The meta-analytical study (ie one that systematically pools the results of other studies) is the work of Dr Irving Kirsch, from the University of Hull, and his colleagues, and is published today, 26th February, in the open access journal PLoS Medicine.
Antidepressants are prescribed for the treatment of clinical depression, and the most widely used are the “new generation” drugs, the SSRIs such as fluoxetine (Prozac) and venlafaxine (Effexor).
Previous meta-analytical studies of antidepressants have already suggested they have only modest benefits over placebos, and the authors pointed out that when data from unpublished trials are included, the benefits are so small they fall below the criteria for clinical significance. What has not been clear in the past however, is whether within this overall result, the effectiveness of antidepressants depends on how severely depressed patients are when they start treatment.
Kirsch and colleagues pooled all the available full data sets from all clinical trials submitted to the US Food and Drug Administration (FDA) for licensing four of the new generation of antidepressants, the SSRIs fluoxetine (Prozac), venlafaxine (Effexor), nefazodone (Serzone), and paroxetine (Seroxat, Paxil). The data came from both published and unpublished trials.
The point of including data from unpublished as well as published trials, is to avoid potential bias arising from the omission of “disappointing” unpublished findings.
SSRI stands for “selective serotonin reuptake inhibitors”. This new type of antidepressant, like the older ones, works by attempting to stabilize chemicals in the brain that influence mood, except that SSRIs specifically target and increase the circulating levels of a brain chemical called serotonin, a neurotransmitter that regulates mood. They do this by inhibiting the reuptake of serotonin so that more of it is available for binding to cell receptors.
Using meta-analytical techniques (a way of pooling data from a range of studies as if they were one big study with broadly the same objectives) the authors assessed the relation between the initial severity of depression and the improvements shown by drug and placebo groups, as well as the relation between initial severity and differences in drug-placebo improvement scores.
The results showed that:
- Drug-placebo differences got bigger as initial severity went up.
- This difference was hardly noticeable at moderate levels of initial depression, went up to a relatively small difference for patients with severe depression, and reached a level that would be classed as clinically significant only in those patients at the extreme end of the very depressed scale.
- The improvement seemed to result from the most severely depressed patients not responding as well to placebo compared to their less depressed counterparts, than because they responded better to the active drug.
Kirsch and colleagues concluded that:
“Drug-placebo differences in antidepressant efficacy increase as a function of baseline severity, but are relatively small even for severely depressed patients.”
“The relationship between initial severity and antidepressant efficacy is attributable to decreased responsiveness to placebo among very severely depressed patients, rather than to increased responsiveness to medication,” they added.
In other words, the difference in effect between drug and placebo was only clinically significant in those patients who were very severely depressed at the start of their treatment and this effect was more likely due to a weaker response to the placebo than a stronger response to the drug itself in that group.
This study is important because although licensing authorities like the FDA in the US and NICE (National Institute for Health and Clinical Excellence) in the UK have approved SSRIs for treating depression, there are nagging doubts about how effective they are.
Depression, which affects about 1 in 6 people at some point in their life, is a serious medical condition characterized by imbalances in brain chemicals that regulate mood. The illness, which can last for months and sometimes years, makes a person feel unmotivated, worthless, hopeless, and sometimes even that life is so futile it would be better to be dead. Depression is often a cause of suicide.
Severity of depression is measured using a questionnaire called the Hamilton Rating Scale of Depression (HRSD) which comprises up to 21 items. If the total score comes to 18 or more, the person is classed as severely depressed.
For an antidepressant to receive a license, clinical trials have to show that it can significantly improve the HRSD score compared to a placebo.
Different countries have slightly different clinical criteria for how much the HRSD score has to improve by before the drug can be licensed to treat depression. In the UK, NICE require that the drug show an improvement in the HRSD score of 3.
A previous meta-analysis of published and unpublished trials sent to the FDA for licensing these drugs showed they only have an average benefit of 1.8 HRSD points.
The reason this study was done was to find out if underneath this 1.8 average there might be subgroups of patients for whom the improvement score was significantly higher, perhaps within the range required by NICE.
And indeed, this is what Kirsch and colleagues found: the clinical criteria were only met when the drugs were used to treat the most severely depressed patients, that is ones with an initial HRSD score of 28 or more, at the extreme end of the scale. And perhaps just as important, is the finding that this effect did not arise as a result of responding to the drug, but because of decreased responsiveness to the placebo.
This last, rather surprising finding, provides a new direction for future research.
In the meantime, the UK’s Royal College of Psychiatrists urges patients not to stop taking their prescribed antidepressants without seeing their doctor first.
“Initial Severity and Antidepressant Benefits: A Meta-Analysis of Data Submitted to the Food and Drug Administration.”
Kirsch I, Deacon BJ, Huedo-Medina TB, Scoboria A, Moore TJ, et al.
PLoS Medicine Vol. 5, No. 2, e45
Published online: February 26, 2008.
Sources: PLoS Medicine press release, journal abstract and article.
Written by: Catharine Paddock, PhD