Conflict of Interest, or Concordance of Interests.

My last blog on ultrasound and fracture healing brought up the topic of authors with conflicts of interest. This is where the authors have something to gain from publishing favourable results of medical therapies; benefits that might include royalties, stock options, research funding and my favourite: consultancy fees (this is fancy term for a relatively simple process whereby a company gives money to a doctor, often large amounts and at regular intervals). Sometimes the authors are direct employees of the company making the device or drug in question. And sometimes these employees write the article for the esteemed senior author to put his or her name to (ghost writing), thereby giving the article authority.

But does the presence of conflict of interest influence whether or not the study is actually favourable – does it bias reporting? I could list examples where sponsored studies show a benefit, and independent studies do not, like in my last blog. But I am probably biased myself, and my recollection or perception does not represent scientific evidence of conflict of interest causing bias in reporting. This, however, does. And this. And this. In fact, I can’t remember the last time I read a sponsored study that was not favourable for the product in question.

Why does a conflict of interest lead to biased (one-sided) reporting of results? Because of what I call a Concordance of Interests. A favourable study supports the interests of all parties involved, as follows:

1.     It is in the interest of the surgeons writing the article, often because they are getting paid (directly or indirectly) by the producer, or because it justifies the surgery by which they earn a living

2.     It is in the interest of the company paying for the study as it will sell more of the product.

3.     It is in the interest of the publishing journal, as positive studies are more newsworthy and more likely to be referenced (therefore raising the impact factor of the journal).

Unfortunately, the outlook is not so good, as more and more studies are being conducted by the manufacturing companies themselves (rather than paying others to do them, they are keeping it more “in-house”). This not only gives them control over the study, but control over whether or not to publish the results.  Guess which ones they publish? A good example of this is with antidepressant drugs. Reviewing all published trials would lead you to conclude that antidepressants are more effective than placebos. In a study by Irving Kirsch, he looked at this by gaining access to all of the studies performed (not just the one that were published: he managed to get access to the unpublished ones held back by the manufacturers, using the Freedom of Information Act). When he summarised the data from all studies performed (published or not), he found that the antidepressant drugs were no better than placebo except for the most severe cases, where their effect was marginal. Yet these drugs are still being used today. Commonly, and for depression of any severity.

I have worse news: the authorities feel that the problem is now solved because of the strict rules that now apply regarding the declaration of a conflict of interest that doctors must make when publishing or presenting research. The objection I have is that declaring a conflict of interest does not remove it. It might lead the reader to be more wary, but it doesn’t change the fact that the conflict exists. It’s like fixing your leaking tap by recording how much water is coming out of it.

Addit 28 July 2012: Former editor of BMJ on conflict of interest, here.