Biostatistician Dr. Donald Berry of MD Anderson Cancer Center wrote to me recently, “My assessment of the landscape of observational studies, including much of epidemiology, ranges from bleak to parched earth.”
That should get your attention about why we – all of us who communicate about research findings – need to do a better job when communicating about observational studies.
That includes medical journals and journalists.
And here is a fresh example where both need improvement.
To begin with, let me acknowledge that the BMJ has a huge job, publishing more than 50 specialist journals and having someone write news releases about stuff in those journals. But that’s their choice.
Today, for the second day in a row (here is yesterday’s), I’m going to criticize a BMJ news release on an observational study – something I’d done for years until late last year when it appeared that improvement had arrived. Now it’s clear that it hasn’t arrived.
The Annals of the Rheumatic Diseases is one of those 50+ specialist journals published by BMJ. This week BMJ sent out a news release based on a paper published in that journal. The news release was headlined:
Gout may lessen chances of developing Alzheimer’s disease
The sub-head was:
Study finds 24% lower risk of Alzheimer’s disease amongst people with a history of gout
The 2nd sentence of the release was:
Gout appears to have a protective effect for the brain.
There was not one word in the news release about the limits of such observational data, nor about how this study can not prove that “gout may lessen chances”…cannot prove a “24% lower risk”….cannot prove a “protective effect.”
Placing qualifiers such as “may lessen” or “appears to have” is akin to whispering “I’m not sure” after you’ve just screamed “FIRE!!!”
Why does it matter? Because news stories then follow the lead of the news release and get it wrong.
Let’s be frank: do you really think that journalists regularly read the Annals of the Rheumatic Diseases? No, but the journal news releases are dropped in their email lap.
And then they write about what they see in the news releases.
As the New York Times Well blog did, reporting: “An Upside to Gout: It May Offer Alzheimer’s Protection.” Or it may not, something the blog post never explained.
Look at the confusing jumble of information journalists fall into when they try to report on observational data. Excerpts of the NYT blog:
The reason for the connection is unclear. But gout is caused by excessive levels of uric acid in the blood, and previous studies have suggested that uric acid protects against oxidative stress. This may play a role in limiting neuron degeneration.
“This is a dilemma, because uric acid is thought to be bad, associated with heart disease and stroke,” said the senior author, Dr. Hyon K. Choi, a professor of medicine at Harvard. “This is the first piece of data suggesting that uric acid isn’t all bad. Maybe there is some benefit. It has to be confirmed in randomized trials, but that’s the interesting twist in this story.”
Or, of course, the honest counter-argument to that second-last sentence is that “maybe there is no benefit.” What did this story deliver as actionable information? Remember: this blog is called Well….not Science. The authors of the blog state that the blog “sifts through medical research to help readers live well every day.” How did this piece help readers do that? By encouraging them to get gout? Imagine the new ad campaign in competition with “Got Milk?” Get Gout!
It’s not the first time I’ve criticized the New York Times Well blog for its coverage of observational studies. See “The NY Times Well blog isn’t always so well.” And it probably won’t be the last.
And this was neither the first time (I think it’s the 7th time!) and it appears it won’t be the last time that I’ve criticized news releases by BMJ for how they describe (or fail to describe) observational studies.
Guaranteed: that’s going to keep us busy.
Follow us on Twitter: