In a conversation about daily life, an older person might be asked whether they still visit friends, take part in community groups or manage their own money. Those ordinary details are at the heart of a new dementia risk score developed for countries where the need for useful research tools is greatest.
Around 60% of the world’s more than 57 million people living with dementia are in low- and middle-income countries, across regions including Latin America, the Caribbean, Asia and Africa. Yet many tools used to predict dementia risk were built with data from wealthier countries, and have not worked as well when applied elsewhere. A team led by Northumbria University in Newcastle upon Tyne set out to help close that gap.
The researchers built and tested their five-factor score using information from more than 11,000 older adults at seven sites in China, Cuba, the Dominican Republic, Puerto Rico, Venezuela, Mexico and Peru. The factors are age, social engagement, physical activity, high blood pressure and difficulty managing money. All can be assessed through a simple conversation with an older person and, when needed, a family member or caregiver. The method does not require blood tests, brain scans or special equipment.
Age is the strongest single predictor in the score. Risk rises noticeably from 70 and more steeply from 80 onward. Social engagement includes whether someone is married or lives with others, how often they see friends or family, and whether they join community or religious groups. Physical activity includes everyday movement, such as farming, walking to market or doing household tasks. The team also included uncontrolled high blood pressure, which can harm blood vessels supplying the brain and is often undiagnosed in lower-income settings. A caregiver’s account of a change in money management can offer another early sign of cognitive change.
The five factors add up to a score from zero to nine. In the study, a score of zero corresponded to about a 1% chance of developing dementia within three years; a score of nine corresponded to nearly 19%. The model achieved an accuracy score of 0.75, above the 0.70 level generally considered useful in clinical settings. The study was published in the American Journal of Epidemiology.
The researchers say the score is intended mainly to support research, not to diagnose dementia. It could give studies and clinical trials a practical, low-cost way to identify and recruit people at higher risk in places where existing prediction tools are scarce. That matters because better representation in research can help future prevention efforts reflect the lives and needs of local communities.
Dr. Eduwin Pakpahan, from Northumbria University’s Applied Statistics and Data Science Research Group, said the model offers a practical way to find people who should be the focus of future studies and trials. Blossom Stephan, chair in dementia at Curtin University in Perth and a study guarantor, described it as a foundation for more inclusive prevention research, while stressing that it needs further testing across diverse regions. The next step is broader validation: a simple score built for these settings could become more useful as researchers test how well it works in more communities.