Ability to Detect Changes and Minimal Important Difference of Real-World Digital Mobility Outcomes in Proximal Femoral Fracture Patients

Abstract Background Older adults’ walking has so far been evaluated using standardised assessments of walking capacity within a clinical setting. By taking the evaluation out of the laboratory into the real world, this study provides first evidence of the ability of Digital Mobility Outcomes (DMOs) to detect changes over time and the Minimal Important Difference (MID) in patients after proximal femoral fracture (PFF). This will guide the implementation of DMOs in research and clinical care. Methods For this multicenter prospective cohort study, 381 community-dwelling older adults were included within one year after sustaining a PFF and assessed at two time points, separated by six months. Walking activity and gait DMOs were measured using a single wearable device worn on the lower back for up to seven days. A global impression of change question and three mobility-related outcome measures (Late-Life Function and Disability Instrument; Short Physical Performance Battery; 4m gait speed) were used as anchor variables. To assess each DMO’s ability to detect changes, we calculated the standardized mean change as effect size. For estimating MIDs, both distribution-based and anchor-based methods were applied, followed by triangulation by experts if at least three anchor-based estimates were available per DMO, resulting in single-point estimates. Results All three anchor variables demonstrated substantial changes. Overall, 10 out of 24 available DMOs showed large and 7 DMOs moderate positive effects in the expected direction of the respective anchors. Seven DMOs showed no or only small effects. For 12 DMOs, at least three anchor-based estimates were available, enabling MID triangulation. MIDs for walking activity DMOs per day were: a walking duration of 10 minutes, a step count of 1,000 steps, 50 walking bouts (WB), and 15 WBs in WBs over 10 seconds. For gait DMOs, depending on the walking bout length, MIDs for walking speed were between 0.04 m/s and 0.08 m/s, and MIDs for cadence between 4 and 6 steps/minute. Almost all DMOs showed a strong ability to detect improvement in mobility, but rarely in detecting decline. Conclusions For the first time, MIDs are presented for real-world DMOs in PFF patients. These MIDs inform sample size requirements and interpretation of intervention effects for clinical trials, thereby providing guidance and reassurance for clinicians and regulatory bodies.