Results
Participant Characteristics
We compared the characteristics of those who developed dementia (n= 289) with the characteristics of those who did not (n=3,643). Those who developed dementia were significantly older on average, were less wealthy, had poorer cognition and mobility and significantly slower walking speed, and were more likely to have had a stroke or depressive symptoms than those who did not (Table 1).
Baseline Walking Speed as a Predictor of Dementia
Walking speed at Wave 1 was a predictor of development of dementia, with those with faster walker speeds being less likely to develop dementia (HR=0.36, 95% CI=0.22–0.60) during follow-up (Table 2). This association was robust to adjustment for covariates. Cognitive function at baseline was also an independent predictor of development of dementia (HR=0.42, 95% CI=0.33–0.52), with those with better cognitive function being less likely to develop dementia.
Changes in Walking Speed and Cognition as Predictors of Dementia
Walking speed decreased on average from 0.86 to 0.85 m/s between Waves 1 and 2. Change in walking speed was a significant predictor of dementia (HR=1.23, 95% CI=1.03–1.47), with those who had a greater decrease in walking speed from Wave 1 to Wave 2 having a greater risk of developing dementia independent of covariates and walking speed in Wave 1 (Table 3). Change in cognition was also a predictor, with participants with a greater significant decline in cognitive function between Waves 1 and 2 being at greater risk of developing dementia (HR=1.78, 95% CI=1.53–2.06), but the interaction between walking speed and cognitive function was not a significant predictor of dementia (HR=1.01, 95% CI=0.88–1.17).
Sensitivity Analyses
Three sensitivity analyses were conducted. First, we tested whether walking speed at Wave 2 (2004–05) predicted development of dementia. Participants with faster walking speeds were less likely to develop dementia (HR=0.25) between Waves 3 and 7 (2006–2015) (Supplementary Table S1).
Second, we only considered new events from Wave 4 to 7, omitting any that occurred within 2 years of Wave 2. This reduced the number of cases from 289 to 225 (Supplementary Table S2), but walking speed at Wave 1 remained a predictor of dementia (HR=0.33). Similarly, those with poorer cognition at Wave 1 (HR=0.29) and those who had a greater decline in cognitive function between Waves 1 and 2 (HR=1.69) remained more likely to develop dementia. The effect size for changes in walking speed was similar to that in the full analysis, but the association was no longer significant because of the smaller number of cases.
For the final sensitivity analysis, we excluded participants who were diagnosed using the IQCODE and limited the analysis to those with physician diagnoses. This reduced the number of cases from 289 to 240 (Supplementary Table S3). Walking speed remained a predictor of dementia (HR=0.36). The findings for cognition also remained for baseline cognition (HR=0.26) and change in cognitive function (HR=1.83), but similar to the other sensitivity analysis, change in walking speed no longer predicted development of dementia.
J Am Geriatr Soc. 2018;66(9):1670-1675. © 2018 Blackwell Publishing
Comments