Dementia Risk From Midlife Onward Predicted With New Tool

— Novel 14-year risk score draws largely on modifiable risk factors

A photo of a mature man sitting on the edge of his bed holding tow different shoes.

A novel 14-year risk score helped identify people from age 50 onward at risk for all-cause dementia, a large U.K. study showed.

The tool, called the U.K. Biobank Dementia Risk Score (UKBDRS), was developed and validated in two U.K. cohorts, reported Raihaan Patel, PhD, of the University of Oxford in England, and co-authors.

The area under the curve (AUC) in the U.K. Biobank test cohort was 0.80 (95% CI 0.78-0.82), the researchers wrote in BMJ Mental Health. The AUC in the Whitehall II validation cohort was 0.77 (95% CI 0.72-0.81).

The score included 11 predictive variables: age, education, parental history of dementia, material deprivation, history of diabetes, stroke, depression, hypertension, high cholesterol, household occupancy (living alone), and sex.

"Importantly, our score highlights the importance of modifiable risk factors," Patel told MedPage Today. "While age and APOE were the strongest predictors, modifiable factors such as diabetes, depression, and high blood pressure also played a key role."

The UKBDRS outperformed three other widely-used risk scores -- the Cardiovascular Risk Factors, Aging, and Dementia (CAIDE) score, the Australian National University Alzheimer's Disease Risk Index (ANU-ADRI), and the Dementia Risk Score (DRS).

An analysis published earlier this year suggested commonly used scores, including CAIDE and ANU-ADRI, had high error rates for predicting 10-year dementia risk.

"Previous risk scores have not been very good, and so a better score is most welcome," Gill Livingston, MD, of University College London in England, who led the Lancet Commission's report on modifiable dementia risk factors, told MedPage Today.

"Dementia risk scores are important as they indicate who is at highest risk of dementia, which is potentially reversible," Livingston observed.

"This gives individuals information and therefore power to change the course of their life," she pointed out. "It helps doctors answer the question about what to do to prevent dementia and intervene. Finally, it helps researchers find out who they should target for interventions to prevent dementia -- if someone is not at risk, then the intervention will not help."

To develop the UKBDRS, Patel and colleagues evaluated data from 220,762 participants in the U.K. Biobank study (mean age 60) and 2,934 participants in the Whitehall II study (median age 57) in their analyses. The U.K. Biobank cohort was split into a training set (80% of sample) and test set (20%). The Whitehall II cohort was used for external validation.

All participants were ages 50 to 73 years. A subset of people in each group had APOE data.

In the U.K. Biobank group, all-cause dementia status was determined by self-reports, primary or secondary diagnoses in medical or death records, or prescriptions for dementia medicines. In Whitehall II, dementia was determined by self-reported data and inpatient hospital records.

The researchers compiled a list of 28 risk and protective factors associated with dementia, including modifiable factors identified by the Lancet Commission. They used LASSO regression to select the strongest predictors of incident dementia from the 28 candidates, then developed the risk score using competing risk regression. Based on available follow-up times, the researchers used a 14-year time horizon to develop the AUC in the U.K. Biobank and a 17-year horizon in Whitehall II.

Overall, 3,813 people (1.7%) in the U.K. Biobank cohort and 93 people (3.2%) in Whitehall II developed dementia. In the subset of people with genotype data, the UKBDRS-APOE achieved an AUC of 0.83 among U.K. Biobank participants and 0.79 in the Whitehall II group.

The study came with some caveats. Dementia was defined differently in each cohort, and the Whitehall II study was made up mostly of men.

Moreover, dementia risk scores may not work well outside the population in which they were initially developed, Patel noted.

"There are still improvements required before this score is suitable for clinical practice," he said. "While the score performed well across two U.K. cohorts, further evaluation across more diverse groups of people both within and beyond the U.K. is required."

  • Judy George covers neurology and neuroscience news for MedPage Today, writing about brain aging, Alzheimer’s, dementia, MS, rare diseases, epilepsy, autism, headache, stroke, Parkinson’s, ALS, concussion, CTE, sleep, pain, and more. Follow


This research was supported by a U.K. Alzheimer's Society Research Fellowship and a grant awarded by the Academy of Medical Sciences, the Wellcome Trust, the Government Department of Business, Energy and Industrial Strategy, the British Heart Foundation, and Diabetes U.K.

The authors reported no competing interests.

Livingston has worked with some of the authors. She has no relationships with industry.

Primary Source

BMJ Mental Health

Source Reference: Anatürk M, et al "Development and validation of a dementia risk score in the UK Biobank and Whitehall II cohorts" BMJ Ment Health 2023; DOI: 10.1136/bmjment-2023-300719.