A new meta-analysis in the Lancet tried to find out:
"Mirtazapine, escitalopram, venlafaxine, and sertraline were significantly more efficacious than duloxetine (odds ratios [OR] 1·39, 1·33, 1·30 and 1·27, respectively), fluoxetine (1·37, 1·32, 1·28, and 1·25, respectively), fluvoxamine (1·41, 1·35, 1·30, and 1·27, respectively), paroxetine (1·35, 1·30, 1·27, and 1·22, respectively), and reboxetine (2·03, 1·95, 1·89, and 1·85, respectively). Reboxetine was significantly less efficacious than all the other antidepressants tested. Escitalopram and sertraline showed the best profile of acceptability, leading to significantly fewer discontinuations than did duloxetine, fluvoxamine, paroxetine, reboxetine, and venlafaxine."I haven't read it yet, so I can't speak for its veracity.
2 comments:
I'd be interested to see what you make of it.
From my reading of the methods section, what they did was clever, but complicated and they don't spell out exactly what they did. So the result is hard to evaluate. I suspect that if they tweaked their model slightly they might have got different results (that's how it always is)...
Hmm...just had a quick look.
Like you I think I understand what they've done from a theoretical point of view but not necessarily from a practical perspective.
But it seems reasonable, they tested for inconsistency (which you must do for this kind of model to ensure that the estimates are transitive) and the estimates using the Markov chain Monte Carlo model fit reasonably with the straight meta-analysis, so looking at the drugs they say are good (with fluoxetine as common reference because this is the most frequent comparator):
Mirtazapine vs fluoxetine: OR 1.4 (MCMC) and 1.6 (MA) both stat sig
Escitalopram vs fluoxetine: 1.3 (MCMC) and 1.2 (MA), only the former is stat sig
Venlafaxine vs fluoxetine: 1.3 (MCMC) and 1.4 (MA) both stat sig
Sertraline vs fluoxetine: 1.3 (MCMC) and 1.4 (MA) both stat sig
I suppose you might ask what MCMC models add to the meta-analytic approach but I think you could argue that testing the assumption of transitivity is a valuable one, since any departure would make you worried about systematic differences between trials with different comparators, and it adds some statistical robustness to your meta-analytic contrasts which might be based on only a couple of trials, although I still have some reservations about how this approach might propagate errors down the chain.
I'm less convinced by the reboxetine conclusion which finds OR .68 compared to fluoxetine but even then looking at the straight meta-analysis shows that reboxetine is less efficacious than citalopram (.58), fluoxetine (.72), sertraline (.73), and venlafaxine (.45) (only the citalopram contrast is stat sig) and it is certainly therefore plausible.
Of course, you can't really trade-off the efficacy against the tolerability in any meaningful way, making first choice selection of an antidepressant still a matter of judgement (and, of course, in the NHS they aren't going to like you starting people straight onto venlafaxine) and these are aggregated efficacy figures so, as you say, since they all mostly seem to have some efficacy, there is still a place for the less effective drugs for those who don't respond to the first line drugs.
But, I must confess, this probably reflects well on sertraline which is cheap, - and I'll likely bear that in mind in the future (sertraline, citalopram and fluoxetine would be common first choices anyway).
Post a Comment