publication bias IV – close encounters of the second kind…

Posted on Sunday 22 May 2016

I doubt that my discussion of the difficulty getting contrarian studies published would stand up in Evidence-Based Medicine court – too few examples to be called anything but anecdotes. But if you asked the few investigators who’ve given it a shot, I’d bet p would approach (1 ÷ ∞). Consider, for example, this contentious [and telling] response to Juireidini et al’s submission:

from Dr. Greenblatt [Editor of the Journal of Clinical Psychopharmacology]

We do not share your concerns about what you term "ghostwriting" — more properly described as manuscript preparation assistance. This is just another form of assistance or collaborative effort in the course of research — similar to technical assistance with measurement of plasma drug levels, or statistical consultant input in analysis of data. In the end, the listed authors take responsibility for the content of the manuscript, and that is what matters.
We also are not worried about the participation of the pharmaceutical manufacturer in the execution of the study, or the preparation of the manuscript. This is to be expected — they are the sponsors, and they have the most knowledge about the candidate drug and the data… We also note that we ARE concerned that you are serving as a reimbursed expert witness on behalf of the plaintiffs, proposing in the present manuscript what we expect are similar arguments as presented in the context of the litigation.
With all of that said, we certainly could reconsider a revised manuscript in which the focus was ONLY on the scientific content of the paper(s) in question. If you disagree with the scientific content or its interpretation, that can be presented, but without the court documents and internal E-mails, and without accusations of malfeasance, misrepresentation, manipulation, whitewashing, complicity, etc. The issue of manuscript preparation assistance is also not in the picture…

Greenblatt’s response misses the central point that by ignoring protocol directed exclusions for the primary outcome variable and changing a secondary outcome altogether, the analysis falsely reported a negative study as positive. Perhaps that’s what he didn’t want to hear.

But even in the BMJ response to Le Noury et al where the journal was sympathetic to the reanalysis, they performed their own audit of the harm findings because of similar concern about COI, and asked for additional post hoc efficacy analyses  [multiple imputation]. To borrow a $5 word from theological scholarship:
     1. a method or theory of interpretation
The RCT submissions are viewed with a hermeneutic of acceptance [innocent until proven guilty], and the contrarian submissions are evaluated under a hermeneutic of suspicion [guilty until proven innocent]. There’s not much a contrarian author can do about that. The journals are independent entities and they set their own sails.

Publication bias of the first kind [not publishing negative studies] presents a false view of a drug by inflating the drug’s profile. Publication bias of this second kind [not publishing or under-publishing contrarian articles – criticism] achieves the same result through a different mechanism. The first is calculated, intentional. The second is more insidious, and often has to do with the general climate. Dr. Greenblatt’s comments aren’t about the specific drug study criticized by Jureidini et al. He’s reacting to their implied criticism of the system that he, himself, is a part of. He conflates the legal challenge that produced the documents used in Jureidini et al and the criticism of the specific drug study itself, and lashes out at both in a single breath [rejecting the article out of hand].

Cosgrove et al has had a similar fate so far. The paper not only presents a right-sized view of Vortioxetine and exposes the mechanisms used to inflate it in the first place, it also looks at the structure of the system that allowed that to happen and suggests reform. We don’t know the details, but the net result is that their paper is buried and largely inaccessible. I doubt anyone thought, "Let’s block this paper from the general discourse" along the way. But that’s what happened. The article says "Your system needs to change" to an audience that is fine with the system as it is. I agree with these authors that the system needs to change. And though my recipe for change might differ somewhat, I would welcome these opinions to the conversation. But that’s not the mainstream point of view – unfortunately reflected in their article’s fate.

I first heard the phrase, "Sunlight is the best disinfectant" in Ben Goldacre’s initial Ted Talk [see something of value…]. And it is dead-on accurate. It’s the motivating force behind AllTrials, around Data Transparency, with articles like Restoring Study 329… and The citalopram CIT-MD-18 pediatric depression trial…, blogs like this, etc. I think the main problem with Cosgrove et al right now is not the Impact Factor of the journal where it’s published, but the fact that it’s not easily accessible on-line. Negotiating ResearchGate is not the same as just clicking a link. The other articles mentioned here are on-line, and I would hope that theirs will be too at some point. It could use the kind of sunlight only the Internet can provide…

Sorry, the comment form is closed at this time.