Dual-task gait from midlife to older age in the population-based Rotterdam Study

Brain function is important for gait, and subtle gait disturbance may identify individuals at high risk of neurodegenerative disease. However, it has been difficult to isolate the neural contribution to gait. We aimed to investigate dual-task gait that places larger demands on neural resources, and compare it to single-task gait. We included 2222 participants of the community-based Rotterdam Study (median age 72 years, 55% women), who completed a single-task and a dual-task (serial 9 s) gait assessment on GAITRite®. First, we used principal component analysis to summarize gait variables into domains. To capture intra-individual changes with dual-task, we projected dual-task measurements onto the single-task principal component space and derived domain scores. We estimated the change between single-task and dual-task measurements by taking the residuals of a Deming regression of dual-task on single-task measurements, and related those to age and sex. Virtually all participants (99.9%) performed worse on dual-task compared to single-task walking, with most profound declines on temporal gait parameters and velocity. Decline in gait upon dual-task was observed from age 45 to 90 years, with somewhat larger absolute decline in younger than in older participants, and similar between women and men. The Deming regression residuals also indicated slightly better dual-task performance at older age than expected based on single-task performance. In this community-based sample, a cognitive challenge during walking caused decline on virtually all observed domains of gait, from mid-life to the oldest old, and most profound for temporal measures of the Rhythm domain and velocity. • A cognitive challenge during dual-task walking caused decline on virtually all observed domains of gait, from midlife to the oldest old, and most profound for temporal measures, including velocity. • The decline upon dual-task gait was somewhat larger in younger than older participants. • We quantified the change between single-task and dual-task performance with Deming regression method, by taking the residuals of dual-task gait regressed on single-task gait.