Nearly 48,000 U.S. adults tracked over two decades showed a clear pattern: the more ultra-processed food they ate, the worse their metabolic health markers looked and the higher their risk of dying early. Published in the American Journal of Public Health, the analysis drew on National Health and Nutrition Examination Survey data collected between 1999 and 2018, with mortality follow-up through the National Death Index. The findings add weight to a growing body of evidence that processing itself, not just poor diet quality in general, may carry distinct cardiovascular and metabolic costs.
Why the 48,000-person NHANES finding matters right now
Ultra-processed foods now account for the majority of calories consumed by American adults. That makes even small per-serving risk increases a population-level concern. The NHANES 1999–2018 analysis found that each 10 percent increment of daily energy from ultra-processed food was associated with higher BMI, higher HbA1c, and higher diastolic blood pressure. Those three markers track directly to the conditions that drive the most deaths in the United States: type 2 diabetes, hypertension, and cardiovascular disease.
The study classified foods using the NOVA system, which groups products by degree of industrial processing rather than nutrient content alone. Soft drinks, packaged snacks, reconstituted meat products, and many breakfast cereals fall into the ultra-processed category. Because the researchers controlled for overall diet quality, the association between ultra-processed food share and worse biomarkers persisted even when total calorie intake and nutrient profiles were comparable. That distinction matters: it suggests something about how these foods are manufactured or consumed, beyond their sugar or fat content, may independently affect health.
Mechanistically, several hypotheses are on the table. Formulation strategies that maximize palatability can encourage faster eating and reduce satiety, which in turn may disrupt appetite regulation and glucose control. Emulsifiers, stabilizers, and other additives could alter the gut microbiome in ways that promote low-grade inflammation. Packaging and processing at high temperatures may introduce compounds that stress metabolic pathways. The NHANES data cannot test these mechanisms directly, but the pattern of metabolic markers it reveals is consistent with those proposed pathways.
A testable next step would be to examine whether adults who cut their ultra-processed food share by 15 percent or more over two years show measurable drops in HbA1c and diastolic blood pressure compared with matched controls who maintain their intake, independent of total calorie change. The observational data from NHANES cannot answer that question on its own, but it sets the parameters for the kind of randomized trial that public health agencies have called for. Such trials would also help clarify whether specific subcategories of ultra-processed foods, such as sugary beverages versus savory snacks, carry different levels of risk.
Converging evidence from three U.S. cohorts and a meta-analysis
The NHANES results do not stand alone. A separate analysis that combined data from three large U.S. prospective cohorts along with a systematic review and meta-analysis of prospective cohort studies reported elevated risks for the highest versus lowest ultra-processed food intake across overall cardiovascular disease, coronary heart disease, and stroke. The consistency across different study designs and populations strengthens the case that the association is not an artifact of a single dataset or analytic method.
In that pooled analysis, participants with the greatest reliance on ultra-processed products had substantially higher rates of incident cardiovascular events than those who ate the least. While individual studies varied in how they measured diet and which confounders they adjusted for, the direction of association was strikingly uniform. That kind of alignment across cohorts is one of the signals epidemiologists look for when judging whether an observed relationship is likely to be robust.
The National Heart, Lung, and Blood Institute at the NIH has highlighted this emerging evidence in a public-facing review, noting that higher ultra-processed food intake is associated with higher cardiovascular risk while also flagging the limits of observational data and the need for controlled trials. That framing reflects the scientific consensus as it stands: the signal is strong and consistent, but causation has not yet been established through experimental evidence.
NHANES itself collects 24-hour dietary recalls and laboratory measures from a nationally representative sample, giving the data both scale and clinical precision. Participants attend in-person exams where trained staff measure blood pressure, draw blood, and record anthropometrics, then complete detailed interviews about their food intake. Those clinical measures are what allowed the American Journal of Public Health team to link ultra-processed food share not only to mortality, but also to intermediate markers like HbA1c that change long before a heart attack or stroke occurs.
Participants in NHANES are then linked to the National Death Index, which provides mortality follow-up periods and cause-of-death information. That linkage is what allows researchers to connect dietary patterns reported during a clinic visit to outcomes years or even decades later. By following tens of thousands of adults over time, analysts can estimate how different levels of ultra-processed food intake relate to the odds of dying from any cause, or from specific conditions such as cardiovascular disease.
Gaps in the data and what to watch next
Several questions remain open. The published summaries of the NHANES analysis do not include exact hazard ratios or confidence intervals for all-cause mortality, which limits how precisely outside researchers can evaluate the size of the risk increase. Cause-specific mortality breakdowns, separating cardiovascular deaths from cancer or other causes, are also absent from the publicly available materials. Without those numbers, the headline finding that ultra-processed food raised early-death odds is directionally clear but not fully quantified in public view.
Dietary recall data also carry known limitations. NHANES relies on participants remembering and accurately reporting what they ate in the previous 24 hours. Under-reporting is common, and ultra-processed foods may be especially prone to it because they are often eaten as snacks or in settings where portion tracking is difficult. Misclassification is another concern: some mixed dishes and restaurant items are hard to assign cleanly to a NOVA category, which could blur distinctions between ultra-processed and less-processed foods.
Residual confounding is a further challenge. People who eat a lot of ultra-processed foods often differ in other ways from those who do not: income, education, work hours, stress levels, and neighborhood food environments can all shape dietary patterns. If these factors are not fully measured and adjusted for, part of the apparent risk attributed to ultra-processed foods may actually reflect broader social and economic disadvantage. The study authors have not publicly detailed how much unmeasured confounding might remain after their adjustments.
Another open question is whether certain demographic groups are more vulnerable. Younger adults, for example, tend to consume the highest share of calories from ultra-processed products, but may not yet show the full burden of cardiovascular disease in follow-up data. Conversely, older adults with existing conditions could be more sensitive to dietary shifts. Future analyses that stratify results by age, sex, race and ethnicity, and baseline health status will be important for tailoring public health recommendations.
How individuals and policymakers can respond
For readers trying to act on this evidence, the practical takeaway is straightforward even if the science is still filling in details. Replacing a portion of daily ultra-processed calories with minimally processed alternatives, such as swapping a packaged snack for whole fruit or replacing a sugary cereal with oatmeal, aligns with the direction every major study in this space points. Even modest substitutions, repeated day after day, can meaningfully change the overall share of ultra-processed foods in a diet.
At the household level, simple strategies can help: cooking more meals at home from basic ingredients, keeping ready-to-eat whole foods like nuts and yogurt on hand, and checking ingredient lists for long strings of additives or refined starches. None of these steps require perfect adherence or a complete overhaul of eating habits. The NHANES findings suggest that incremental reductions in ultra-processed intake are likely to move metabolic markers in a healthier direction, even if some convenience foods remain in the mix.
For policymakers, the evidence base is beginning to justify broader interventions. Front-of-package labeling that clearly distinguishes minimally processed from ultra-processed products, procurement standards for schools and hospitals that favor less-processed options, and incentives for retailers to stock affordable whole foods in underserved neighborhoods are all under discussion in various jurisdictions. As randomized trials and more detailed cohort analyses emerge, those policy debates will have sharper numbers to work with.
In the meantime, the core message is unlikely to reverse: diets dominated by ultra-processed foods track with higher metabolic risk and higher odds of early death, while patterns centered on minimally processed foods appear more protective. The exact size of that risk, and the mechanisms that drive it, are still being quantified. But for individuals and health systems looking for levers they can pull now, dialing down ultra-processed intake remains one of the clearest, most actionable signals coming out of modern nutrition science.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
