Sunday, August 2, 2009

Citation distortion and unfounded authority

We all know about publication bias—wherein those studies generating “positive” results are more likely to be submitted and published than are studies generating negative results. It’s easy to understand why, especially as it pertains to the single-center, before-after studies that predominate in the infection control literature. Who wants to tell the world that they spent hundreds of thousands of dollars swabbing noses and isolating patients, but didn’t see any reduction in infection rates? Likewise, if you invest a lot of time and energy in an intervention that works, you want to tell people about it (and, of course, add a publication to your CV).

A related phenomenon is “citation bias”. This occurs when authors selectively cite only those papers that support a claim, and either do not cite or distort those papers that are critical of the claim. Repeated citation of supportive claims and lack of citation of data that undercut the claims can then serve as “amplifiers” of the claim in question. Pure invention of facts can also occur in this way, as a proponent of a claim may state something as a hypothesis, and a subsequent author may cite the statement as a fact—a “fact” that then becomes firmly embedded through further amplification and citation bias.

I just ran across a fascinating paper in the British Medical Journal that examines these phenomena, using a specific belief system in the field of neurology as an example (the belief that beta-amyloid is produced by, and injures, skeletal muscle in patients with inclusion body myositis). The author, Steven Greenberg, uses social network theory to construct and analyze the “claim specific citation network” that undergirds this belief.

Many firmly held beliefs in the infection control world (e.g. that contact precautions prevent MDRO transmission in endemic settings, or that active surveillance is essential for MRSA control) are overdue for such an analysis.

Though I would add one additional form of bias to that discussed by Greenberg: editorial bias. For several years, the editorial leadership of one of our main infection control journals was held by persons with firmly fixed beliefs about how to control MRSA. Early in my career I submitted a letter to the editor that was critical of one of the papers published in the journal. I received a personal call from the editor telling me he couldn’t publish the letter. Not because he could identify anything factually incorrect about the letter, but because the letter would give “aid and comfort” to those who didn’t agree with his beliefs about MRSA control. Since then, I’ve spoken with investigators who have submitted papers to other journals that challenge the claim that active surveillance is essential to MRSA control, and have also had editors or reviewers express concern that the data undercut their beliefs about MRSA control (and therefore should not be published!).

1 comment:

  1. An amazing find! A home grown Richmond physician writer with an important message to share.

    Thanks for a readable blog with a valuable and valued message to share.

    Gayle Schrier Smith, MD

    ReplyDelete