Who Loves Tamiflu?

Those who are paid to love it, by a wide margin.

This brief evaluation, published in Annals of Internal Medicine, asks the question: is there a relationship between financial conflicts-of-interest, and the outcomes of systematic reviews regarding the use of neuraminidase inhibitors for influenza?  To answer such a question, these authors reviewed 37 assessments in 26 systematic reviews, published between 2005 and 2014, and evaluated the concluding language of each as “favorable” or “unfavorable”.  They then checked each author of each systematic review for relevant conflicts of interest with GlaxoSmithKline and Roche Pharmaceuticals.

Among those systematic reviews associated with author COI, 7 of 8 assessments were rated as “favorable”.  Among the remaining 29 assessments made without author COI, only 5 were favorable.  Of the reviews published with COI, only 1 made mention of limitations due to publication bias or incomplete outcomes reporting, versus most of those published without COI.

Shocking findings to all frequent readers, I’m sure.

“Financial Conflicts of Interest and Conclusions About Neuraminidase Inhibitors for Influenza”

Original link in error … although, it’s a good article, too!

5 thoughts on “Who Loves Tamiflu?”

  1. Hi Ryan, thought this would have quieten down after this annals piece but new inbox message today from Public Health England asking me to prescribe Neuraminidase Inhibitors for Influenza as "There is good evidence that antivirals can reduce the risk of death in patients hospitalised with influenza." https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/370673/AV_full_guidance.pdf

    All seems to be based on this.

    any thoughts…



  2. So, there have been a couple responses to that Lancet Respiratory Medicine article in the BMJ.

    Initial comment: http://www.bmj.com/content/348/bmj.g2228
    Lancet author response: http://www.ncbi.nlm.nih.gov/pubmed/24788485
    Comment author response: http://www.ncbi.nlm.nih.gov/pubmed/24788584
    Lancet author second response: http://www.ncbi.nlm.nih.gov/pubmed/24788389

    Basically, Mark Jones critiques their results for lacking face validity, and mostly disagrees with them regarding how they handle immortal time bias. Essentially, from what I gather, immortal time bias magnifies effect sizes in observational comparisons because, by definition, patients in the cohort exposed to a medication intervention survived the initial period necessary to receive the intervention. Alternatively, patients who did not survive that period count as poor outcomes in the non-exposed group. Essentially, it's a survival bias. The original authors from the Lancet group, of course, disagree, and believe they adjusted for it using some sort of frailty model. Frankly, I'm underqualified to pick a winner here regarding these esoteric statistical bits.

    But, back to the mandate itself. While the term "good evidence" is bandied about, I think it more appropriate to suggest "weak evidence suggests a possible reduction in risk of death". I would further say this reduction is only likely in the setting of high-mortality influenza strains, such as the current prevalent H1N1. But, at the end of the day – once someone is ill enough to be hospitalized for a respiratory infection complicated by influenza, the relative additive cost of the neuraminidase inhibitor is negligible. And, if the patient-oriented outcome of interest is mortality, I am generally not opposed to initiating neuraminidase inhibitors on hospital inpatients with influenza.

Comments are closed.