BREAKING: Eating organic food cuts risk of POVERTY by 36%!
That's just one of the many things you'll learn about organic eaters in that new cancer study that is a LESSON in crap nutritional epidemiology! Where shall we begin ... I know, DATA COLLECTION! jamanetwork.com/journals/jamai…
That's just one of the many things you'll learn about organic eaters in that new cancer study that is a LESSON in crap nutritional epidemiology! Where shall we begin ... I know, DATA COLLECTION! jamanetwork.com/journals/jamai…
All of the study data are self-reported. Organic consumption, and also diet quality, weight, income, exercise, and other data points that are NOTORIOUS for being inaccurately reported. Here’s an example from Mintel on organic consumption: mintel.com/press-centre/f…
What’s notable about that study is that consumers report buying MUCH MORE organic food than is sold in the US. People ALWAYS also report being thinner, richer, and more active than they actually are. You CANNOT take these data to the bank.
But let’s move on to the actual data. Among the nearly 70,000 participants, there were 1340 cancer cases, which is 2% (I’m assuming one cancer per person, which of course might not be the case). We’re starting with a SMALL number, and then there’s CONFOUNDING.
Organic food consumption data are CONFOUNDED every which way from Sunday, because people who eat a lot of organic food are different from other people in many other ways. They’re 36% less likely to be poor (as defined by the bottom income tier in this study).
They’re also more likely to have a college degree, be female, and be older. They eat more vegetables, less processed meat, and a more healthful diet in general. They weigh less. They drink less.
To figure out whether decreased cancer incidence is due to organic food or any of that other stuff, including stuff the study DOESN'T COVER, you have to CORRECT for it all. GOOD LUCK WITH THAT. I’m no statistician, but I feel statistical significance slipping away.
Not to mention that WHICH of those you correct for, and HOW you correct for them, can yield WILDLY different answers. I learned at my father’s knee (he’s a mathematician) that if you can CORRECT the data you can show anything you want.
But one final (promise!) issue looms large. The data come from the NutriNet-Sante study, a population study that tracks MANY health outcomes, of which cancer is only one. Did the researchers look at OTHER outcomes, and pick the one with positive results? clinicaltrials.gov/ct2/show/NCT03…
We don’t know because they don’t tell us. The pre-registration is only for the main study, not this particular study. Researchers are under intense pressure to produce positive results, and a paper that showed no associations either way might go nowhere.
This all makes me sad because I support organic agriculture. Just today, I visited an astonishing organic farm. Super high yields from double-cropped, biodiverse complex rotations of grains & legumes. I want THAT to be the story. Not bullshit epidemiology THE END.
