Different Combinations of Mobility Metrics Derived From a Wearable Sensor Are Associated With Distinct Health Outcomes in Older Adults

Abstract

Background: Gait speed is a robust nonspecific predictor of health outcomes. We examined if combinations of gait speed and other mobility metrics are associated with specific health outcomes.

Methods: A sensor (triaxial accelerometer and gyroscope) placed on the lower back, measured mobility in the homes of 1,249 older adults (77% female; 80.0, SD = 7.72 years). Twelve gait scores were extracted from five performances, including (a) walking, (b) transition from sit to stand, (c) transition from stand to sit, (d) turning, and (e) standing posture. Using separate Cox proportional hazards models, we examined which metrics were associated with time to mortality, incident activities of daily living disability, mobility disability, mild cognitive impairment, and Alzheimer's disease dementia. We used a single integrated analytic framework to determine which gait scores survived to predict each outcome.

Results: During 3.6 years of follow-up, 10 of the 12 gait scores predicted one or more of the five health outcomes. In further analyses, different combinations of 2-3 gait scores survived backward elimination and were associated with the five outcomes. Sway was one of the three scores that predicted activities of daily living disability but was not included in the final models for other outcomes. Gait speed was included along with other metrics in the final models predicting mortality and activities of daily living disability but not for other outcomes.

Conclusions: When analyzing multiple mobility metrics together, different combinations of mobility metrics are related to specific adverse health outcomes. Digital technology enhances our understanding of impaired mobility and may provide mobility biomarkers that predict distinct health outcomes.

Keywords: Aging; Gait; Health outcomes; Wearable sensor.