Temporal sequencing of sensory impairments and frailty reveals vulnerability-dependent dementia risk.

Although hearing and vision impairments are considered modifiable dementia risk factors, it remains unclear whether their associations reflect sensory-specific pathways or broader ageing-related vulnerability. Using harmonised data from the Health and Retirement Study (n=22,756) and the English Longitudinal Study of Ageing (n=12,921), we applied multi-state Markov models, time-varying Cox models and disaggregated population attributable fractions (PAFs) to test whether associations depend on their temporal sequence with frailty. Associations weakened after frailty adjustment (adjusted hazard ratios, aHRs, 1.17-1.80 [PAFs 3.4-9.2%] vs 1.08-1.56 [PAFs 1.5-6.6%]). Sensory impairments arising before frailty showed no independent association with incident dementia (aHRs 0.88-1.22 [PAFs -1.3 to 1.5%]), whereas those emerging with or after frailty were consistently associated (aHRs 1.31-2.28 [PAFs 3.4-6.4%]). Conventional estimates may over-attribute dementia burden to sensory impairments by neglecting their ageing context. Frailty-informed approaches may improve targeting and interpretation of sensory interventions for dementia risk reduction.