Effects of Using a Lumbar-Type Hybrid Assistive Limb by Older Adults With Frailty on Exercise Self-Efficacy: Retrospective Analysis Based on the Kanagawa Prospective ME-BYO Cohort Study

BACKGROUND: Improving exercise self-efficacy (ESE) is a key strategy for promoting sustained physical activity among older adults with frailty. The lumbar-type Hybrid Assistive Limb (HAL), a wearable device that supports voluntary movement through bioelectric signals, may facilitate mastery experiences in individuals with low baseline self-efficacy via adjustable assistance. However, its effects on ESE remain unclear. OBJECTIVE: This study evaluated the effects of lumbar-type HAL exercise on ESE in older adults with frailty and examined whether responsiveness differed according to baseline readiness for exercise. METHODS: A retrospective pooled analysis was conducted using data from 2 prospective studies of a lumbar-type HAL-based frailty prevention program: a feasibility study (n=20) and a randomized controlled trial (RCT; n=76; lumbar-type HAL group: n=38; waitlist control: n=38). Participants aged 50 to 84 years performed sit-to-stand and squat exercises using the lumbar-type HAL twice weekly. ESE was assessed using a validated 4-item questionnaire (range 4-20). The prespecified primary analysis examined within-person changes in ESE in the lumbar-type HAL group (2-tailed paired t test; n=96). A secondary analysis compared changes in ESE between the lumbar-type HAL and waitlist control groups in the RCT using the Welch t test and baseline-adjusted analysis of covariance. A prespecified subgroup analysis compared participants in stages 1 to 4 of the transtheoretical model of behavior change (exercise habit not established) with those in stage 5 (maintenance). All statistical tests were 2 sided, with significance at P<.05. RESULTS: The primary pooled paired analysis demonstrated a significant increase in ESE of 1.05 points (95% CI 0.41-1.69; P=.002). The subgroup analysis was also significant: those in stages 1 to 4 (n=56) improved by 1.73 points vs 0.05 points for those in stage 5 (n=38), a difference of 1.68 points (95% CI 0.41-2.95; P=.01). The largest within-group improvement was observed in stage 1 (precontemplation: +3.50 points, 95% CI 0.27-6.73; P=.04; Cohen d=1.14); 4 of 6 participants improved and none deteriorated. In the secondary RCT between-group comparison, the lumbar-type HAL group showed a directionally favorable improvement (between-group difference=1.21 points, 95% CI -0.33 to 2.75; P=.12; Cohen d=0.36). The subgroup differences were 2.10 points in stages 1 to 4 (P=.07; Cohen d=0.58) and 0.02 points in stage 5 (P=.99). Sensitivity analyses confirmed robustness to source-related heterogeneity (heterogeneity test: P=.74; source-adjusted regression: P=.76). CONCLUSIONS: The pooled analysis demonstrated significant within-person improvement in ESE among older adults with frailty following lumbar-type HAL training, with the greatest improvement in individuals without established exercise habits. The RCT indicated a favorable but inconclusive between-group effect, and adequately powered confirmatory RCTs remain necessary. The exploratory response among precontemplators-typically difficult to engage in interventions-suggests that HAL-supported mastery experiences may support disengaged older adults at high risk of frailty in starting physical exercise.