Hacker News Reader

Comment by Karunamon | Hacker News Reader

I'd personally take this with a grain of salt - if I recall correctly, the effectiveness of antidepressant medication is roughly equal to that of random chance.

Yeah.. high 40's percent relapse rate in the article.

View on HN

Take what with a pinch of salt? That MCBT is about the same as meds for chronic recurrent depression? Because the paper appears to show that it is as effective. Or that it works at all? Well, they didn't have a group that they took off medication and who didn't do therapy so we don't know what the relapse rate would have been for that group. I suspect that t would be higher than 45%

You also should remember that the placebo effect does actually contribute a large part of the effect of antidepressants. The research just shows that antidepressants don't have much of an effect above and beyond placebo.

I see therapy as being more useful and reliable in the long term than a placebo.

> The research just shows that antidepressants don't have much of an effect above and beyond placebo

For mild depression. For moderate depression they're better. For severe depression they're better when you get the right one.

An evidence based therapy is always a good idea.

I just did a search and came across this meta-analysis, which confirms your statement:

http://jama.jamanetwork.com/article.aspx?articleid=185157

Bad choice of words.. maybe "perspective" more than the suspicion that "grain of salt" implies, but the point being that meds aren't terribly effective anyways, so something that's nearly as effective of them is.. somewhat less impressive.

The studies can also take advantage of mean reversion and test patients that are experiencing uncommonly high levels of depression (eg, unresponsive to medication). Over time patients return to their usual level of depression (or whatever's being tested) and the treatment is deemed effective.

No, it cannot, the design is a a randomized controlled trial, the controls would also go back, what is measured is the difference.

Reply on news.ycombinator.com