Email list hosting service & mailing list manager


Re: JAMA article Charlie Hathaway 29 Jan 2003 17:06 EST

The meta-analysis report in JAMA on financial conflicts of interest could show little difference between the "quality" of industry-sponsored and non-industry-sponsored research.  However, the study clearly shows that industry sponsorship leads to pro-industry conclusions in the published literature.  Why?

The authors acknowledge that assessment of research quality is difficult. Even if a basic study design is acceptable, you can never really tell from published reports exactly what people are doing.  Using or ignoring data points?  Repeating an experiment?  Employing the subtleties of English and paragraph organization to attract attention to something?  Or make it inconspicuous?  Or just being very selective about what gets mentioned in the discussion and the abstract?

Of course, what we would really like to know is WHY a researcher is doing something.  The standard definition of scientific misconduct includes falsification, which includes manipulation and omission of data that leads to inaccurate representation of research.  The definition also says that "honest error or differences of opinion" are OK.  What does "honest" mean here?  That someone was ignorant about the pitfalls of post-hoc analyses?  Try telling the pharmaceutical companies that "presenting one's research in the best light" is honest in academia but a crime in industry.  Convince them that strong defense of a grant renewal is all that different from strong defense of market share?

Our current ethos seems to view traditional COI and scientific misconduct policies as surrogate standards of truthfulness. Certainly, we don't want rules we can't enforce.  But public confidence in and support of biomedical research, if not all scholarly activity, hinges on perceptions of honesty.  The drug industry likes to say that there is no one more conflicted than an NIH principal investigator.  Let's face it, if they can get well people to demand that their doc give them Vioxx by showing nothing but Dorothy Hamill twirling, they can certainly convince the average tax payer that a big research grant that pays salaries can be a great incentive to stretch the truth.

Many like to say that we should not automatically be skeptical simply because a pharmaceutical company is providing financial support.  I disagree.  Industry influence is ubiquitous and WE DO KNOW WHY they are involved.  Regardless of how nice one little research grant may be, the overall result of industrial marketing and promotion is bad for medicine, bad for research, and bad for us.  If we are committed to preserving the research enterprise, we must do more than play public relations and damage control.  We need to combat any mechanism anywhere that diminishes objectivity and enhances bias.  And to be consistent in this commitment we must do more, in our own houses, to promote the ethical conduct of research.

Charlie Hathaway

At 11:39 AM 1/29/03 -0600, you wrote:
>Good Morning Group,
>
>I wanted to alert you to this article of interest in the latest issue of JAMA.
>
>
>Scope and Impact of Financial Conflicts of Interest in Biomedical Research
>A Systematic Review
>Justin E. Bekelman; Yan Li; Cary P. Gross
>JAMA. 2003;289:454-465
>
>
>
>
>scott
>
>
>Scott Davis
>Sponsored Programs Administrator
>University of Oklahoma Health Sciences Center
>Tel:     405.271.2090
>Fax:    405.271.8651
>email:  xxxxxx@ouhsc.edu
>
>
>======================================================================
> Instructions on how to use the RESADM-L Mailing List, including
> subscription information and a web-searchable archive, are available
> via our web site at http://www.hrinet.org (click on "Listserv Lists")
>======================================================================
>
>

======================================================================
 Instructions on how to use the RESADM-L Mailing List, including
 subscription information and a web-searchable archive, are available
 via our web site at http://www.hrinet.org (click on "Listserv Lists")
======================================================================