A few weeks back I blogged about an error-ridden paper from pharmaceutical company Janssen. It reported on the failure of a candidate antidepressant, JNJ-18038683. But it was rubbish: the Abstract contradicted the results of the study, wrongly claiming that the drug did statistically significantly better than placebo in a particular analysis, when it didn't. And they repeatedly mixed up the names of two other antidepressants.
I posted on May 17th. On May 23rd, a new version of the offending article quietly appeared on the journal's website: here it is. But the old version is still up, although I'm not sure it's intended to be, because the link is gone, so I think the only way to find it is from this blog.
Interestingly, the new manuscript corrects those two issues I noted. As far as I can see those are the only changes. Coincidence?
The original paper was a "Fast Forward" online accelerated publication manuscript, and it's common for these to be corrected in minor ways when they're officially published (i.e. when they appear in print), but that's not what happened here, because the new version is still a Fast Forward. This kind of revision is unusual.
So it seems that someone at Janssen is reading this blog!
If so: Hey. Believe it or not, I'm on your side, I want you to succeed. I suffer from depression, and I've love it if you came up with an actually good new antidepressant. But your industry hasn't released one in at least 10 years (be honest), and you're not fooling anyone with clever statistical tricks.
Forget that, and invest in some proper science. For example, ketamine is clearly the most exciting potential antidepressant right now, and glutamate could be the next serotonin. But because no-one has tested it against an active placebo, no-one, including you, knows whether it's really working. Run a really serious trial of ketamine, if it does work, then that's your next generation of antidepressants right there. If not, better to know that now than later when you've sunk billions into the idea.