Article Text
Abstract
Background Current policies do not support screening for future dementia among apparently healthy individuals. This is due to the lack of certainty of the clinical outcome and effective interventions. Numerous dementia risk models have been developed, currently, all within the research setting only. Studies have shown impairment in multiple cognitive domains several years before a clinical diagnosis of dementia.
Methods We examined the utility of an extensive battery assessing a range of specific cognitive abilities as well as a composite global score, to predict dementia. Dementia outcomes were ascertained using electronic health record linkage in 8581 individuals (aged 48–92 years) taking part in the EPIC-Norfolk study. Participants were followed for 15 years (2004–2019). Risk of dementia was estimated using Cox proportional hazard models adjusting for sociodemographic, lifestyle and health variables, evaluating discriminative accuracy of the models by analysing receiver operating characteristic (ROC) curves.
Results Poor cognition was predictive of incident dementia, even after adjustment for co-variates. Those with a poor performance score for global cognition (bottom 10%) were almost four times as likely to get a dementia diagnosis than those who performed well (HR=3.51 (95%CI 2.61, 4.71) P<0.001). Associations were observed for specific as well as global cognitive abilities. The test for episodic (verbal) memory outperformed other tests and was comparable to global cognition scores. Poor cognition in four or more tests was associated with 10-fold increased risk of developing dementia compared to those not performing poorly in any test (HR=10.82 (95% CI 6.85, 17.10) P<0.001)). Cognitive measures strengthen prediction models of dementia (AUC = 0.85 (95%CI 0.82, 0.87) P<0.001).
Discussion This study provides further insight on poor cognition predicting future dementia. This association was observed for global cognition and specific abilities, particularly for verbal episodic memory. Impairment occurs in multiple domains several years prior to a clinical diagnosis, and the more pervasive and greater the variability, the higher the risk of dementia. Deficits across multiple domains predict over and above individual test scores.