Alzheimer’s Disease and Mobility Multimodal Prediction of Gait Decline, Fall Risk, and Loss of Functional Independence in Mild Cognitive Impairment and Early Alzheimer’s Disease

Abstract

Mobility deterioration is an important component of disease progression in mild cognitive impairment and early Alzheimer’s disease, yet gait decline, falls, and functional dependence are frequently evaluated as separate outcomes. This study examined whether multimodal information improved prediction of cognitive-motor deterioration and adverse mobility outcomes. A multicenter, prospective longitudinal quantitative cohort design was used to follow 420 community-dwelling adults, including 210 participants with mild cognitive impairment and 210 with early Alzheimer’s disease, for 24 months. Comprehensive assessments were conducted at baseline and 6, 12, 18, and 24 months, while falls were monitored monthly. Clinical characteristics, cognition, instrumented gait, dual-task performance, balance, functional status, medication exposure, free-living mobility, magnetic-resonance imaging, and blood biomarkers were measured. At 24 months, 351 participants remained under observation, representing 83.6% retention. Mean annual usual-pace gait-speed decline was 0.035 m/s in mild cognitive impairment and 0.061 m/s in early Alzheimer’s disease. Clinically meaningful gait decline of at least 0.10 m/s occurred in 37.7% and 64.9% of the respective groups, with an adjusted odds ratio of 2.64 for early Alzheimer’s disease. At least one fall was experienced by 43.8% of participants with mild cognitive impairment and 66.2% with early Alzheimer’s disease. Fall rates were 44.7 and 87.4 events per 100 person-years, respectively, while injurious falls occurred in 16.7% and 31.9%. Sustained functional dependence developed in 21.9% of the mild cognitive impairment group and 51.4% of the early Alzheimer’s disease group. Previous falls, gait variability, dual-task cost, turning instability, orthostatic hypotension, medication burden, executive decline, reduced daily activity, and abnormal phosphorylated tau independently predicted adverse outcomes. The multimodal model explained 63% of gait-decline variance, achieved concordance statistics of .83 for first falls and .87 for functional dependence, and outperformed clinical-only, cognition-only, and gait-only models. Multimodal assessment provided a more complete quantitative representation of mobility vulnerability than any single measurement domain and identified distinct but interconnected pathways leading to gait deterioration, recurrent falling, injury, and loss of independence.