GlaxoSmithKline shares are down and the company thinks it’s our fault – we, here, being the media.
GSK saw sales of diabetes drug Avandia drop off after studies linked it to heart attacks. Now it has issued a profit warning for 2008 (details from: WSJ, NY Times, BBC, FT).
“My wish for the media is to be more sophisticated when they report scientific news. Debates now are being thrown into the public domain before scientists have given their opinion,” the Guardian quotes the company’s soon to retire CEO J P Garnier as saying at a talk this week.
A slide for the talk declares:
Media translation of scientific facts
Incidence: Less than 5 out of 10,000 patient years
As reported: “43% Increase in Heart Attacks”
Let’s have a look at this in an official Great Beyond analysis…
First up: the study triggering the Avandia stories last year was a meta-analysis (published in the New England Journal of Medicine), not a randomized controlled trial. Many criticisms have been leveled at this study (see coverage in Nature, Nature Reviews Drug Discovery; subscription required). Elsewhere in our stable, Nature Clinical Practice Cardiovascular Medicine called the paper “a rushed and incomplete examination”.
These concerns weren’t present in a lot of the reporting. Score one for Garnier.
Media 0, Garnier 1.
But are these debates really being “thrown into the public domain before scientists have given their opinion”. Well the NEJM paper was written by, err, scientists. Reporters covering the story quoted scientists. When would you like us to report something, JP? When you tell us to?
Media 1, Garnier 1.
How about ‘media translation of scientific facts’. Well the NEJM paper says, “In the rosiglitazone group, as compared with the control group, the odds ratio for myocardial infarction was 1.43 … and the odds ratio for death from cardiovascular causes was 1.64 …”
If you’re not familiar with odds ratios, 1.0 would mean there was no difference in risk of a heart attack between the groups; 1.43 means people taking Avandia (aka rosiglitazone) were 43% more likely to get heart attacks. No media fabrication there. And if you pull the PDF of the paper itself you can search it for the phrase ‘patient years’ which the media supposedly translated. You won’t find it. So it’s hard to see how the 43% figure — which is in the paper — can be seen as a poor “translation” of the incidence in person years — that isn’t.
Media 2, Garnier 1
That said, reporting on such things should in general always contain incidence figures as well as relative risk. It’s just good practice. And a lot of this reporting didn’t. Enough that Mr Garnier earns a draw? Hard to say – and I may be biased on the subject….
Anyway, here in all fairness are documents on what GSK thinks about Avandia and the presentations are here.
Unrelated Avandia story: as revealed by Nature News last month a peer reviewer for the NEJM review broke confidentiality and leaked the report to GSK weeks ahead of publication.