Age-Dependent Nonlinear Association Between Gait Impairment and Incident Dementia

Background and Objectives Gait impairment is associated with incident dementia, yet whether this association varies with age or follows a linear dose-response relationship remains unclear. The aim of this study was to examine the nonlinear dose-response relationship between Performance-Oriented Mobility Assessment (POMA) gait performance and incident dementia and to test whether this association is modified by age. Methods This prospective cohort study used data from the Korean Longitudinal Study on Cognitive Aging and Dementia (baseline enrollment, 2010–2012; follow-up through 2020; median follow-up, 6.2 years). The primary outcome was incident dementia diagnosed by Diagnostic and Statistical Manual of Mental Disorders IV (DSM-IV) criteria. Gait and balance were assessed using the POMA. Cause-specific Cox proportional hazards regression with death treated as a competing event was used. Restricted cubic splines (RCS) modeled the nonlinear dose-response relationship between gait performance and incident dementia, and RCS-by-age interaction terms were used to test age modification and estimate age-specific hazard ratios and 95% CIs. Results Of 6,818 eligible participants, 4,822 met inclusion criteria (mean (SD) age, 69.6 (6.3) years; 2,719 (56.4%) female; median follow-up 6.2 years [IQR, 3.8–8.1]). During follow-up, 251 developed dementia and 458 died. The gait-dementia association was significantly nonlinear ( p = 0.002), with hazard ratios rising steeply as scores departed from the ceiling and peaking near 9 to 10, below which estimates were imprecise. The association was strongly and inversely modified by age (interaction p < 0.001), indicating that the predictive value of gait deficits for dementia progressively diminishes with advancing age: the hazard ratio for a gait score of 11 vs 12 was 3.13 (95% CI 1.82–5.37) at age 65 years, 1.36 (95% CI 0.98–1.87) at age 75 years, and 0.59 (95% CI 0.37–0.94) at age 85 years. The estimated crossing age was 78.6 years. The age-dependent attenuation was specific to gait and not observed for balance in models including both components. Discussion Gait impairment predicted incident dementia in a nonlinear, age-dependent manner, with the association strongest in adults aged 65–75 years and not evident beyond approximately age 80 years. These findings suggest gait assessment may be most informative for dementia screening in younger-old adults. Limitations include an ethnically homogeneous cohort.