Gait Speed and Decline in Gait Speed as Predictors of Incident Dementia

{Background: Slow gait is common in dementia, but it remains unknown whether the slowing happens many years prior to dementia onset. We therefore examined the relationship between slow baseline gait speed (GS), change in GS, and the hazard of incident dementia in a community dwelling of elderly people. Methods: A total of 3,663 participants dementia-free at baseline (mean age, 73.5 years) were followed up for 9 years from a prospective cohort (Three-City study, France) for incident dementia (all-cause, Alzheimer's disease, vascular dementia, and other causes). GS over 6 m was assessed 4 times over the follow-up using two photoelectric cells. We used a multistate model to estimate the hazard ratio (HR) of dementia for baseline GS and tested a washout period of 4 to 7 years. The role of GS change between 65 and 85 years was examined using linear mixed models and joint models for survival and longitudinal data. Results: A total of 296 participants developed dementia during the follow-up. In age/sex-adjusted models, 1-SD (0.204 m/s) lower GS was associated with an increased hazard of dementia (HR = 1.59, 95% confidence interval [CI] = 1.39, 1.81, p < .001), with associations evident when gait assessments were taken from 4 years (HR = 1.46; CI = 1.26, 1.68) and 7 years (HR= 1.30; CI = 1.00, 1.70) prior to dementia onset. Independently of baseline GS, those with a steeper decline had a higher hazard of dementia (HR per 1 SD [0.007 m/s/year] decrease = 3.39 [1.37-8.43]