A new study published today is sure to set off another storm in the ongoing debate about the widespread prescription of antidepressants. Professor Irving Kirsch at the University of Hull and colleagues in the US and Canada report that new generation ‘SSRI’ antidepressants like Prozac or Seroxat mostly fall, “below the recommended criteria for clinical significance” (Kirsch et al. 2008). In other words, the most modern drugs prescribed for depression generally don’t work.
The study was particularly interested in whether the drugs had different effects on people with different levels of depression. Here is what they found:
- Mild depression: not tested as mild depression is usually treated with a ‘talk therapy’ rather than antidepressants.
- Moderate depression: antidepressants made “virtually no difference”.
- Severe depression: antidepressants had a “small and clinically insignificant” effect.
- Most severe depression: antidepressants had a significant clinical benefit – but see below…
Effectiveness limited even for severe depression
When Professor Kirsch and colleagues looked more closely at the data for those who were most severely depressed they uncovered more bad news for drug manufacturers. The antidepressant effect the drugs appeared to have, though small, was largely due to differences in the effects that the placebo had on the control group rather than better response to the drug.
Let’s unpack this a little.
The placebo effect means that even when you give someone a ‘fake’ antidepressant they still improve a little, simply because they expect to. This effect is so powerful and reliable that to be taken seriously drug studies have to compare depressed people taking an antidepressant to a control group taking a placebo.
What Professor Kirsch and colleagues found was that while the placebo effect was present for moderately depressed people, it disappeared for those who were the most severely depressed. This meant that antidepressants weren’t having any more effect on those who were more depressed, it’s just that in comparison to the control group that’s how it appeared. In reality what was happening was that the control group weren’t responding to the placebo.
The authors, therefore, conclude that there’s no point prescribing SSRI antidepressants to anyone but the most severely depressed people, unless other treatments have been tried and have failed.
Can we believe this study?
So the question is: can we believe the results? Well, the study used data from 47 clinical trials that had been submitted to the US Food and Drugs Administration (FDA). The FDA already has a rigorous set of criteria for including studies, so this suggests only quality studies were included.
The data from all these studies were then combined using a statistical technique called ‘meta-analysis’. This means all the studies were collected together and analysed as though they were all one huge study. By doing this you can increase the power of the study significantly.
Like many statistical techniques, though, there is some debate about the use of meta-analyses. For example it is often argued that they lump together studies with different protocols so that effectively you end up comparing apples with oranges. Whether this sort of criticism is valid depends on the study’s nitty-gritty details.
More broadly, we have to be careful about drawing conclusions from a single piece of work. There’s no doubt how high the stakes are for everyone: Professor Irving Kirsch has built a career on showing the power of the placebo effect, pharmaceutical companies have built their fortunes on studies proclaiming the benefits of SSRI antidepressants, while patients are stuck in the middle.
Despite this, the evidence does seem to be mounting up against SSRI antidepressants. Although previous studies seemed to show SSRIs were effective, recent work has suggested this might be due to a bias in the way research is reported (Turner et al., 2008). Studies which show no effect have a tendency to be ‘filed’ rather than being submitted for publication. This can result in a much more rosy picture being painted of a drug’s effectiveness than is really the case.
Either way, considering the number of people worldwide currently taking SSRI antidepressants, we can be sure this isn’t the end of the story.
» Read some of the strong reactions to the antidepressant study.
[Image credit: selva]
Kirsch, I., Deacon, B. J., Huedo-Medina, T. B., Scoboria, A., Moore, T. J., & Johnson, B. T. (2008). Initial severity and antidepressant benefits: a meta-analysis of data submitted to the food and drug administration, PLoS Medicine, 5(2), e45 EP
→ Try one of PsyBlog’s ebooks, all written by Dr Jeremy Dean:
Turner, E. H., Matthews, A. M., Linardatos, E., Tell, R. A., & Rosenthal, R. (2008). Selective publication of antidepressant trials and its influence on apparent efficacy, New England Journal of Medicine, 358(3), 252-260.