People whose hemoglobin was unusually low or unusually high on a routine health check were more likely to be diagnosed with dementia over the next 15 years, according to a study of more than 316,000 adults in South Korea1. The work was presented in October at the 39th ECNP Congress in Munich and was published earlier in the year in the Journal of Alzheimer’s Disease1. It shows a link between a common blood measure and later dementia, not a cause, and a study author said so directly1.
What happened
The researchers used data from 316,542 people aged 40 and over who had a national health screening in 2004 to 20051. They then tracked who received a dementia diagnosis over the following 15 years, until 20191. The aim was to test whether hemoglobin levels were linked to dementia risk among Korean adults, and whether any link was stronger in men or in women1.
Participants were sorted into five groups, from the lowest hemoglobin levels to the highest1. People at both ends were the most likely to be diagnosed with dementia later on1.
The study also looked at two common kinds of dementia. Vascular dementia, which is tied to problems with blood flow in the brain, was more common at both ends: an 18% higher risk in the highest group and a 19% higher risk in the lowest group1. Low hemoglobin was also linked to a higher risk of Alzheimer’s disease1.
The evidence so far
This is an observational study. The researchers did not change anyone’s hemoglobin; they looked at records and saw which patterns went together over time. Its strengths are its size and length: hundreds of thousands of people followed for 15 years1.
Hyun Soo Kim, a study author and assistant professor at Dong-A University College of Medicine in Busan, South Korea, explained what the results do and do not mean in a press release1:
“The findings do not show that hemoglobin itself causes dementia, but suggest that abnormal hemoglobin levels may be a visible marker of wider physical vulnerability relevant to brain health.”
Experts quoted in coverage said that hemoglobin levels may reflect other health problems, such as iron deficiency, chronic kidney disease and heart and blood vessel issues1.
What it does not show yet
The size of the risk needs care. Dung Trinh, an internist with MemorialCare Medical Group who was not involved in the study, told Medical News Today that the 16% figure is a relative increase compared with the study’s reference group, not a 16% chance of getting dementia1. He said the rise is “relatively modest for any one individual” but “could have meaningful implications at the population level”1.
He also said the study cannot answer the question many readers will ask: “we cannot conclude from this study that raising hemoglobin levels will prevent dementia”1.
Other open questions remain. The study examined whether the link differed between men and women, but the coverage reviewed here does not report that result1. The people studied were all Korean adults, so it is not known how closely the findings apply elsewhere1. And because hemoglobin was measured at a screening in 2004 to 2005, the coverage does not describe how levels changed over the following years1.
Who is behind it
The study comes from university researchers in South Korea, including Kim at Dong-A University College of Medicine1. It used data from a national health screening program1. The coverage reviewed for this article does not describe who funded the work or list any company ties. Trinh, the outside doctor quoted, is also chief medical officer of Healthy Brain Clinic in Irvine, California1.
How it fits the field
The idea that low hemoglobin, known as anemia, could be tied to dementia is not new. In a study from the Kungsholmen Project in Stockholm, researchers followed 1,435 people aged 75 to 95 without dementia for 3 years2. Those who met World Health Organization criteria for anemia, with hemoglobin below 130 g/L for men and 120 g/L for women, had a higher risk of developing dementia (hazard ratio 1.6, roughly a 60% higher risk)2. Those authors noted that earlier long-term studies had given contradictory results, and suggested that anemia might raise dementia risk through long-term low oxygen supply to the brain2.
A larger Swedish study published in JAMA Network Open in April 2026, led by Martina Valletta of the Aging Research Center at Karolinska Institutet, followed 2,282 dementia-free adults aged 60 or older for an average of 9.3 years3. At the start, 199 of them (8.7%) had anemia3. After adjusting for kidney disease, heart disease, cancer, inflammation and other factors, anemia was tied to a 66% higher risk of dementia3. People with anemia also had higher blood levels of proteins linked to Alzheimer’s disease and nerve damage, and those with both anemia and a high level of one such protein had about 3.6 times the risk of people with neither3. That study’s own limits included a mostly White population and blood markers measured only once3.
Both Swedish studies looked at anemia, meaning low hemoglobin32. The Korean study adds a signal at the high end as well, especially for vascular dementia1.
In an editorial alongside the Swedish JAMA study, Frank Wolters of Erasmus MC in Rotterdam wrote that “in order to move findings from observation to effective intervention, mechanistic insight is mandatory”3. He also called for testing the findings in geographically diverse settings, with attention to differences between men and women3.
What to watch next
The next steps the researchers and outside experts point to are studies that explain why the link exists. Wolters suggested studies using repeated hemoglobin measurements, analyses designed to mimic trials, and small physiological trials3. Kim said that identifying “simple, accessible markers of future risk” could help researchers and clinicians understand how whole-body health relates to brain health over time1.
For now, hemoglobin is something doctors already measure with a simple blood test, as Trinh noted1. Whether that number can help predict dementia for an individual, or whether treating the conditions behind abnormal levels would change anyone’s risk, has not been tested.
