This study investigates how grip strength, cardiorespiratory fitness, and lung function relate to incident dementia in a large prospective cohort from the UK Biobank.
Prospective cohort study. Participants followed from baseline to first dementia diagnosis, death, or end of follow-up (mean ~12 years).
51,517 participants aged 37–73 at baseline. 628 incident dementia cases. Complete physical fitness measurements required.
Grip Strength (mean kg), Cardiorespiratory Fitness (max workload, W), Lung Function (FEV₁, L).
All-cause dementia, Alzheimer’s disease, vascular dementia, other dementia (ICD-10 coded).
Cox proportional hazards (Models 1–3), Elastic Net prediction, causal mediation (regmedint), restricted cubic splines.
2,918 Olink NPX proteins. Multivariable linear regression + Elastic Net selection. GO enrichment for pathway characterization.
All primary analyses adjust for the following covariates.
Multivariable linear regression of each fitness measure against 2,918 Olink proteins. FDR correction (Benjamini–Hochberg).
Elastic Net Cox regression across fitness-associated proteins to identify the most predictive protein signatures for dementia.
Cox models linking selected proteins to dementia risk, with GO enrichment analysis to characterize biological pathways.