Americans who eat the most ultra-processed foods face a 47 percent higher risk of having had a heart attack or stroke compared with those who eat the least, according to a peer-reviewed analysis of federal dietary data from 4,787 U.S. adults. The finding, drawn from the most recent cycle of the National Health and Nutrition Examination Survey (NHANES 2021-2023), puts a hard number on a pattern that nutrition researchers have tracked for years but that now carries fresh statistical weight from a nationally representative sample.
Why a 47 percent cardiovascular risk gap demands attention
The study, published in the American Journal of Medicine and accessible via its journal article, compared adults in the highest quartile of ultra-processed food intake against those in the lowest quartile. After adjusting for confounders, the top-quartile group showed an adjusted relative risk of 1.47, with a 95 percent confidence interval of 1.06 to 2.04. That interval clears the threshold of statistical significance, meaning the association is unlikely to be explained by chance alone.
Cardiovascular disease in this analysis was defined as self-reported prior heart attack or stroke. The researchers used 24-hour dietary recalls collected through the What We Eat in America component of NHANES, with food items coded using USDA food codes from the Day 1 Individual Foods file. Those codes were then mapped to the NOVA classification system, which groups foods by their degree of industrial processing. Products such as packaged snacks, reconstituted meat products, and sugar-sweetened beverages typically land in the ultra-processed category.
One question the data raises but does not settle is whether the risk increase stems from something specific about industrial processing or simply from the fact that ultra-processed foods crowd out fruits, vegetables, whole grains, and other nutrient-dense options. If the real driver is displacement of better foods rather than a unique harm from additives or processing techniques, the policy and dietary response would look different. Re-running the NHANES models with a nutrient-density covariate in place of the NOVA ultra-processed food score could help distinguish between these two explanations, but that analysis has not yet been published.
NHANES data and a separate meta-analysis point in the same direction
The NHANES 2021-2023 dietary data files that underpin this study are publicly available through the National Center for Health Statistics’ online dietary portal. The raw records, including the Day 1 Individual Foods file (DR1IFF_L), document every food and beverage each participant reported consuming in a single 24-hour period. The USDA Agricultural Research Service’s Food Surveys Research Group provides companion documentation explaining how those recalls were collected, processed, and weighted to represent the broader U.S. population.
The 47 percent figure from the NHANES-based study does not stand alone. A separate systematic review and dose-response meta-analysis, available through PubMed Central, pooled results from multiple cohorts and examined how cardiovascular event risk changes across different levels of ultra-processed food exposure. While the meta-analysis drew on diverse populations and study designs, its pooled estimates pointed in the same direction as the single-study finding, reinforcing the pattern across independent datasets.
The convergence matters because no single observational study can prove causation. When a nationally representative cross-sectional analysis and a pooled review of longitudinal cohorts both identify elevated cardiovascular risk tied to ultra-processed food consumption, the evidence base becomes harder to dismiss as a statistical artifact of one dataset or one research team’s choices. The consistency across designs and populations is one of the key signals epidemiologists look for when judging whether an association might reflect a real underlying biological effect.
Self-reported outcomes and missing causal proof limit the 47 percent finding
Several gaps in the evidence deserve direct acknowledgment. The NHANES study relied on self-reported cardiovascular events rather than verified hospital records, death certificates, or clinical adjudication. People can misremember or mischaracterize a prior diagnosis, and those who eat differently after a heart attack may show up in a different intake quartile than they occupied before the event. This reverse-causation problem is a standard limitation of cross-sectional designs, and the authors’ adjusted models can reduce but not eliminate it.
The dietary exposure itself comes from a single day of recall. While NHANES protocols are well validated and the survey weights are designed to produce nationally representative estimates, one day of eating does not capture long-term dietary habits with the same precision as repeated measurements over months or years. The NOVA classification system, which sorts foods into four processing tiers, also groups together a wide range of products. A diet soda and a frozen pizza both count as ultra-processed, but their nutrient profiles and potential biological effects differ sharply.
Another limitation is that ultra-processed intake may be acting as a marker for broader lifestyle patterns. People who rely heavily on packaged snacks and ready-to-eat meals may also be more sedentary, sleep less, or have less access to preventive medical care. Although the NHANES models adjusted for several demographic and behavioral factors, unmeasured confounding remains a plausible explanation for part of the observed risk gap.
No direct author interviews or supplementary statements from the research team are available in the current record. The dose-response curves from the meta-analysis use exposure metrics that do not map precisely onto the NHANES 2021-2023 coding scheme, making direct comparison between the two studies less straightforward than their shared direction might suggest. Readers who want to examine the underlying methods and statistical models can find both papers and related work through the National Library of Medicine, which indexes peer-reviewed biomedical research.
How much weight should individuals and policymakers give this signal?
For individuals, the practical takeaway is cautious but actionable. The new NHANES analysis and the corroborating meta-analysis both suggest that diets dominated by ultra-processed foods are associated with a substantially higher likelihood of having had a heart attack or stroke. That pattern holds even after accounting for age, sex, and several lifestyle factors, and it aligns with broader evidence linking ultra-processed intake to obesity, hypertension, and type 2 diabetes.
Because the data are observational, they cannot prove that simply swapping ultra-processed foods for less processed options will cut cardiovascular risk by 47 percent. Still, the direction of the association, its consistency across datasets, and the absence of any known health benefit from heavy ultra-processed consumption make a shift toward minimally processed foods a low-regret move. Emphasizing vegetables, fruits, legumes, nuts, whole grains, and basic protein sources while reducing packaged snacks, sugary drinks, and ready-to-heat meals is compatible with many existing dietary guidelines and carries little downside for most adults.
For policymakers and public health agencies, the findings strengthen the case for monitoring ultra-processed food intake as a distinct dietary risk factor. That could mean incorporating processing-based metrics into national dietary surveillance, funding longitudinal studies that track ultra-processed consumption over time, or testing interventions that make healthier, less processed options more accessible and affordable. At the same time, the methodological limits of the current evidence argue against sweeping, processing-only regulations without parallel work to clarify which specific products or patterns drive the observed risk.
The 47 percent figure is not a definitive verdict on any one food or ingredient. Instead, it is a statistical signal that a certain style of eating-one heavily built around industrially formulated products-is linked with more cardiovascular disease in the U.S. adult population. As more NHANES cycles accumulate and additional cohort studies report their findings, that signal will either sharpen or fade. For now, it is strong enough to warrant attention, further research, and modest but concrete shifts in how Americans fill their plates.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
