Further reflection on antidepressant efficacy

Having first reflected (if that’s the right word) in the media, Cipriani et al now reflect on their network meta-analysis (see previous post) in Lancet Psychiatry (see article), which is a more sober affair. Again they admit that differences between antidepressants are small, but qualify this by saying that “exceptions exist”, but then do not spell out what those exceptions are. They come up with a slightly lower figure for placebo response in clinical trials than a BMJ editorial (see previous post), but who’s going to argue about a few percentage points in a matter like this? They don’t really say that the range of antidepressant responses for different drugs in the trials included in the analysis could be due to the trials themselves rather than the drugs.

As I said previously, none of this is new. To me, all the fuss about this study, which has led Cipriani et al to reflect on it, seems to have been created because this is 6 years work which only reaches weak conclusions. Going slightly off message, they go on to argue for open access to data from clinical trials at the anonymised individual patient level. Having always been open in this blog, I couldn’t agree more, although I suspect that all this will do is show up even more biases in the data.

Anyway, Cipriani et al seem to be agreeing the debate about antidepressant efficacy has not been ended by their study. As I have said several times previously, I would also encourage them to research the placebo amplification hypothesis, however difficult this may be.