *be absolutely sure* to slightly bend your knees, sit up ramrod straight, take three deep breaths, and passionately recite a mantra like "Pfizer and I are One" just before you took a pill-- and to only do the ritual in that *exact* order-- you'd probably get better results with that pill than if you were told, "Hey, try this, and let's see what happens."

When it comes to change and healing, belief-- no matter how arbitrarily generated-- is your friend.
posted by darth_tedious at 7:46 PM on February 25

and this is against a Nietzsche-level, staring into the abyss, six-month long depression

To those who have not have depression sufficiently severe or long in course that you felt you needed to seek treatment: Imagine you feel sad and tired, for no apparent reason. Then you feel more sad, and tired, and maybe panicked because you need to get on with things but you just don't feel up to it. And you, being not a complainer, tell yourself, "shake it off, boy! Get out for a walk in the fresh air--you'll feel better, your just being lazy!" Then you go to your wife's work function and try to maintain a normal facade but in ten minutes you need to run out of the room to the stairwell because you've burst into tears (and you're a 250 lb. man so this is really bad). And you walk around the city for a couple of hours trying to talk sense into yourself. But you're hopeless, and sad, and even worse, self-critical. And nothing gives you pleasure any more except sleeping. And drinking, which seems to work miracles. And this goes on for a couple of months and you've pretty much given up hope of being normal and doing all those things you intended to do. And your family and friends has no idea why you're being so lame and can't go to a family function without crying.

Now imagine you get a prescription for a drug. And over the next several weeks things get slowly better. To the point where without you noticing it, you're back to normal. You feel a certain allegiance to the pharmaceutical.
posted by Turtles all the way down at 7:46 PM on February 25 [7 favorites]

If you know it's a placebo, you lose the placebo effect.

Has anyone tested this? Could the placebo effect recurse? That is, can people knowingly take a placebo hoping that the placebo effect will help them and have that twice-removed hope help them?
posted by TheOnlyCoolTim at 7:48 PM on February 25

Why is this even an issue? People put things into their bodies to make themselves feel better. If it works for them, great. If it doesn't, try something else. Why all the gnashing of teeth over brain pills?

For the record: Zoloft works for me. I was originally prescribed it for anxiety and severe hypochondria ("Doctor, I think I have stomach cancer!!!11!!!11!" "Uh, no, Mr. Avenger, you have gas.") to the point where I literally couldn't leave the house some days. I have to say that in the few times when I forgot to take it for a week or two, my weird thinking comes back -- full force. When I'm on them, like today, I can think like a normal human being and not be paralyzed with Dread of Impending Doom.

Something in these pills works. I'd be interested to know as a scientific curiosity -- but even if we never discover the mechanism of action (or even if we discover that Zoloft is nothing but a $3 sugar pill) then I'll still be just fine with that. Make me feel better, doc, I don't care how.
posted by Avenger at 7:50 PM on February 25

The gnashing of teeth, I believe, is mainly over marketing.
posted by dminor at 7:54 PM on February 25

That is, can people knowingly take a placebo hoping that the placebo effect will help them and have that twice-removed hope help them?

Probably not. In standard clinical trials the identity of the placebos is strictly controlled.
The subjects, the clinic, the sponsor and the the formulators do not know which dosages are active and which are placebos - nor does the packaging provide any hints.
Any information on which dosages are which, even if the information is incorrect, can screw up the study.
The human mind appears to be pretty finicky in this regard.
posted by speug at 9:07 PM on February 25

Coming off Effexor XR makes Heroin withdraw easy in comparison. Oh yea and the weight gain on the Effexor is stupendous, which portends diabetes, and now you look like a big pile of shit. So Effexor is addictive physically damaging, and minimally effective at best; but is legal to prescribe. What a wonder drug for the drug companies. At least opiates work right away and aside from the tolerance (fixed with DMX or with naltrexone) and legal issues; is a lot cheaper. Oh and the SSRI's; just try and have a climax during sex, you get to the edge but cannot go over. Contrary to myth women (or men for that matter) do not care for a fuckathons. It has been suggested that Ketamine might be in the treatment of depression. Like Ketamine, DMX is a dissocative anesthetic.
posted by Rancid Badger at 10:11 PM on February 25 [2 favorites]

Please correct me if I am wrong, but using a handful of drugs to treat depression seems like using a mallet to fix a watch that runs slow. I don't doubt that depression is a real medical condition, and perhaps throwing a pill at the problem can help some people, but one pill to treat 40 million people seems to lack the nuance and insight that is really needed. The human brain is a vastly complex organ and I don't think we have a good enough grasp on the problem to create a real solution. We don't have a quantitative model of the biochemistry associated with depression to define specifically what is or is not effective. SSRIs may work for some people with a specific flavor of depression, and may offer effectiveness that is on par with placebo for others, but they do have side effects and may not work for many people. These drugs are created by groups of people who may arrogantly assume that they understand depression and can fix it, and then marketed by people who have an economic incentive to see the drugs brought to market. I really feel that the combination of side effects from these drugs, and a willingness to treat the symptoms with a pill rather than really engage the problem are leaving many without proper treatment. Sixty years ago it was acceptable to shove ice picks in people's eye sockets to treat a whole host of conditions. We have learned a great deal about the brain since then and we have much more to learn still. I would be wary of any invasive technique for treating psychological conditions until we have a better understanding of what we are doing.
posted by kscottz at 10:26 PM on February 25

Yes the Black Dog ( Churchill's expression) sleep's at the foot of my bed and at times I would have had to gotten better just to have been able to kill myself. Exercise is amazing in it's restorive powers but sometimes you need a push from a friend or spouse. Having friends who have gone through it and are willing to let you bitch and moan are a true God send. Pets also work well (except for that GD black dog). Meditation and cuddling also help. Last resort is ECT and though it get bad press it really can work.
posted by Rancid Badger at 10:28 PM on February 25

Speaking of invasive surgery for depression....
posted by Rumple at 10:33 PM on February 25

Rumple - (this is a tad off topic) I think deep brain stimulation may have some practical applications for treating Parkinsons disease as we have a decent understanding of what the disease is, how it works, and potentially how to reduce or eliminate the problems that cause the symptoms. That said, I still think DBS is fairly primitive and questionably applicable to treating depression or other psychological conditions. Depression is a real condition but the causes are much more subtle and varied and we don't have a mechanism or a set of mechanisms that describe it. Since we can't precisely define the condition (chemically) we really can't propose solutions or measure their effect in a quantitative manner.
posted by kscottz at 10:58 PM on February 25

kscottz --

Very wise, except that most people who commit suicide are actually doing it to make the pain of depression go away: anything that is so torturous that death is preferable is worth taking a risk or two, if it might actually help.
posted by jrochest at 11:06 PM on February 25

kscottz -- I know it works for some conditions, Parkinsons, some epilepsies I think. But like you imply, I don't think there is a 1cc part of a brain that if you fry it with and electrode it makes you happy. It'd be awesome if there was, though
posted by Rumple at 11:24 PM on February 25

There was an awesome Radiolab episode on the power of placebo.
posted by kaibutsu at 11:28 PM on February 25

but one pill to treat 40 million people seems to lack the nuance and insight that is really needed.

Yeah, screw those aspirin manufacturers!

(What? Biochemistry is 99% the same for all humans? You don't say).
posted by tkolar at 11:43 PM on February 25

jrochest - Suicide is horrible ( I have seen a few friends and family try and some succeed). I often wonder though if the SSRIs really do anything to prevent it. I know in the case of adolescents it can actually make things worse. I tend to think that SSRIs probably don't do much but allow the depression suffers and those around them sleep easier knowing they are being "proactive" in the problem. If the SSRIs worked maybe the side effects would be worthwhile, but if we are just deluding ourselves into thinking we are fixing the problem what is the point? Perhaps instead of having false hope we should pursue methods that are actually effective.

My opinion is that the placebo effect of SSRI is less useful than the psychology of a doctor having empathy for the sufferer's depression, and having the drugs as a "tool" to ask for the assistance, emotional sensitivity, and extra space/time needed by depressed people. We as a society tend to ignore people's emotional needs, while giving those with medical conditions the extra consideration they deserve. It is a shame we need to apply a clinical name to what is really a combination of biology and stressful social norms to get people the help they need.
posted by kscottz at 11:45 PM on February 25

tkolar - The brain is way more complex than simple muscular pain. Just look at how aspirin works. Pretty complex eh? The brain has a bunch of different organs, half a dozen neurotransmitters that each have a different function in each organ (and different receptors), and a bunch of different kinds of electrical signals. Throw in genetics and you are looking at combinatoric nightmare (or a beautiful mystery). Some simple problems in the body can be fixed by simple solutions, but I doubt that depression is really going to be fixed by a one size fits all solution. Precisely diagnosing what is causing a person's depression would be a problem in itself. Biochemistry is uniform across humans but the devil is in the details.
posted by kscottz at 12:00 AM on February 26

TheOnlyCoolTim: Has anyone tested this? Could the placebo effect recurse? That is, can people knowingly take a placebo hoping that the placebo effect will help them and have that twice-removed hope help them?

Not really.
posted by Gyan at 12:58 AM on February 26

I do think it's interesting how the following statements have become confused by some people:
1) According to (a subset of) the studies originally filed with FDA, the SSRIs were (statistically) significantly more effective than placebo overall -- though if you use Europe's NICE criteria, the effect isn't 'clinically significant' -- except for the treatment of like severely depressed people, in which case, uh... it is.
2) According to all the data out there, SSRIs are no better than placebo, man! 850
3) SSRIs "do not work." 900
4) SSRIs don't really do anything, they are placebos. 1300

I've appended a number of points after some of the answers. Please add them to your personal dumbass score. Note that if you believe 4 you probably also believe 2 and maybe 3, go ahead and score all points as appropriate.
posted by fleacircus at 1:11 AM on February 26

Meh. I'm pretty skeptical of meta-studies in general. By choosing which studies to include, which to exclude and how to weight them, you can pretty much make them give any conclusion you want.

Given that the studies that were published did give significant results, what was wrong with their implementation or methodology? Did they all use flawed statistics, or sample sizes that were too small?

I wouldn't start binning the pills on the basis of a meta-study alone.
posted by TheophileEscargot at 2:21 AM on February 26

I tend to think that SSRIs probably don't do much but allow the depression suffers and those around them sleep easier knowing they are being "proactive" in the problem.

Bullshit. SSRIs are not the answer for all people or all cases of depression. Having done a fair amount of research into the subject, both out of curiousity and out of an attempt to save my own skin, it's pretty damn clear that depression's a fun sort of multifactorial disorder. In many cases, you've got a mix of outside problems (situational depression) and chronic depression, and I wouldn't be at all surprised to find out that altering any number of neurological pathways might lead to depressive symptoms. I personally suspect that this might partially account for the fact that some people have very strong and positive reactions to SSRIs, while others see no effects at all.

Certainly [anacdata warning], there are a great many people suffering from chronic depression who have tried to handle it with other meds or with therapy alone, without success - despite the fact that they know they're "being 'proactive' in the problem." Yet some of these types do find the help they were looking for when supplementing therapy with SSRIs. I should know: I'm one of them. Tried for years to get by with therapy and determination, and while that helped me devise coping strategies and support networks, it couldn't stop depression from periodically swamping me. SSRIs haven't been a magical cure, either, but they've lessened the incidence and severity of those depressive sloughs. Furthermore, "not much better than a placebo" isn't the same as "not much better than nothing" - and hey, if I'm paying for a placebo that's working, it's still better than my status quo.

Beyond personal experience, though, it's certainly not the case that:
These drugs are created by groups of people who may arrogantly assume that they understand depression and can fix it... Quite the opposite. We think seratonin is probably involved in depression, or at least some widespread types of depression, but it's pretty obvious to people doing the research that it's still something of a shot in the dark, and really, everyone concerned wishes that we understood much more about depression so that we could target it more accurately. There are other treatments that work without us really understanding why: electroshock therapy and the nascent technology of deep brain stimulation. But it's not quite as popular to mock those therapies currently. They're all blunt weapons, but right now, they're all we've got, and the lack of anything better has nothing to do with scientific arrogance or drug company greed.

Furthermore, I'm sure there are doctors and psychiatrists who are just pill-pushers. But I've seen several studies suggesting that a combination of meds and therapy tends to be the best treatment of all, and indeed, every psychiatrist I've gone to has gone to great lengths to make sure I was combining both. They've also all been pretty clear about potential med side effects, and the possibility that one might need to try a few types of drugs before giving up on medication, since some medications seem to work especially well on some people (and, again, we don't know why).

So hey: many of those of us on meds are quite aware of their failures, and would be happy to have more and better options. But for now, this is what we've got. Stories like this attest to the inadequacy of current treatment options, but don't throw the baby out of the bathwater with blanket statements about how SSRIs don't work and are a waste of our money and time.
posted by ubersturm at 2:22 AM on February 26 [3 favorites]

"Coming off Effexor XR makes Heroin withdraw easy in comparison. Oh yea and the weight gain on the Effexor is stupendous,... Oh and the SSRI's; just try and have a climax during sex, you get to the edge but cannot go over."

Hmmm, you have just described pretty much my entire side effect history with Effexor, including the infinitely prolonged orgasm (although this last seemed to not happen after a few months). I would still recommend it above SSRIs to anyone with panic attacks or mild to fairly serious depression.
posted by Henry C. Mabuse at 3:34 AM on February 26

I know lots of people who use straight-up placebos: echinacea, ginseng, super-vitamin-C pills, etc.
posted by a robot made out of meat at 4:32 AM on February 26

Options include: Kick in the nuts

I believe the technical term is CBT.
posted by hoverboards don't work on water at 5:06 AM on February 26 [1 favorite]

Don't do drugs. I've said it before. Not only do they not do what you want them to do (fix your brain) but they can seriously screw you up.

Psychiatry: Still shafting us with bad science after all these years.
posted by ewkpates at 5:32 AM on February 26

Oh shush with your scientology claptrap.

I'm not exactly a fan of SSRI's, but there's a difference between an industry out of balance and some Markov alien mind control conspiracy.
posted by cytherea at 8:05 AM on February 26

Meta-analyses should never be performed. Data obtained under different conditions are analyzed, using statistical methods which require as a prerequisite for validity that the data under analysis was all have been obtained under the same conditions. That's a meta-analysis. Meta-analysis is shorthand for "invalid methods of analysis;" or in layman's terms, "comparing apples to oranges."

Particularly specious is the fact that this data ("all" clinical trials) includes phase I and phase II data. These trials are conducted to determine safety (Phase I) and to determine effective dose (phase II), not to determine medicaton efficacy (Phase III). If you use safety and effective-dose-determination trial data to determine efficacy, you are going to include a lot of people who got the meds at doses less than the doses eventually determined to be effective. Throw them into your efficacy analysis as "treated" and, yes, you are going to show that the medicine is inefficacious.

Damn lies and statistics, people. I like to see truthiness imprisoned in Guantanamo and subjected to cruel and unusual punishment as much as the next neocon, but this is really egregious.
posted by ikkyu2 at 8:13 AM on February 26 [6 favorites]

some Markov alien mind control conspiracy.

Does Markov mean something other than I think it means?
posted by TheOnlyCoolTim at 8:27 AM on February 26

Sorry, Marcab. Though markov chain might be a more apt description of hubbard's writing.
posted by cytherea at 9:16 AM on February 26

kscottz, you seem to be suggesting that until we understand everything about a part of the body we should not attempt to treat it. But consider how medicine began: trial and error over centuries. At best you can say that psych treatment today is in its infancy and primitive.
posted by phliar at 9:51 AM on February 26

An SSRI saved my life. Don't care how. Just glad it did.
posted by kebnabi at 10:39 AM on February 26 [1 favorite]

I think that a lot of people here are misunderstanding this study.

(1) It is true that meta-analyses are purely statistical exercises, one more step removed from actual patient experience than clinical trials, and thus should be taken with a grain of salt. However, this does not render them useless - any more than the fact that most clinical trials use complicated statistical methods does not render their data useless. A well-conducted meta-analysis can provide useful information, and be as statistically valid and reliable as a clinical trial or other methodology.

(2) A second point of doing the meta-analysis here is that it used the data provided by the drug companies in trying to get the drugs approved. This is not a bunch of psychologists conducting a random study to critique pharmaceuticals, it is a reanalysis of precisely the data that led to the approval of these drugs, which we would expect would show the greatest effects, since drug companies cherry-pick their best results for publication and for submission to the FDA.

(3) This study does not prove or even suggest that the drugs "don't work." It argues that, on average, several classes of SSRI do not provide a significantly better benefit than placebos do. Like it or not, this is the gold standard by which drugs are measured: how much better do they do than the placebo effect?

(4) Anecdotal evidence in this thread is irrelevant to the findings of this study. the authors would readily admit that plenty of individuals in this study benefited from SSRIs. The point of the study is not that SSRIs won't work for you (particularly if you are "severely" depressed), but rather that on average they seem to be only marginally better than placebos. Saying that Prozac works for you does not disprove this study, any more than saying that Ambien didn't make you sleepy disproves that Ambien "works" on average. In just about any clinical trial, some patients benefit from the drug (according to the trials' outcomes) and some don't; but whether or not it is approved for use depends upon how well the drug performs in the population, not any given individual.

(5) PLoS Medicine is a new but fairly well-respected publication. Its no JAMA, but its hardly a fringe broadsheet.

My sense on reading the study is that the authors would not suggest that we should get rid of SSRIs, nor that they don't "work." Rather, I think it suggests that they have been approved for a much larger patient population than may be warranted, and that we might consider reducing the amount of people on SSRIs, because many may be taking them without deriving any real benefit.
posted by googly at 10:43 AM on February 26

I think the real story here is the difference in the data presented for approval by the drug companies, and the additional unpublished data included in this analysis.

The difference being that the favorable stuff was published and the nonfavorable was not published. Of course, everyone suspected this, but now we see it laid out in gory detail.
posted by Mr. Gunn at 11:13 AM on February 26

I've used Paxil and now Effexor. Paxil had some side-effects I didn't like. But placebo-effect or not, they both have made a pronounced, positive change in my mood. When I don't take them, everyone can tell, sometimes before I can.

I would bet that a large number of people prescribed antidepressants have short-term depression (or anger or anxiety). These people don't need to be on antidepressants. If any drug, maybe Xanax or maybe a glass of wine. Most people are probably better helped by doing something positive for themselves: like exercise or art or charity.

I love my Effexor. But it's not for the occassional down period. It's for a chronic like me. :) I'm not surprised that a placebo is as effective for a lot of people.
posted by muzzlecough at 12:25 PM on February 26

For most people, depression goes away on it's own: the situation improves or you learn to live with the situation. If the problem isn't permanently screwed up brain chemistry, focusing on the brain chemistry might be counterproductive for these people. I'm glad this is being made public.

I'm probably belaboring this, so I'll shut up now.
posted by muzzlecough at 12:34 PM on February 26

It argues that, on average, several classes of SSRI do not provide a significantly better benefit than placebos do.

Actually, while some of the editorializing as well as the primary author's writing (who apologies for a bit of ad hominem, seems to be making a career off convincing people SSRIs don't work) would seem to have you believe this, that is not a fair interpretation of what the study shows. The study uses what it feels is the NICE definition of improvement on the HRSD (Hamilton Depression scale) to define a "clinically significant" improvement in depression. On this scale they feel that a 3 point improvement (based on who knows what) is significant whereas the pooled data from multiple studies with different time courses, different drugs, varying baseline severity, and varying dosages (did it really including Phase I data? If so, that's patently ridiculous), only showed a 1.8 point average increase compared to placebo on the HRSD.

The first issue here is the difference between a somewhat arbitrarily defined consensus notion of "clinical significance" versus the more concrete notion of statistical significance. The drugs provide a statistically significant improvement in the bulk of studies included. Period. Now if in pooling the data, you feel that the magnitude in a difference which is statistically real is somehow inadequate, that's fine. But that's a terrible conclusion to jump to via meta-analysis where magnitudes of effects are so washed out by the data pooling. As it is, the magnitude of benefit in most individual SSRI studies which are single-step is muted by a fair fraction of total non-responders to a drug. Anybody that has experience with SSRIs knows that at least anecdotally (and I think also there is fair research to back up) the fact that some patient's respond to trials of alternative agents when a first line drug doesn't yield a response. In fact some of the better designed studies on these drugs are conducted outside of pharmaceutical support and use a tiered approach that takes the cohort of patient's deemed non-responders after a period of time and places them on a new arm using an alternative drug. There's no way to take the results from that kind of nuanced prospective research design and pool it with other data in an honest way.

It's utterly disingenuous. Extroplating from this that SSRI drugs don't provide a significantly better benefit is just not a fair conclusion by the vary nature of the meta-analysis they performed. Even if you choose to use the NICE standard of clinical significance, you have to accept that that's a consensus standard designed for application to primary data from single clinical trials. That standard as far as I can tell IS NOT VALIDATED STATISTICALLY/SCIENTIFICALLY for use in secondary meta-analysis precisely for the reasons I am alluding to.

Now, there are other perfectly good arguments to make regarding the weaknesses of the data out there on SSRIs, including the publication biases OmieWise refers to. But this is a stellar case like I said before of how some (not all) meta-analysis can twist the primary data out there. Moreover, it again reflects the dangers of publishing mainstream fourth hand pop-sci media interpretations of the discussion section from papers written by authors who may or may not have there own aggendas, without having the experience or training to evaluate the data and study design at face value. Go read what the monkeys over at Digg are commenting about this to get an idea of what I mean.
posted by drpynchon at 12:39 PM on February 26 [2 favorites]

I think this discussion is kind of missing the point. It's not necessarily about the effectiveness or lack thereof of SSRI-drugs, it's about the abuse of the public trust the FDA and the drug companies are engaging in. Defending the idea that "the drugs work" based on personal experience may be gratifying for you or your patients, but it does not really figure into the bigger issue.

The bigger issue is that people are (potentially) being taken advantage of by both big Pharma and the government organization that is designed to protect the weakest in society from practices which do not benefit the public or (potentially) the individual.
posted by cell divide at 12:47 PM on February 26

Mind Hacks has a nice summary of this topic.

BTW, the Mind Hacks book is also great, as is the site.
posted by muzzlecough at 1:49 PM on February 26

Good points all, drpynchon. I should have clarified that my post was aimed primarily at commenters who were citing anecdotal evidence against this study, or interpreting it to say that their particular SSRI of choice didn't work, not trying to defend the merits of this particular study.
posted by googly at 3:23 PM on February 26

And this goes on for a couple of months and you've pretty much given up hope of being normal and doing all those things you intended to do....Now imagine you get a prescription for a drug. And over the next several weeks things get slowly better.

Two potential alternative explanations for your subjective experience. The first, is that whatever actions you took to get a drug required you to take initiative, something that becomes extremely difficult in a severely depressed person. So your motivation to seek therapy may have been evidence of the beginning of a natural improvement in your condition. Can't be sure, obviously, but it competes with the hypothesis that the drug itself did something.

A second, purely statistical explanation that is sort of like the first but differs in that it doesn't require that seeking therapy be the leading edge of an improvement in the depression: regression to the mean inevitably occurs if you select a fluctuating measurement when it is at its extreme and then measure it again later. The second measurement will probably be closer to the mean than the first. If I look up the hottest days of the year, and then look up the temperature of the next day, it will be cooler on average. And vice versa for the coolest day. Your state of depression slowly fluctuates over time, and if you start trying therapies when it gets really bad then there is a good chance that later you will feel better regardless of which drug you're taking at the time.

Randomized trials are for understanding whether drugs work on average. Even if they, on average, don't have a beneficial effect, it may be that they work in some people and do harm in others. So your individual experience may be truly the effect of your medication, but if one can't determine in advance who will benefit and who will be harmed, then the physician is rolling the dice every time she prescribes the drug.

What really blows me away is that drug companies are not required to actually do large enough trials to establish efficacy unequivocally before they are allowed to make billions off of them. Or at least be required to do a large enough trial as they make their billions.

Imagine I'm a drug company doing work with an anti-cancer drug. We do three studies on small cell lung cancer, pancreatic cancer and melanoma and find that while our drug is great for pancreatic cancer and melanoma, it doesn't do diddly for lung cancer, so we go to the FDA with those two indications. Am I to understand from this that he's including the data for the indication or population segments and/or indications that did not get label inclusion? Does anyone see anything that would confirm or deny this suspicion?

The only indication for which the meta-analysis was done is depression.

Particularly specious is the fact that this data ("all" clinical trials) includes phase I and phase II data. These trials are conducted to determine safety (Phase I) and to determine effective dose (phase II), not to determine medicaton efficacy (Phase III).

I think you may have misunderstood the methods. The meta-analysis only used randomized placebo-controlled trials, so they were all phase II or phase III with efficacy measurement. Also, note that they were only able to get data from alternative sources from 4 of the 9 non-significant trials that the FDA did not give them. This means that there were 5 negative trials that were not included, so this analysis is optimistic.

All-in-all, I'd say that if SSRIs were the godsend that the drug industry and the enthusiastic users say they are, this meta-analysis should have produced an astoundingly large effect with very small confidence limits. It did not. The study was pretty well done and only included randomized trials with placebo controls, so the result cannot be explained away as bias or selection effects. This is pretty damaging for the hypothesis that the average effect of SSRIs is positive. Of course, as with any psychoactive drug, YMMV.
posted by Mental Wimp at 5:36 PM on February 26 [2 favorites]

What drpynchon said really hit the nails on the head wrt to the issues with this study. The summaries of this study in the news media, reporting the authors' statements as "fact," are just sad.

Meta-analyses can be great tools when wielded by great researchers. Wielded by researchers with clear philosophical axes to grind, they show the researcher exactly what he or she was aiming for. This study is a perfect example of such bias being portrayed as objective.

And it really begs the question -- what about all the research published in the past 19 years? Sure, some of it like the NEJM study mirror this data (on some level), but there's also a fair amount of solid, independent research (e.g., not pharma funded) that confirms SSRIs' effectiveness.

So instead of helping clear the waters, I'm afraid this study has only muddied them further.
posted by docjohn at 6:01 PM on February 26

The water is muddy. This is the whole point. Mental Wimp said it, and you missed it John:

This is pretty damaging for the hypothesis that the average effect of SSRIs is positive.

Now add that there is also a question of how thorough and transparent FDA's approval processes is, and how effective it is in identifying safe and effective drugs.

The water is muddy. Some people have made alot of money and other people have built careers in this muddy water. The appearance of impropriety has been breached.
posted by ewkpates at 4:12 AM on February 27

Bad Science has some more discussion on this paper and its implications. One of the points about another recent paper may illustrate the thought behind this latest analysis.


a paper in the New England Journal of Medicine dug out a list of all trials on SSRIs which had ever been registered with the Food and Drug Administration, and then went to look for those same trials in the academic literature. There were 37 studies which were assessed by the FDA as positive and, with a single exception, every one of those positive trials was written up, proudly, and published in full. But there were also 33 studies which had negative or iffy results and, of those, 22 were simply not published at all — they were buried — while 11 were written up and published in a way that portrayed them as having a positive outcome.

It seems to me that it is part of a building movement to counter the selective publication of much of the drug industry research. I think there may also be a bit of a backlash against SSRIs in particular, or more specifically, of the image of them that is being fostered by the pharma industry marketing. They are not a panacea that people with life issues should be hounding GPs to dish out, but should be prescribed in specific circumstances only. We don't need a valium for the 21st century.
posted by Jakey at 5:35 AM on February 27

Thanks for all the thoughtful comments, everyone.

I had the exact same reaction to the muddy water comment. The waters are muddy, dammit, and yet we are not supposed to know that? I was vaguely aware of the burying of the negative results trials, thanks Jakey for finding that .

It's worth noting that the PLOS group of journals is new, but pretty well respected and, most importantly for a study such as this, is non-commercial. Look at the unholy mess that includes conflicted physicians/researchers funded by pharma, at GPs bombarded by advertising and bombarded by patients who have been told by their TV to talk to their doctors about a drug, and, most germanely, the big journals. Note the multi-whole-page advertisements for drugs in those -- you are seeing the healthcare-industrial complex at work, and journals like PLOS are actually part of the solution. If this was in some other top-ranked medical journals, half of us would only have the press release to read as the actual article would be behind a subscription wall or available for the cost of hardcover book.

"Clinically significant" is, in effect, the most powerful marketing slogan any drug has. We need to keep a very very careful watch on how that is applied and, to be completely frank, I do not trust physicians to watch the henhouse on this issue [NOT PHYSICIANIST].
posted by Rumple at 9:15 AM on February 27

So, ahem............. I guess, what Ikkuyu2 said?

Medications + whatever else works is the way forward. No-one these days argues that meds alone are the panacea. (other than drug companies, or HMO's who don't want any form of mental illness on their books?)
posted by Wilder at 3:08 PM on February 27

The Medicated Americans: Antidepressant Prescriptions on the Rise. Close to 10 percent of men and women in America are now taking drugs to combat depression. How did a once rare condition become so common?
posted by homunculus at 12:29 PM on February 28

From Ben Goldacre:

It was fun to hear universal jubilation over the new meta-analysis showing once again that some antidepressants aren’t much cop in mild or moderate depression: most of all on the Today programme, where a newsreader said the industry was contesting the study on the basis that it was not in line “with patient experience”. I’ve always said that homeopaths mimic big pharma in their marketing spiel, but this is the first time I’ve seen it done the other way around, so bravo to pill peddlers of all shades.

In fact the new study added nothing (and it was ridiculously badly reported, see further down this page): we already knew that antidepressants perform only marginally better than placebo, and the National Institute for Health and Clinical Excellence guidelines have actively advised against using them in milder depression since 2004.

But the more interesting questions are around placebo.

Do drugs stop working if you know they are little better than a sugar pill? And do cultural factors, like our collective faith in a treatment, have a measurable effect on the benefits? On this, there has been a only tiny amount of highly tenuous research...............................
posted by Rumple at 9:18 AM on March 5