Tuesday, October 06, 2015

Take medical studies with a pinch of salt

Another article on medical issues in today's Globe is worth relaying. A study published recently in the journal Plos One suggests that psychotherapy (including cognitive behaviour therapy and interpersonal therapy) for the alleviation of depression is perhaps 25% less effective than previously thought.
This finding is the result of some medical detective sleuthing into publication bias in favour of encouraging findings. Published journal articles tend to exaggerate the benefits of psychiatric treatments, in much the same way as other studies have shown that they tend to exaggerate the benefits of antidepressant drugs, and for much the same reasons and to the same extent.
Many completed studies are never published, or even submitted for publication, mainly because the researchers think that a finding of no net benefit stands little chance of actually being published. Including such unpublished studies with negative or inconclusive outcomes in the overall results, as this review has tried to do, substantially reduces the observed effectiveness of the treatments, just as has been previously found in pharmacological trials. The study was not able to determine whether the authors of published papers have massaged data to make the treatment appear better than it really was  (as had happened in some of the drug trials), but that also seems distinctly possible. What's more, one assumes that a similar level of publication bias exists in other fields as well.
So, in conclusion, psychotherapy, and particularly cognitive behaviour therapy, is indeed modestly effective in treating depression, especially in combination with antidepressant drugs, but perhaps not to the extent previously thought. And peer-reviewed medical literature, although the best guide we have to the effectiveness of current practices and the promise of future developments, is good but not completely reliable.
The system is not broken, then, just slightly rickety. But maybe medical journal articles should come with a government health warning, and an added pinch of salt.

No comments: