The paper in the journal BMJ Open was entitled, “Association of skirt size and postmenopausal breast cancer risk in older women.” Association, not causation. That, alone, should have been a clue to journalists.
If they missed that, certainly they’d follow the news release from the journal, right?
That news release got it right:
“Going up a skirt size over a period of 10 years between your mid 20s and mid 60s is linked to a 33% greater risk of developing breast cancer after the menopause, finds a large observational study.”
Is linked to, not causes.
And “finds an observational study.”
These were key phrases apparently missed or misunderstood by many news organizations.But even if all of these clues were missed, how could you miss this from the news release:
“As this is an observational study, no definitive conclusions can be drawn about cause and effect, and there is likely to have been some variation in skirt sizing over the years, say the researchers.”
But look at the cause and effect statements appearing in many news stories. Some examples:
- HealthDay News tweeted: “Going up a skirt size every 10 years raised chances of developing breast cancer by 33 percent.” Nope. Sorry, but you need to use your 140 characters more wisely than that. The HealthDay story itself – if you dug that far in – did include the researchers’ caveats about not drawing cause and effect conclusions. But Tweets matter. They should be accurate as well.
- The Telegraph in the UK headlined it “Women who go up a skirt size ‘raise breast cancer risks’ ” – wrong again, even though, again, the body of the story included the appropriate caveats.
- The Times of London reported: “The risk of breast cancer increases with skirt size.” I always say that you can’t make a statement of increased risk if you haven’t established cause and effect. All you can talk about is a statistical association.
- The Daily Mail was wrong with: “Going up a dress size raises breast cancer risk 33%.”
- Oncology Nurse Advisor used causal language: “One skirt size increase every 10 years elevates breast cancer risk.”
The NHS Choices’ Behind the Headlines site analyzed the study even further, explaining other limitations of the research.
As always, I point journalists to our primer, “Does the Language Fit the Evidence? Association Versus Causation.” It provides good and bad examples of language used to describe observational studies.
Follow us on Twitter: