Photo: Volkova (Shutterstock)
Ultra-processed foods have been linked with increased risk of cancer, heart disease, and dementia. These foods include a lot of fast foods and factory-produced snacks, which probably sounds about right—they are the things we might categorize as junk food. But to focus on the “processing” leaves more questions than it answers.
Who decides what is “ultra-processed,” anyway?
The definition of ultra-processed food, as the term is used in these studies, comes from an organization called NOVA. They divide foods into four categories.
Category 1 includes plants, animals, animal products, and fungi that we can eat, and they may stay in this category if they are dried or otherwise minimally processed. (Grapes and raisins both count.)Category 2 includes sugars, oils, and salt, and items made by combining these (like salted butter).Category 3 includes combinations of the above. Home-cooked meals pretty much all fall into this category. Category 4 is the “ultra-processed” designation. These include foods with additives that are considered industrial, and foods that are sold as ready-to-eat meals.
I get that scientists want some way to separate different types of cuisine, but this system has never had any kind of internal consistency. As we’ve previously covered, liquor is considered ultra-processed, but wine is not. A ready-to-eat burger is ultra-processed, but a plate of steak with a roll on the side is not.
It’s hard not to see this as an attempt to create artificial divisions between nutritionally similar foods. Sweetened cranberry juice is ultra-processed, but a bottle of grape juice that has the same amount of sugar naturally is not. You can see the official definitions, with examples, here.
How, exactly, are ultra-processed foods bad for us?
If studies compared these similar foods, and concluded that something about the processing was bad for us, that would be a scientific finding worth talking about. But the studies that supposedly spotlight the risks of ultra-processed foods aren’t like that.
They typically ask a group of people to recall what they ate recently—a method that is known to be unreliable and that some researchers have argued is “fatally flawed.” Then they score the food items against the NOVA scale. Then, if they’re doing their jobs reasonably well, the researchers will attempt to control for other factors like income. (Many cheap foods are ultra-processed, so they will be more commonly eaten by people who don’t have as much money.)
From there, it’s often possible to say that people who ate the most ultra-processed foods had a higher risk of some health condition than people who did not. But does that actually tell us anything about the processed food?
Not necessarily. Even if the controls were done properly and we have ruled out income, smoking status, and other factors as driving the differences between groups, the foods themselves aren’t necessarily comparable. Is the issue the salt content of ultra-processed foods? Sugar content? Specific preservatives or coloring or texture additives?
We can’t really take nutrition advice from these studies
It’s easy to jump to the conclusion that processed foods are bad for us, but we really don’t have the data to say which ones and why. We already know from the NOVA definitions that nutritionally similar foods can end up in entirely different categories depending on where they came from; compare the cranberry juice and the grape juice, for example.
In one recent press release (about a study that found a cancer link), the lead author said that “We need clear front of pack warning labels for ultra-processed foods to aid consumer choices.” But then the last paragraph of the press release states: “The researchers note that their study is observational, so does not show a causal link between ultra-processed foods and cancer due to the observational nature of the research. More work is needed in this area to establish a causal link.”
As a scientific study, it’s interesting and valid. It sets up researchers to do the promised “more work” to find out what, if anything, is protective or harmful about the different diets they identified.
The problem is when we take these preliminary results as warnings against eating certain foods. Ultimately we’re trying to use the word “ultra-processed” as a measure of nutrition even though it is not defined in terms of nutrition. If the problem with ultra-processed foods turns out to be their sugar and salt content, for example, then the issue would be with sugar and salt, not with whether we bought a burger from a fast food restaurant (Group 4) or made our own burger at home (Group 3).