Older Americans who get the largest share of their calories from ultra-processed foods face a 58 percent higher risk of developing dementia, according to a peer-reviewed study of more than 5,300 adults tracked over seven years through the University of Michigan Health and Retirement Study. The research, published in the American Journal of Public Health and led by investigators at the Harvard T.H. Chan School of Public Health, also found a 46 percent higher risk of cognitive impairment among those in the highest consumption group compared with the lowest. With ultra-processed items now supplying more than half of total calories in many American diets, the findings sharpen a question millions of families will eventually confront: whether everyday food choices are quietly accelerating cognitive decline.
Why a 58 percent dementia gap demands attention right now
The size of the risk increase matters because dementia already affects roughly one in nine Americans aged 65 and older, and no approved drug reverses the disease once it takes hold. A 58 percent elevation tied to dietary pattern, rather than genetics or age alone, points to a modifiable exposure. The study drew on nationally representative data collected between 2013 and 2020, meaning the dietary habits it captured reflect the same packaged-food environment that still dominates U.S. grocery aisles and fast-food menus.
One testable question emerging from the data is whether a modest dietary shift could meaningfully lower risk. If a person aged 60 to 75 replaced roughly 15 percent of ultra-processed calories with minimally processed alternatives over two years, would that produce at least a 25 percent relative reduction in new cognitive impairment? The observational record cannot answer that directly, but the dose-response pattern in the Health and Retirement Study cohort suggests the relationship is graded: more ultra-processed food, more risk. That gradient is exactly the kind of signal researchers look for before designing intervention trials.
How HRS data and parallel cohorts built the evidence base
The primary analysis, using HRS dietary records, classified intake with the NOVA food-grouping system and tracked participants for incident dementia and cognitive impairment across multiple waves. Participants in the top tier of ultra-processed food consumption showed both the 58 percent higher dementia risk and the 46 percent higher cognitive impairment risk after adjusting for confounders including age, education, and chronic disease burden. The NOVA system distinguishes minimally processed foods like fruits, vegetables, and plain grains from ultra-processed items such as packaged snacks, sugary drinks, and many frozen meals, allowing researchers to focus on the most industrially altered products.
A separate analysis of the same HRS panel broke ultra-processed foods into specific categories and found that distinct product groups carried steeper risk. Processed meats, sugary beverages, and ultra-processed animal products stood out as categories with the strongest individual associations with incident cognitive impairment. That granularity is useful because it moves the conversation beyond a blanket warning about “processed food” and toward specific items people can identify in their own kitchens, from deli meats and sausages to sweetened sodas and flavored dairy desserts.
Additional peer-reviewed work using the same HRS-linked dietary and cognitive data examined performance across several domains, including memory, executive function, and processing speed. The pattern held across domains, reinforcing the idea that ultra-processed food exposure is not selectively harming one type of thinking but is associated with broad cognitive erosion. Participants with higher intakes tended to show faster declines on composite cognitive scores, suggesting that the impact is diffuse rather than limited to a single mental skill.
Outside the HRS cohort, other longitudinal studies have reported comparable patterns. In the Framingham Heart Study, a prospective analysis linked higher daily servings of ultra-processed food to elevated dementia and Alzheimer’s disease risk, echoing the magnitude and direction of the HRS findings. A systematic review and meta-analysis in the Journal of Neurology pooled results from multiple observational cohorts and concluded that greater consumption of ultra-processed products was consistently associated with higher odds of cognitive impairment or dementia, although effect sizes varied depending on whether exposure was measured by quartiles, servings, or percentage of total energy. Together, these independent datasets strengthen confidence that the HRS results are not a statistical fluke.
What ultra-processed diets may be doing to the brain
Researchers are still mapping the biological pathways that could connect a supermarket cart full of ultra-processed items to a higher dementia risk years later. One leading hypothesis centers on chronic, low-grade inflammation. Diets high in refined carbohydrates, added sugars, and certain fats can drive systemic inflammation, which in turn is linked to vascular damage and neurodegenerative changes in the brain. Over time, that inflammatory burden may contribute to the accumulation of abnormal proteins and the loss of neurons involved in memory and reasoning.
Another proposed pathway involves the gut-brain axis. Ultra-processed foods often contain emulsifiers, artificial sweeteners, and other additives that may alter the composition of gut microbes. Shifts in the microbiome can influence immune signaling, blood-brain barrier integrity, and the production of neurotransmitters. If ultra-processed diets disrupt this ecosystem, they could indirectly accelerate cognitive decline. Metabolic effects are also plausible: ultra-processed patterns are strongly associated with obesity, type 2 diabetes, and hypertension, all established dementia risk factors. However, the HRS and Framingham analyses did not include detailed biomarker panels, so these mechanisms remain speculative rather than proven.
Limits of the current data
Despite the consistency of the associations, several limits constrain how far the findings can be pushed. Every study in the evidence base is observational. People who eat the most ultra-processed food differ from those who eat the least in ways that statistical adjustment can reduce but not eliminate. Income, food access, occupational demands, stress, sleep quality, and physical activity all cluster with dietary quality, and residual confounding is a real concern. It is possible that some of the measured dementia risk reflects these co-traveling disadvantages rather than ultra-processed food itself.
Reverse causation is another possibility. Subtle cognitive changes often precede a formal dementia diagnosis by years. Individuals in the earliest stages of decline may rely more heavily on ready-to-eat packaged foods because they are easier to prepare, which could inflate the apparent risk associated with ultra-processed diets. The HRS analyses attempted to reduce this bias by excluding participants with baseline impairment and by adjusting for health status, but no observational design can rule it out entirely.
Finally, there are no large randomized trials in which participants are assigned to swap ultra-processed calories for whole foods and then followed for dementia incidence over many years. Such trials are expensive and logistically challenging, given the long time horizon of neurodegenerative disease. Until they exist, the evidence will rest on converging observational lines rather than definitive proof of causation.
Practical implications for older adults and caregivers
Even with those caveats, the data carry practical implications for families trying to lower dementia risk in the absence of curative drugs. The strongest signals in the HRS work point toward specific categories-processed meats, sugary drinks, and ultra-processed animal products-as priorities for reduction. Replacing a daily soda with water or unsweetened tea, choosing fresh or frozen poultry and fish instead of hot dogs and bacon, and favoring plain yogurt over flavored varieties with long ingredient lists are all concrete steps that move intake away from the highest-risk patterns identified in the research.
For many older adults, especially those on fixed incomes or living in areas with limited grocery options, ultra-processed products are cheap, shelf-stable, and heavily marketed. That reality means individual behavior change will be easier if it is supported by broader policy shifts, such as clearer front-of-package labeling, incentives for retailers to stock minimally processed staples, and community programs that make fresh produce and whole grains more affordable. Clinicians and caregivers can also play a role by discussing diet in the same conversation as blood pressure, blood sugar, and physical activity when talking about brain health.
What to watch next
Researchers are now pushing for more detailed mechanistic studies that pair dietary data with brain imaging, blood-based biomarkers, and microbiome analyses to clarify how ultra-processed patterns exert their effects. Shorter-term randomized feeding trials, focused on intermediate outcomes such as inflammatory markers or cognitive test scores rather than dementia diagnoses, may also be feasible and could strengthen the case for causality. As that work unfolds, the current evidence already points toward a cautious but actionable message: while no single food choice determines whether someone will develop dementia, consistently relying on ultra-processed products appears to tilt the odds in the wrong direction, and shifting even part of the diet toward simpler, less processed foods is a step most people can take now.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
