Sunday, March 22, 2009

March 20: more JAMA backpedaling on COI / Robinson

Over at JAMA, the COI issue continues to roll on. On March 20, JAMA posted a letter from DeAngelis, explaining how JAMA has been virtuous throughout the Robinson/COI deal --essentially, since Robinson had not received pharma money for this exact specific Escitalopram study in 2008, he decided that it was unnecessary to note that he has been a hired spokesmodel for the makers of escitalpram; others later pointed this out, in a letter to JAMA, and JAMA had to acknowledge.

At this point, WSJ blog picked up the story and discussed the issue.

http://blogs.wsj.com/health/2009/03/13/jama-editor-calls-critic-a-nobody-and-a-nothing/

I blogged abt this study here:
http://www.medsvstherapy.com/2008/07/license-please-or-what-happened-to.html

In a Mar 20 JAMA letter, DeAngelis described the WSJ blog coverage as "sensationalized media accounts..." She also declared: “JAMA editors take issues of undisclosed conflicts of interest seriously...In investigating this allegation, we followed our standard polices and procedures..."

1. “Sensationalized”?
The blog was not “sensationalized.” It was reporting and discussing an important issue: this Robinson article had a ridiculous approach to analyzing the data, quite at odds with the quasi-survival analysis illustrated in the same article. There has already been the (unsupported) report of DeAngelis name-callling over this issue. I think the "sensationalized" description amounts to more name-calling.

2. Even more ridiculous:
Well, this “standard procedure” to deal with COI seems all fine and good. But what if JAMA assumed, with all of the abundent evidence, that there WILL be researchers trying to scam the system, and "investigate" proactively, rather than when caught?

Here is all ya gotta do:
look at the lead author. Look at the drug seemingly promoted. Figure out the pharmaceutical company that markets the drug. Enter author name and drug company into Google. Hit enter.
That is how I quickly and easily discovered Robinson’s COI when the article appeared.

Is it that hard? Is there no editor at JAMA with a computer with internet access? No one can assess this proactively? If you say you take this seriously, then take it seriously. Call me up – I will do it for ya.
Maybe it is the age-old strategy: it is better to ask for forgiveness than for permission. Is JAMA hoping no one notices these COI, then putting together a good press release on the portion of occasions when it is detected?

DeAngelis, and JAMA, falter on one simple scientific concept: the theoretical threat of bias must be respected. In other words, the mere presence of bias indicates that the scientific community should be conservative in accepting the results of such a study, and the burden is on the pharma-sponsored researcher to design methods, and report this design in the manuscript or, when not permitted by space, to be gracious and understanding when requested for such information. Anything less is less than scientific. If you are on the payroll, it is fair for us to assume COI. Even if the pharmaceutical company did not pay you for this specific exact study.

How can JAMA fail to get this? A guess: The editors are medical doctors first, and scientisits second. Medical doctors, AKA physicians, are generally not scientists. The exceptional medical doctor who goes out of his or her way to receive education, formally or informally, in science, may step up to the level of scientist, along with being at the level of clinician. Either way: the rules of science do not change. Science does not care what letters are behind your name. Science does not care how many lives you believe you have saved. Science does not care about your knowledge of Latin. Science cares about hypothesis testing, empirical evidence, and bias. If you conduct a study evaluating a patented drug whose manufacturer has paid you any money in the past several years, you are tainted. That is just the way it is. Do the right thing: share the info regarding the taint. Game over. Please try again.

No comments: