Associations of Smartwatch-Derived Step Count and Active Minutes with Patient-Reported Outcomes Collected from Smartphone: electronic Framingham Heart Study (Preprint)

BACKGROUND Higher physical activity levels are associated with better symptoms, functional health, and mobility. However, few studies have examined these relationships using real world data collected by wearable devices in community samples of older adults. OBJECTIVE We sought to investigate the cross-sectional associations between smartwatch-derived physical activity metrics with smartphone-collected patient-reported outcomes and mobility measures. METHODS Among the electronic Framingham Heart Study (eFHS) participants, smartphone survey responses were used to compute life-space assessment (LSA) scores to measure mobility outside the home and to compile T scores for seven patient-reported outcomes measurement information system (PROMIS) domains, including cognitive ability, cognitive function, anxiety, depression, fatigue, sleep, and physical function. Average daily steps and active minutes (steps per minute ≥40) were derived from ≥30 smartwatch days with 8-18 hours of watch wear. Linear regression was used to evaluate the associations of average daily steps and active minutes (independent variables), separately with PROMIS T-scores and LSA scores (dependent variables). Models were adjusted for covariates, including age, sex, body mass index, eFHS cohort membership, and average daily smartwatch wear time. We used the false discovery rate (FDR) < 0.05 to account for multiple testing. We categorized participants by tertile with the lowest group as reference, and calculated adjusted differences in dependent variables. RESULTS Among 392 eFHS participants in the study sample (mean age 73 years, 59% women, 15% non-White), median average daily steps and active minutes were 4,811 and 34. Median PROMIS domain scores ranged from 41-56, and the median LSA was 96 (25th, 75th percentiles, 81, 110). Each 1,000 higher average daily steps was associated with 0.68 lower fatigue T score, 0.97 higher physical function T score, and 1.25 higher LSA scores (all false discovery rate (FDR) <0.05). Each 5-minute higher average daily active minutes was associated with 0.25 lower fatigue, 0.48 higher physical function, 0.23 higher cognitive ability, 0.21 higher cognitive function T scores, and 0.61 higher LSA scores (all FDR <0.05). Participants with medium and high average daily step counts had lower fatigue (adjusted mean difference (95% CI), -3.14 (-5.28, -1.01) and -4.50 (-6.98, -2.02)), higher physical function T scores (4.64 (2.96, 6.33) and 5.74 (3.79, 7.69)) and LSA scores (6.90 (1.50, 12.30) and 8.05 (1.81, 14.29)), compared with the lowest tertile and additional associations were observed with cognitive function, cognitive ability, and anxiety. Associations across active minute tertiles remained consistent with the primary analysis. CONCLUSIONS Our findings add to the growing evidence in older adults supporting the association between higher physical activity levels and improved patient-reported outcomes and mobility. Leveraging wearable and smartwatch technologies provides opportunities to collect real world data on outcomes important to the health of older adults more often and more conveniently outside research and clinical settings.