Ultra-Processed Food: Is the Problem the Processing—or Everything We Put Into It?

Walk into any grocery store on Chicago's North Side and you can watch the ultra-processed food debate play out in real time: shoppers scanning labels for "no seed oils," clutching bars that promise to be "clean," while a cart two aisles over is stacked with frozen dinners nobody's pretending are farm-to-table. Somewhere in between is where most of us actually eat, and that's precisely the terrain this story lives in—not the extremes, but the murky middle where scientists are still arguing about what, exactly, is going wrong.

The term "ultra-processed food" has become a cultural shorthand for everything wrong with the modern diet, blamed for rising obesity, diabetes, and a general sense that food isn't food anymore. And there's real science behind the alarm: population studies across multiple countries link diets high in ultra-processed food to worse metabolic health, and a landmark NIH-funded trial found people simply eat more of it. But "ultra-processed" is a classification system, not a nutrient, and lumping together diet soda, industrial white bread, breakfast cereal, and plant-based meat substitutes under one label invites a question the public conversation mostly skips: is it the processing that hurts you, or is processing just how we happen to deliver the sugar, salt, fat, and calorie density that were already known to be problems?

This isn't an academic quibble. If it's the processing itself—some novel mechanism tied to industrial manufacturing—then reformulating a product's texture or additive profile might not help much. If it's the calorie density, hyper-palatability, and nutrient displacement that tend to travel with heavy processing, then the fix looks different: less about banning a category and more about reengineering what's actually in it, and how much of it we're induced to eat.

Umbrella reviews—studies that pool findings across many systematic reviews and meta-analyses—paint a more mixed picture than headlines usually suggest. The evidence linking higher UPF intake to obesity and to markers of cardiometabolic risk is comparatively consistent and appears across many independent cohorts. The evidence for some other outcomes sometimes attributed to UPF, including certain cancers or specific mental health conditions, is thinner, more heterogeneous, and more prone to reversal as new studies arrive. That unevenness is itself informative: if "processing" were a single unified causal mechanism, you'd expect a more consistent signal across outcomes rather than strong evidence for some and weak, conflicting evidence for others.

One of the more interesting frontiers is NIH-funded metabolomics research attempting to identify objective biological biomarkers of ultra-processed food consumption—chemical signatures in blood or urine that correlate with UPF intake, bypassing the well-documented unreliability of self-reported dietary recall, where people misremember or under-report what they eat by wide margins. If validated, such biomarkers could let researchers test the UPF-disease link with far less measurement error than current cohort studies allow. That work is promising but still in progress; no validated, widely used UPF biomarker exists yet, which means most of the current evidence base still rests on the exact kind of self-report data known to be noisy.

There's also the question of reverse causation and residual confounding common to all diet epidemiology: people who eat heavily ultra-processed diets may differ from those who don't in ways that are hard to fully measure—income, time pressure, baseline health literacy, activity levels, or existing metabolic conditions that shape both food choice and disease risk. Cohort studies do their best to statistically control for these factors, but no amount of adjustment fully substitutes for randomization, which is precisely why the Hall trial carries outsized weight in this debate despite its small size—and precisely why one small trial, run in a metabolic ward for two weeks, cannot settle a question about years of everyday eating.

Even researchers sympathetic to the UPF framework increasingly argue that the category needs to be split. Some proposed refinements to NOVA or entirely new frameworks attempt to separate ultra-processed foods that are nutritionally reasonable (fortified plant milks, whole-grain packaged bread, some plant-based proteins) from those that are calorie-dense and nutrient-poor (sugary drinks, reconstituted meat products, packaged snack cakes). If those refinements gain traction, "ultra-processed" as a single monolithic villain may eventually give way to a more granular set of categories—which would vindicate the critics who've argued all along that NOVA, as currently used, is too blunt an instrument to carry the weight put on it.