A longitudinal study of more than 5,000 older adults revealed that individuals classified as overweight had a 14% lower risk of developing dementia compared to those with normal weight, while obese participants had a 19% lower risk. However, further analysis indicated that individuals who lost weight from midlife to late life had an increased risk of dementia. Those whose weight remained stable had the lowest risk. The paper was published in Neurology.
Late-life dementia refers to dementia that develops in older adulthood, typically after the age of 65. It is most often caused by Alzheimer’s disease, but vascular dementia, Lewy body dementia, and mixed forms are also common in later life. Risk increases sharply with age, with prevalence doubling approximately every five years after age 65.
Symptoms typically begin with mild memory loss and confusion and gradually progress to impaired judgment, disorientation, language difficulties, and loss of independence. Age-related changes in blood vessels, the accumulation of abnormal proteins in the brain, and other chronic conditions contribute to its onset. Midlife risk factors such as hypertension, diabetes, and obesity are important predictors of late-life dementia. Lifestyle factors such as physical activity, cognitive engagement, and social connectedness may delay or reduce risk.
Study author Ethan J. Cannon and his colleagues noted that obesity in midlife is a known risk factor for dementia. However, studies indicate that obesity in late life is associated with a lower risk. The researchers sought to better investigate this paradox and identify the factors that might explain it. They particularly focused on changes in body weight between midlife and late life. The authors hypothesized that individuals who lost weight during this period would be at increased risk of dementia, even more so than normal-weight individuals who maintained a stable weight.
They analyzed data from the Atherosclerosis Risk in Communities (ARIC) study, a longitudinal project conducted at four U.S. sites: suburban Minneapolis, MN; Jackson, MS; Forsyth County, NC; and Washington County, MD. From 1987 to 1989, the study enrolled a total of 15,792 participants. Multiple follow-up visits occurred through 2020. For the present analysis, 5,129 participants who remained dementia-free at Visit 5 (2011–2013) were included. At enrollment, participants were 45–64 years old. At Visit 5, the average age was approximately 75 years.
The researchers used data on participants’ body weight and BMI collected at two key timepoints—midlife (1996–1998) and late life (2011–2013)—and linked it to dementia diagnoses through a combination of in-person neuropsychological testing, structured telephone interviews, hospital discharge records, and death certificates.
They first examined whether BMI in late life alone predicted future dementia risk. Consistent with previous studies, individuals classified as overweight or obese in 2011–2013 were less likely to develop dementia in the years that followed. Specifically, those who were overweight had a 14% lower risk, and those who were obese had a 19% lower risk than normal-weight individuals.
However, when the researchers accounted for weight change from midlife to late life, the pattern shifted. Participants who maintained a normal weight over this 15-year period had the lowest risk of dementia. In contrast, individuals who lost at least 2 kg/m² in BMI during this period were at elevated risk—regardless of their BMI category in late life.
Compared to normal-weight individuals who remained weight stable, the hazard of developing dementia was, 2.22 times higher for those who lost weight but remained normal weight in late life, 1.78 times higher for those who lost weight and were overweight in late life, and 1.80 times higher for those who lost weight and were obese in late life.
Even individuals who gained weight and were classified as obese in late life had a slightly increased dementia risk—about 29% higher than the reference group. However, after adjusting for factors such as smoking, alcohol use, depressive symptoms, frailty, hypertension, and diabetes, this association was weakened.
For example, after full statistical adjustment overweight individuals who had lost weight from midlife to late life had a 53% higher risk of dementia, while normal-weight individuals who had lost weight had a 2.05-fold increased risk.
Meanwhile, overweight and obese individuals whose weight remained stable—or increased—did not differ significantly in dementia risk from the reference group of weight-stable, normal-weight individuals.
“In this study of older adults (mean age 75 years), we found that overweight or obesity in late life was associated with a lower risk of dementia, but that when stratified by weight change history, this result was explained in part by a disproportionate number of low-BMI [body mass index] participants who had experienced weight loss,” the study authors wrote. “Regardless of late-life BMI category, weight loss from midlife to late life was associated with a higher risk of dementia, compared with being normal weight in late life and weight stable from midlife to late life.”
The findings suggest that the so-called obesity paradox may be at least partly explained by unintentional weight loss associated with underlying health decline. The study could not determine whether weight loss was intentional or unintentional. Unintentional weight loss is common in older adults and is often a sign of deteriorating health. In such cases, both weight loss and cognitive deterioration can be the consequence of a decline in physical health.
Additionally, it is important to note that the design of the study does not allow any causal inferences to be derived from the results. That is, higher weight in late life should not be interpreted as protective per se, nor should normal weight be viewed as inherently risky. Rather, trajectories of weight change may serve as an important signal for dementia risk assessment.
The paper, “Association of Body Mass Index in Late Life, and Change from Midlife to Late Life, With Incident Dementia in the ARIC Study Participants,” was authored by Ethan J. Cannon, B. Gwen Windham, Michael Griswold, Priya Palta, David S. Knopman, Sanaz Sedaghat, and Pamela L. Lutsey.
