The roller coaster ride of uneven quality of the New York Times Well blog was on display again as they posted, “Some Fruits Are Better Than Others.” Excerpt:
Recent studies have found that eating a greater variety, but not a greater quantity, of fruit significantly reduces the risk for Type 2 diabetes. This made researchers wonder whether some fruits might have a stronger effect than others.
Using data from three large health studies, they tracked diet and disease prospectively over a 12-year period in more than 185,000 people, of whom 12,198 developed Type 2 diabetes. The analysis appears online in BMJ.
After controlling for many health and behavioral factors, researchers found that some fruits — strawberries, oranges, peaches, plums and apricots — had no significant effect on the risk for Type 2 diabetes. But eating grapes, apples and grapefruit all significantly reduced the risk. The big winner: blueberries. Eating one to three servings a month decreased the risk by about 11 percent, and having five servings a week reduced it by 26 percent.
Substituting fruit juice for whole fruits significantly increased the risk for disease.
As a journalist-reader of ours pointed out, there was a lot of causal language used inappropriately in this brief story about observational studies. Examples – all highlighted in bold above.
When I read the lines emboldened, all the other text becomes blah, blah, blah in my mind because the story has crossed the line from reporting on statistical associations and entered the world of making cause-and-effect statements.
We’ve written about it before many times. Examples:
Jane Brody’s One-Sided Take on Sodium. Excerpt of what we wrote: “There has never been a study that’s definitively proven that we can save any lives – let alone a million of them – by reducing our intake of sodium from current levels. All of the figures Brody cites are derived from observational studies that cannot prove cause and effect, or else from clinical trials of blood pressure drugs that assume a similar benefit for sodium restriction (even though the effects on health outcomes may well be different).”
Please, Grey Lady, don’t spill more coffee observational studies on us. Excerpt of what we wrote: “On its blog, at least, if not in print, the New York Times has all the room in the world to explain things like this. Use links if you must. But please, Grey Lady, don’t let your writers contribute to the back-and-forth ping-pong games of “coffee lowers risk
coffee heightens risk” stories that seem to endlessly pour forth from the coffee pot of observational studies.”
It wouldn’t hurt if journalists and readers reviewed our primer, “Does the Language Fit the Evidence? Association Versus Causation.”
The words matter. Accuracy matters. I shouldn’t have to tell the New York Times that. But read some of the comments left online by readers of the “Some Fruits Are Better Than Others” piece. Readers are telling the Times to get with it as well. Excerpts of those comments:
- Type 2 diabetes is to a large degree a disease of poor people. So any factor that is related to socioeconomic status will correlate to lower diabetes risk, because wealthier, more highly educated people tend to have a lower diabetes risk.Blueberries are probably a good surrogate for wealth/education. If you can afford $5 a pint for blueberries twice a week, you probably have a higher income and education level and less chance of coming down with diabetes.The exact same correlation would be found for French wine, truffles, and good balsamic vinegar.Likewise, you can “reduce your diabetes risk” by buying Fendi purses, Ferraris and first-class airline tickets.
- Anyone who buys blueberries five times a week would probably have an intense interest in health & probably not be rounding out their diet at McDonalds the rest of the week.I doubt blueberries cause reduction in diabetes. They are just proof that having access to fresh produce you can afford, year-round, will keep you healthier. I’m not sure that’s news.
- As my Statistics professor tried to drum into our heads:“Correlation is not causation!”
- These comments are very helpful. It is a new journalistic phenomenon that the NYTimes comments contain as much wisdom as the original articles.
- This was a poorly written article that created more questions than it answered.
- I wish these articles would always put a disclaimer that studies like these do not prove causation or prevention.
It is weak science and worse reportage; it states straight-up that “eating grapes, apples and grapefruit all significantly reduced the risk (of diabetes)”. Obviously that is not what the study says; it could show a CORRELATION between eating certain fruits and remaining non-diabetic — not a causation — and we (should) all know that CORRELATION IS NOT CAUSATION
- when every new correlative study is heralded in the media without any caveats, it really does a disservice to the general public. With these sorts of epidemiological studies, there are irreducible numbers of confounding factors, but the media presents the study as if showing a CORRELATION between a certain fruit and a certain disease means that if you change your diet you will reduce your chances of developing that disease.By that line of reasoning (which is known in logic as a post hoc ergo propter hoc fallacy), we can scientifically establish that the crowing of roosters is correlated with the sun rising. However, to suggest that roosters cause the sun to rise would be folly.Before you can conclude, for instance, that eating blueberries reduces the probability of developing type II diabetes, you need to establish the biochemical pathways through which they might work. Then, from a Bayesian perspective, the correlative evidence becomes a much stronger basis for establishing a causal relation, because you have a strong prior probability based on chemistry.But the media either does not have good science writers or they feel that pushing studies in such a way to encourage people to change their lifestyle based on inadequate medical evidence is a way to sell advertisements and subscriptions.
Follow us on Twitter:
and on Facebook.