Sensory impairment and cognitive-state transitions in middle-aged and older adults: a multistate Markov analysis of the health and retirement study
Abstract
Background Sensory impairment has been associated with cognitive decline and dementia. However, endpoint-based longitudinal analyses often cannot determine whether sensory impairment is related to early cognitive deterioration, reduced cognitive recovery, dementia progression, or mortality-related study exit. Methods We analyzed 2004–2022 Health and Retirement Study data using a continuous-time multistate Markov model. Four states were defined: normal cognition, cognitive impairment no dementia (CIND), dementia, and death. The analytic cohort included 18,292 participants, and 16,083 were included in the primary complete-case adjusted model. Baseline self-reported sensory status was classified as no impairment, visual impairment only, hearing impairment only, or dual sensory impairment. Primary models adjusted for age, sex, education, wealth, depressive symptoms, and chronic disease count; sensitivity analyses included time-varying and lagged exposures, inverse-probability weighting, and expanded confounding adjustment. Results Compared with no sensory impairment, visual impairment only, hearing impairment only, and dual sensory impairment were associated with faster transitions from normal cognition to CIND. The corresponding HRs were 1.18 (95% CI, 1.10–1.27), 1.16 (95% CI, 1.07–1.25), and 1.37 (95% CI, 1.26–1.50). Visual impairment only and dual sensory impairment were also associated with slower transitions from CIND to normal cognition, with HRs of 0.79 (95% CI, 0.72–0.86) and 0.84 (95% CI, 0.75–0.94), respectively. Associations with CIND-to-dementia progression and mortality-related transitions were less consistent. Time-varying, lagged, weighted, and expanded-confounding analyses supported the same early-transition pattern. Conclusion Sensory impairment, particularly dual sensory impairment, was most consistently associated with early cognitive deterioration and reduced cognitive recovery. It was not consistently associated with higher dementia progression or mortality transition rates. Death-inclusive multistate modeling may help identify where potentially modifiable sensory impairment is related to cognitive ageing trajectories.