Conditions such as coronary artery disease, osteoarthritis, and cancer and its treatments are considered high risk for the onset of frailty, which may worsen disease responsiveness, activities of daily living (ADL), and life prognosis. Therefore, prevention of frailty in non-older adults at high risk holds significant individual and societal value. However, the evidence remains insufficient regarding whether digital health service interventions should be conducted in this population, which led to the development of this recommendation. A total of 11 673 articles were initially identified. After removing 2239 duplicates, 9434 were included in the primary screening. Following abstract and full-text screening, 64 articles were ultimately included in the systematic review (Figure 1) 1. Participants included patients with cancer, coronary artery disease, stroke, neurodegenerative disorders such as multiple sclerosis, obesity, and type 2 diabetes. Settings included in-home, outpatient, and inpatient environments. Interventions involved remote exercise instruction via videoconference, web-based, or app-based programs, and virtual reality (VR) training, with intervention durations ranging from 3 weeks to 1 year. Meta-analyses were conducted for outcomes including physical activity (activity levels, time, steps), endurance (6-min walk test 6MWT, VO2 peak), physical function (gait speed, Dynamic Gait Index DGI, Timed Up and Go test TUG), and quality of life (SF-36 Social Functioning). Eight studies were included in the meta-analysis on physical activity volume, comprising six asynchronous interventions (e.g., apps, wearable devices) and two synchronous interventions (e.g., VR, videoconference). Both asynchronous (mean difference: 0.21, 95% CI: 0.11–0.31) and synchronous interventions (mean difference: 0.61, 95% CI: 0.11–1.12) significantly increased physical activity volume. When all eight studies were pooled, digital health interventions significantly improved physical activity volume (mean difference: 0.23, 95% CI: 0.13–0.32) (Figure 2) 1. Seven studies assessed physical activity duration, all of which used asynchronous interventions. The meta-analysis showed no significant effect on activity duration (mean difference: 2.54 min/day, 95% CI: −3.38 to 8.46) (Figure 3) 1. Seven studies were included for daily step count: six asynchronous and one synchronous. Asynchronous interventions significantly increased step count (mean difference: 0.27, 95% CI: 0.10–0.43), while the effect of synchronous interventions was not significant (mean difference: –0.11, 95% CI: −0.58 to 0.37). The pooled analysis of all seven studies showed a significant increase in step count with digital health interventions (mean difference: 0.22, 95% CI: 0.07–0.38) (Figure 4) 1. Twenty studies were included in the meta-analysis of the 6MWT, comprising four asynchronous and 16 synchronous interventions. Asynchronous interventions showed no significant effect (mean difference: 13.79 m, 95% CI: −4.65 to 32.23), whereas synchronous interventions significantly increased walking distance (mean difference: 31.56 m, 95% CI: 20.92–42.20). When all studies were pooled, digital health interventions significantly improved 6MWT performance (mean difference: 27.12 m, 95% CI: 17.90–36.33) (Figure 5) 1. For VO2 peak, three studies were included: one asynchronous and two synchronous. The meta-analysis showed that synchronous interventions significantly increased VO2 peak (mean difference: 2.87 mL/kg/min, 95% CI: 1.42–4.33). Overall, digital health interventions significantly improved VO2 peak (mean difference: 2.81 mL/kg/min, 95% CI: 1.49–4.12) (Figure 6) 1. Twelve studies were included in the meta-analysis of gait speed: one asynchronous and 11 synchronous interventions. Synchronous interventions significantly improved gait speed (mean difference: 0.10 m/s, 95% CI: 0.02–0.18). Overall, pooled analysis of all 12 studies confirmed a significant improvement in gait speed with digital health interventions (mean difference: 0.10 m/s, 95% CI: 0.02–0.18) (Figure 7) 1. Two studies assessed the DGI, both of which involved synchronous interventions. The pooled analysis demonstrated a significant improvement in DGI (mean difference: 3.54 points, 95% CI: 1.81–5.27) (Figure 8) 1. For the TUG test, 16 studies were included: one asynchronous and 15 synchronous. Synchronous interventions significantly improved TUG performance (mean difference: −1.52 s, 95% CI: −1.82 to −1.22). The pooled analysis of all 16 studies showed a significant improvement in TUG (mean difference: −1.47 s, 95% CI: −1.76 to −1.18) (Figure 9) 1. Eight studies assessed SF-36 Social Functioning, including three asynchronous and five synchronous interventions. Both asynchronous (mean difference: 6.51, 95% CI: 0.96–12.06) and synchronous interventions (mean difference: 10.05, 95% CI: 4.83–15.27) significantly improved social functioning. The pooled analysis confirmed that digital health interventions significantly improved SF-36 Social Functioning (mean difference: 8.39, 95% CI: 4.58–12.19) (Figure 10) 1. Taken together, digital health interventions may improve frailty-related indicators such as physical activity, endurance, and physical function in non-older adults at risk of frailty. However, since most included studies had a risk of bias due to lack of blinding, we conclude with a weak recommendation to implement digital health interventions in this population. Type of digital device or intervention method and study outcomes (Table 1) 1. RCT IP RCT HB RCT HB RCT HB RCT OP RCT OP RCT OP RCT HB RCT HB RCT HB RCT HB RCT IP Cluster RCT OP RCT HB RCT HB/OP RCT HB RCT HB RCT HB/OP RCT HB RCT HB RCT OP RCT HB RCT Fitness center RCT HB RCT HB RCT OP RCT OP RCT HB RCT OP RCT HB RCT OP RCT (Delayed Control Design) HB RCT HB RCT HB RCT IP RCT HB RCT OP RCT HB RCT HB RCT HB RCT HB RCT HB RCT HB RCT HB RCT Unknown RCT HB RCT Not specified RCT HB RCT HB RCT HB RCT HB RCT HB RCT OP RCT OP RCT OP RCT OP RCT OP RCT HB RCT OP (4 weeks) + HB (4 weeks) RCT HB RCT HB RCT OP RCT OP This systematic review included a broad array of studies that employed digital devices, yet pooling the results by individual device type proved challenging. For several outcomes, we performed meta-analyses that categorized the interventions as either synchronous or asynchronous. However, the findings may have been heavily influenced by limited sample sizes, and the available evidence is insufficient to determine clearly whether synchronous or asynchronous interventions are more effective. A meta-analysis revealed that digital interventions improved several frailty-related outcomes, suggesting potential benefits for high-risk non-older adults. However, a higher dropout rate was noted in asynchronous interventions (risk ratio: 1.28, 95% CI: 1.13–1.44), raising concerns about the type of intervention, device selection, and adherence. A public representative participated in the guideline development meetings and the recommendation voting, and their input was considered as much as possible. One study included a cost-effectiveness analysis, reporting that the incremental cost for preventing or treating one case of overweight/obesity over 1 year was 10 665 Indian rupees (INR), equivalent to approximately £112.30 or about 20 000 Japanese yen as of early December 2024 (Figure 11). Considering the potential future burden of frailty-related health complications, this suggests that such interventions may be cost-effective. The included studies had small sample sizes and short intervention durations. High-quality studies with longer interventions and evaluation of asynchronous versus synchronous approaches are needed to clarify the effectiveness and provide more specific recommendations. This document is the official English translation of the Japanese version published by the Japanese Association for Sarcopenia and Frailty and the National Center for Geriatrics and Gerontology in March 2025 1. This work was supported by the Japan Agency for Medical Research and Development (AMED) under the project “Project for Establishing Research and Development Infrastructure for the Social Implementation of Prevention and Health Promotion/Healthcare Social Implementation Infrastructure Development Project” (Management No. JP22rea522005). M.A. has disclosed financial relationships with Daiichi Sankyo Co. Ltd., MSD K.K., Toa Eiyo Co. Ltd., Towa Pharmaceutical Co. Ltd., Eisai Co. Ltd., Kracie Pharmaceutical Co. Ltd., Tsumura Pharmaceutical Co. Ltd., Tanabe Mitsubishi Pharma Corporation, Chugai Pharmaceutical Co. Ltd., Ono Pharmaceutical Co. Ltd., Takeda Pharmaceutical Co. Ltd., Astellas Pharma Inc., Bayer Yakuhin Ltd.
Inoue et al. (Wed,) studied this question.