Feb 26, 2008
A fascinating research study was published today by Irving Kirsch of the University of Hull, questioning whether or not 4 popular antidepressants are any better than placebo, and found that they weren't any better. Here's the study:
Both published and unpublished data on fluoxetine (Prozac), venlafaxine (Effexor), nefazodone (Serzone), and paroxetine (Paxil) were combined in a technique called meta-analysis, which brings together the data from many different research studies to see if they agree.
The researchers concluded that the antidepressants were not any better than placebo for mildly depressed patients. They found that although there was a benefit from antidepressants, it was not significantly better than what a placebo (pill without any medicine in it) did. The placebo treatments had a great deal of benefit, almost 80% improvement in depression symptoms. The antidepressants didn't do any better than placebo.
They further concluded that the benefit that these antidepressants had on severely depressed patients was from these patient's lack of benefit from placebo treatment. That is to say, severely depressed patients didn't get better from a placebo (pill without any medicine in it). Since severely depressed patients got better at about the same rate as the mildly depressed patients, by comparing this benefit to the lower benefit from placebo, overall they showed a benefit. Another way to think of it is that really depressed patients get better from antidepressants at about the same rate as mildly depressed patients get better, but since severely depressed patients don't get better from placebo, by comparing treatment versus placebo, the difference is greater.
It's important that patients on these medicines don't stop taking their medicines because serious side effects can occur, such as worsening of the depressive symptoms that they initially took them for, including thoughts of suicide for some patients.
This is just one study that will be further examined. The great thing is that it's been published in the Public Library of Science online journal. This is a place where the data and paper are publicly available, without any advertising or corporate support.
Reasons to think twice about the results of this study, is that the technique of meta-analysis is only as good as how well the 47 different data sets they've combined match. These data sets have to be combined in a valid way, controlling for different factors in each study. Some researchers disagree with this kind of process, saying that it's very difficult to control for confounding factors in the separate data sets, calling into question the validity of the results. I'm expecting the drug companies to be fighting these studies powerfully, and be curious to see how they try to disprove the results.