photo by iStock/Ponomariova_Maria

Research

Study Finds How People Approach Test Problems—Not Just How Many Answers They Get Right—Can Help Predict Future Dementia Risk

New research helps detect dementia risk earlier, when there is the greatest opportunity to intervene and preserve independence.

Scores from neuropsychological assessments (in-depth, standardized evaluations of how a person’s brain functions in various cognitive areas) are widely used to identify underlying cognitive abilities. However, a single score representing a specific cognitive domain often obscures the diverse strategies individuals may employ to complete the test. Additionally, reliance solely on scores can hide subtle cognitive changes, which are especially critical for early detection of cognitive disorders.

A new study from researchers at Boston University Chobanian & Avedisian School of Medicine and the VA Boston Healthcare System has found how people complete cognitive tests — including the small mistakes they make and the strategies they use — can predict who will later develop dementia, even years before diagnosis.

In our study, subtle cognitive features predicted dementia more than five years before diagnosis.

Brandon Frank, PhD

“Many research studies focus on total test scores (for example, how many words someone remembers on a memory test). We showed that looking closely at patterns of errors and response styles provides deeper information. In our study, subtle cognitive features predicted dementia more than five years before diagnosis,” explains corresponding author Brandon Frank, PhD, assistant professor of neurology at the school and clinical neuropsychologist at the VA.

The researchers analyzed data from 2,363 adults over the age of 60 (none who had dementia prior to completing their testing) who were part of the Framingham Heart Study Brain Aging Program. The program, which has collected detailed cognitive data since 2005, not only records final test scores, but carefully and extensively documents how people perform on cognitive tests — including specific types of errors, problem-solving strategies, and subtle response patterns. The researchers then used advanced statistical modeling and machine learning to organize hundreds of small test details into meaningful brain-related patterns — and demonstrated that these patterns improve early detection of dementia risk.

According to the researchers, subtle changes in thinking may be detectable earlier than is previously believed. “Clinically, this could improve early detection of dementia and help doctors identify who may benefit from closer monitoring, support earlier intervention when treatments are most effective, and reduce misdiagnosis caused by relying on single test scores,” adds co-first author Ashita Gurnani, PhD, clinical neuropsychologist, assistant professor of neurology, and co-lead for the clinical core of the Framingham Heart Study Brain Aging Program.

These findings appear online in the Journal of the International Neuropsychological Society.