Thread by @HengartnerMP

There are so many misrepresentations in this new commentary by Goodwin and Nutt about the efficacy of antidepressants… I don't know where to start. So here are just 3 examples of misleading statements in this paper (thread): cambridge.org/core/journals/…

First misleading claim by Goodwin & Nutt: “The number needed to treat (NNT) in studies with a mean drug–placebo difference on the HDRS scale of around 3 is between 5 and 7…”

Here they selectively report efficacy from the most favourable trials and they ignore that in most trials, the drug-placebo difference is much smaller than 3 points. The mean difference across all studies is 2, not 3, so NNT is not between 5 and 7, but rather between 8 and 10.

I give an example to demonstrate how misleading this claim is: If you select the tallest women in a population and compare their mean height to the average men, would that be a valid approach to claim that women are almost as tall as men? Not so...

Second misleading claim by Goodwin & Nutt: “[European regulators] found around an average 16% greater response rate following active treatment than placebo for newer antidepressants”.

Here they ignore that response rates (≥50% symptom reduction) are no valid outcome. If response rate for AD is 56% and that for PBO is 40% (16% difference), it still could be that on average AD were not better than PBO. An example follows...

Imagine change ≥10 points on depression scale is considered response and change <10 points non-response. Of 100 AD users, 56 people improved by exactly 10 points, the remaining 44 had 8 points. In PBO group, 40 people improved by 10 points, the remaining 60 people had 9 points.

So what’s the result? Response rate was 56% for AD vs. 40% for PBO, but mean change was 9.12 for AD vs. 9.40 for PBO. So, mean improvement would be even slightly better for PBO! See why differences in response rates can be so misleading? Why are response rates still in use?

Third misleading claim: “analysis of the long-term efficacy of antidepressants shows that in terms of protecting patients against a subsequent relapse to depression these medicines have an NNT of less than 3…”

Here they refer to discontinuation trials, where some patients who improved on AD are abruptly put on PBO. However, these trials cannot say anything about efficacy of AD, they merely indicate that abruptly discontinuing AD causes significant problems (patients in withdrawal know)

So, in conclusion, I am still waiting for a sound and convincing argument why antidepressants should be much more effective than indicated by the marginally small drug-placebo difference of 2 points. The arguments of Goodwin and Nutt certainly will not end this debate.

Will I write a letter to Acta Neuropsychiatrica in response to Goodwin and Nutt? No, probably not. Last time Martin @Martin_Ploederl and I wrote a letter our arguments were rejected as a minority opinion not worth reading in this journal. Yes, that's very scientific...

No, wait, one last misleading claim by Goodwin and Nutt, because this one is too good: “For no SSRI is the drop out rate statistically higher than placebo, which is what the literal interpretation of doing more harm than good would require.”

Strange argument, but conversely, according to their logic, doing more good than harm would require that drop out rate for AD is significantly lower than PBO, which it is not. So according to Goodwin and Nutt’s very own argument, AD are not useful…

Why is this a strange argument? Say, for instance, 10% of AD users drop out due to suicide attempts and 10% of PBO users drop out due to inefficacy (so drop out rate 10% each), according to Goodwin & Nutt this means that the drug does no more harm than good? Are they serious?

And by the way, in the meta-analysis they quote (Geddes et al 2003) the NNT for SSRI was 4.5, not „less than 3“