Community Walking Speed as a New Predictor of Disability Incidence in Older Adults: A Prospective Cohort Study

Walking speed is widely recognized as an informative indicator of physical capability in later life and is frequently used to evaluate mobility and functional health in older adults. Although the measurement of community walking speed (CWS) has been made possible using accelerometers, it is unclear whether discrepancies between laboratory walking speed (LWS) and CWS make a difference in predicting the incidence of future disabilities in older adults. Using data from the National Center for Geriatrics and Gerontology-Study of Geriatric Syndromes (NCGG-SGS), we examined whether LWS and CWS each independently predict incident disability. The discrepancy was defined as the difference between LWS and CWS, and we further analyzed the characteristics of older adults who exhibited such discrepancies.

Participants comprised 1,631 older adults (mean age = 70 years; 62.7% women). LWS was measured using a WalkWay device that measures the distribution of foot pressure during walking. CWS was derived using a model that estimated gait speed based on composite acceleration calculated from the mean triaxial acceleration within a single gait cycle. Participants were instructed to wear a triaxial accelerometer for at least 14 days. The incidence of disability was prospectively determined for 5 years.

The Cox proportional hazards regression models revealed a statistically significant association between the z scores of LWS (hazard ratio [HR], 0.54; 95% confidence interval [CI], 0.41-0.72) and CWS (HR, 0.59; 95% CI, 0.41-0.83) and disability incidence. There were no statistically significant interactions between LWS and CWS for disability (HR, 1.00; 95% CI, 0.76-1.31). The area under the receiver operating characteristics curve of the LWS and CWS was 0.752 and 0.709, respectively. Regarding discrepancies between the fast LWS group and the fast CWS group, residual analysis showed that the fast LWS group had a higher proportion of women than the fast CWS group.

LWS and CWS were found to be independently associated with disability. Further research must determine how CWS should be interpreted because participants who showed discrepancies between CWS and LWS did not exhibit a significant association with incident disability. Healthcare providers could use either CWS and LWS as a significant indicator in healthcare practice for older populations.