Composite risk scores for predicting dementia

Stephan, Blossom C. M.; Tang, Eugene; Muniz-Terrera, Graciela

Abstract

Purpose of reviewA key priority in dementia research is the development of tools to identify individuals at high risk of dementia. This is important to prevent or delay dementia onset and as we move towards personalized medicine.Recent findingsNumerous models (n>50) for predicting dementia have been developed. These vary in the number (0 to 20+) and type (e.g. demographics, health, neuropsychological, and genetic) of predictor variables used for risk calculation, follow-up length (1-20 years) and age at screening (mid vs laterlife). Evaluation of the models shows that most have moderate-to-poor predictive accuracy. Few have been externally validated, raising questions about their generalizability outside the cohorts from which they were developed. The results highlight that if additional models are proposed the field will be overwhelmed with many competing risk models, making it difficult to reach consensus on which is best.SummaryNumerous models for predicting dementia have been proposed but are limited by a lack of external validation and evaluation of economic impact. Innovative methods and data designs may be needed to improve derivation of dementia risk scores. Having a method for predicting dementia risk could transform medical research and allow for earlier testing of intervention strategies.

Más información

Título según WOS: ID WOS:000369605700012 Not found in local WOS DB
Título de la Revista: CURRENT OPINION IN PSYCHIATRY
Volumen: 29
Número: 2
Editorial: LIPPINCOTT WILLIAMS & WILKINS
Fecha de publicación: 2016
Página de inicio: 174
Página final: 180
DOI:

10.1097/YCO.0000000000000235

Notas: ISI