Wednesday, February 27, 2008

Some Depressing News

A friend of mine sent me a link to an article yesterday, thinking it might be useful blog fodder. However, I didn't have the actual study in front of me--but thanks to The Frontal Cortex, now I do. If you ask me, Jonah has hit the nail on the head when he pointed out the biggest weakness in this study--all of the trials involved were, at most, eight weeks long.

For those unaware, this study was a meta-analysis. A meta-analysis involves doing no new research trials; instead, the results of a large number of old trials are compiled and analyzed using statistical methods. Results from fourty-seven trials were compiled for the meta-analysis; thirty-three of those were only six weeks long. Short trials are known to be less valuable in testing the efficacy of antidepressants; the longer the studies run, the farther apart the placebo and treatment groups usually get. For example, this study demonstrated that if you give patients escitalopram (an SSRI) for sixteen weeks and then switch half of them to placebo without them knowing about it, the placebo group will do significantly worse in terms of probability of relapse into depression.

It's interesting to see placebos compared to antidepressants. According to the PLoS study, placebos had about 80% of the therapeutic activity of SSRIs in the short term--that is, they were about "80% as good" in the six-to-eight week studies they analyzed. This is not actually all that surprising. For many depressed patients--or those with mental illness in general--the placebo effect is huge. Sometimes, the pro-active nature of seeking treatment is enough to bring depressed patients out of a slump, if only temporarily. Patients who feel as though they're finally doing something to fix themselves are likely to experience an emotional boost. And let's face it--not everyone on SSRIs is experiencing symptoms because they have low serotonin, one of the major neurotransmitters behind depression. Other diseases can cause patients to present with depression-like symptoms if they aren't properly ruled out before initiating treatment. That isn't likely the case with this study; before patients can participate in a drug trial for a new antidepressant, other causes of depressed mood must be ruled out.

The problem is that the mainstream media tends to get ahold of things like this and run amok making claims that "a new study proves antidepressants don't work," which is both incorrect and irresponsible. It's incorrect because all this meta-analysis (like many individual trials on the subject) suggests is that SSRIs must be taken long-term to have benefit, and that short-term reactions are largely attributable to placebo effects. There's a big difference between "antidepressants don't work" and "antidepressants have to be taken for months to see major improvement." The rate of relapse for patients switched abruptly to placebo is higher, even if the authors of the meta-analysis aren't convinced that antidepressants create "enough" improvement in the short term to be clinically significant.

To continue, this style of reporting is irresponsible because the everyman is a lot more likely to read a news article with a catchy headline than he is to seek out the original study and pore over the methods of the researchers, either because he has no idea where to look or because he hasn't taken enough statistics to have any idea what the hell these sciencey people are talking about. Which means that the everyman on antidepressants is going to take the words of some reporter at face value--perhaps higher than the words of his doctor, if enough reporters are all saying the same thing--and they're going to throw away their meds, decrying them as a useless waste of money. I've seen this happen before in clinical practice; many mentally ill patients take meds for a while, decide they don't need them any more (either because they "aren't working" or they "feel better"), pitch them, relapse, and start the whole vicious cycle over again.

Is there a tendency to "cover up" the results of clinical trials with negative outcomes? Sure. That's publication bias. But antidepressants aren't the only medications for which reporting bias is an issue--and just because some negative studies don't get published doesn't mean that the positive ones are automatically worthless. The negative studies don't necessarily get swept under the rug as part of some "grand pharma conspiracy." One negative study doesn't prove a drug is ineffective any more than one positive study proves that it is. If the overwhelming trend is positive or negative, then we can make judgments. But even trends can be misleading; you can connect the dots many different ways, depending on where the dots are and what kind of analysis you're doing. And that's one of the risks of meta-analysis studies. There's just as much room for spin there as anywhere else.

So don't throw out the Prozac. There's no such thing as an "official study" that proclaims antidepressants to be therapeutically useless. And any journalist who tells you otherwise is being sucked in by spin just as badly as the drug reps who are trying to push the products--they're just being spun in the opposite direction.

No comments: