Adherence to in-home gait retraining using feedback from a smartphone app

Abstract

Purpose: A high knee adduction moment (KAM) has repeatedly been linked to progression and severity of medial knee osteoarthritis (OA). It places higher than normal loads on the medial compartment of the knee and has become a target for biomechanical intervention. To lower KAM and slow OA progression, gait modification and retraining have been explored. However, the challenge has been incorporating feasible, longitudinal methods for gait retraining. We recently demonstrated that healthy subjects can quickly respond to auditory feedback from a pressure-detecting shoe insole to lower their KAM by shifting their foot pressure medially. Encouraged by these findings, we started a randomized clinical trial enrolling subjects with medial knee OA and outfitting them with a sensor shoe insole and a smartphone feedback kit. Here, we report about our initial experience with adherence to a three-week long program using this technology for daily in-home gait retraining. We were specifically concerned that negative feedback may cause training avoidance and decreased adherence over time and therefore sought to evaluate if there were differences in adherence between this active group and one that handles similar technology without receiving feedback. Methods: Subjects with mild to moderate clinical and radiographic knee OA, BMI <38, and age >40 yrs were recruited and informed consent was obtained. At the beginning of their lab visit, all subjects received a standardized pair of flexible shoes, a pair of pressure-detecting shoe insoles (Moticon GmbH, Munich, Germany), which communicate wirelessly with a smartphone that generates auditory cues for subjects to follow. Subjects were randomized into two groups, Group 1 (‘active group’) which received and trained with auditory feedback cues, and Group 2 (‘passive control’) that interacted with the smartphone without receiving feedback. Both groups learned to connect the insole with the smartphone via a smartphone application. After lab-based instructions, subjects in Group 1 and Group 2 were sent home to complete a three-week, six-days-per-week self-directed feedback training or recording, respectively. All subjects were asked to document the duration of each session in their journal. Subjects were instructed to log a total of 15 min per day, which could be broken into smaller increments (e.g., 5-5-5 min). After 3 weeks of training, the journals were returned to study staff and daily logged minutes were summed. Missing entries were treated as “0”. Statistical analyses were performed using chi-square and repeated measures ANOVA. Results: The journal entries of the first 22 subjects were reviewed. Four subjects were excluded from analysis: one because he withdrew from the study (Group 1), two because journals were not returned (Group 1), and one because excessive training times were logged (143 min daily; Group 2). This left 18 subjects for analysis: 10 were in Group 1 (2M/8F, 61+/−11 yrs, KL = 2.4), and eight were in Group 2 (2M/6F, 62+/−9 yrs, KL = 2.6) (P > .343). On average, Group 1 trained 12.6 min/day, while Group 2 recorded 14.2 min/day during the three week period (P = .679). There was no Group effect on daily logged journal entries (P = .578), and both groups started similarly motivated (17.2 and 17.4 min/day during the first three days); however, with time, the logged journal times decayed (P = .003). Regression analysis suggested a daily reduction of 32 s (Group 1) and 38 s (Group 2), finishing clearly under 10 min/day on Day 18 (Figure). Conclusions: Our initial experience with self-driven, in-home gait retraining using a pressure insole and negative feedback is promising in terms of overall adherence. Although we saw a decrease in adherence over time, the difference was likely not related to negative feedback from training since the no-feedback group also experienced a similar trend in decreased adherence rates over 3 weeks. In that ‘no journal entry’ was treated equally to ‘no training’, a worst case scenario has been created. Treating ‘no journal entries’ as ‘lost data’ would increase the daily average session times to 17.3 and 16.7 min for Group 1 and 2, respectively, with both groups finishing at 12.5 min on Day 18. Nevertheless, continued follow up and encouragement techniques may be necessary to help maintain adherence.